Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS20050071328 A1
Publication typeApplication
Application numberUS 10/676,711
Publication dateMar 31, 2005
Filing dateSep 30, 2003
Priority dateSep 30, 2003
Also published asCA2540679A1, CA2540679C, EP1673703A1, US20100228715, US20120233142, US20130282713, WO2005033979A1
Publication number10676711, 676711, US 2005/0071328 A1, US 2005/071328 A1, US 20050071328 A1, US 20050071328A1, US 2005071328 A1, US 2005071328A1, US-A1-20050071328, US-A1-2005071328, US2005/0071328A1, US2005/071328A1, US20050071328 A1, US20050071328A1, US2005071328 A1, US2005071328A1
InventorsStephen Lawrence
Original AssigneeLawrence Stephen R.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Personalization of web search
US 20050071328 A1
Abstract
A system and method for creating a user profile and for using the user profile to order search results returned by a search engine. The user profile is based on search queries submitted by a user, the user's specific interaction with the documents identified by the search engine and personal information provided by the user. Generic scores associated with the search results are modulated by the user profile to measure their relevance to a user's preference and interest. The search results are re-ordered accordingly so that the most relevant results appear on the top of the list. User profiles can be created and/or stored on the client side or server side of a client-server network environment.
Images(13)
Previous page
Next page
Claims(58)
1. A method of personalizing search results of a search engine, comprising:
accessing a user profile for a user based on information about the user, the user information including information derived from a set of documents, the set of documents comprising a plurality of documents selected from the set consisting of documents identified by search results from the search engine, documents accessed by the user, documents linked to the documents identified by search results from the search engine, and documents linked to the documents accessed by the user;
receiving a search query from the user;
identifying a set of search result documents that match the search query;
assigning a generic score to each of at least a plurality of the search result documents;
assigning a personalized score to each document of the plurality of search result documents in accordance with the generic score assigned to the document and the user profile; and
ranking the set of search result documents according to their personalized scores.
2. The method of claim 1, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have been viewed by the user.
3. The method of claim 1, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have not been viewed by the user.
4. The method of claim 1, including updating the user profile by:
updating a term-based profile by identifying a set of terms from a document in the set of documents, and adding information about the identified set of terms to the term-based profile; and
updating a category-based profile by classifying the document into a plurality of categories, and adding information about the plurality of categories to the category-based profile.
5. The method of claim 4, including updating the user profile by:
updating a link-based profile by analyzing links in the document, and adding information derived from the analyzed links to the link-based profile.
6. The method of claim 5, wherein the link-based profile includes information about URLs or portions of URLs.
7. The method of claim 6, wherein the link-based profile comprises:
a plurality of URLs and a weight associated with each URL, wherein the weight is based on one or more factors selected from the group consisting of frequency with which the user visits the URL, time the user has spent viewing a document associated with the URL and quantity of the user's scrolling activity at the document; and
a plurality of hosts and a weight associated with each host, wherein the weight is based on frequency of the user's visits to the host.
8. The method of claim 7, wherein the URLs further include URLs that have not been visited by the user, but are related to the URLs that have been visited by the user and the weight of an unvisited URL depends on its distance to at least one related URLs that have been visited.
9. The method of claim 4, wherein a term in the term-based profile is an expression comprising at least one word and a weight.
10. The method of claim 9, wherein the weight is a weight associated with occurrences of the term in the set of documents.
11. The method of claim 9, wherein the weight of a term depends at least partially on the term's term frequency and inverse document frequency in said set of documents.
12. The method of claim 4, wherein a category in the category-based profile characterizes at least one aspect of documents in the category and the category is associated with a weight indicative of the category's importance relative to other categories.
13. The method of claim 12, wherein the at least one aspect of the documents in the category is selected from the group consisting of: document format, document type, document topic and document origin.
14. The method of claim 1, including updating the user profile by:
updating a category-based profile by classifying a document in the set of documents into a plurality of categories, and adding information about the plurality of categories to the category-based profile.
15. The method of claim 1, wherein the user profile is stored on a server of the search engine.
16. The method of claim 1, wherein the user profile is stored on a client associated with the user.
17. The method of claim 1, wherein the user is a group of users.
18. A method of personalizing search results of a search engine, comprising:
creating a plurality of user profiles for a plurality of users, each user profile including at least a user's identification number and information derived from documents visited by the user;
receiving a search query from a user of the plurality of users, the search query including at least one query term and the user's identification number;
selecting a set of documents from the Internet, assigning to each document in the set a generic score that characterizes the relevance of the document to the at least one query term;
retrieving the user's user profile and assigning to each of the set of documents a profile score based on the user profile; and
ranking the set of documents according to their generic and profile scores.
19. The method of claim 18, wherein the step of creating a user's user profile further comprises:
creating a term-based profile by extracting a set of terms from documents visited by the user and associating a weight with each extracted term; and
creating a category-based profile by determining a plurality of categories associated with documents visited by the user and associating a weight with each determined category.
20. The method of claim 18, wherein the step of creating a user's user profile further comprises:
creating a link-based profile by analyzing links in documents visited by the user and associating weights with the link.
21. The method of claim 18, wherein the user profile for a particular user includes demographic and geographic information provided by the user.
22. The method of claim 18, wherein the documents visited by the user from which information is derived for use in a particular user's user profile are selected based on the user's activities when visiting the documents.
23. The method of claim 18, including storing the plurality of user profiles on a server of the search engine; and
the retrieving including identifying the user's user profile based on a user identification number associated with both the user and the user's user profile.
24. The method of claim 18, including storing the plurality of user profiles on client computers associated with the plurality of users.
25. The method of claim 18, wherein the ranked set of documents comprises two sets of documents, one set of documents ordered by their generic scores and the other set of documents ordered by personalized scored generated by combining the document's generic and profile scores.
26. The method of claim 25, including interleaving or intermixing the two sets of documents to form the ranked set of documents.
27. A search engine system, comprising:
one or more central processing units for executing programs;
an interface for receiving event messages; and
a search engine module executable by the one or more central processing units, the module comprising:
instructions for accessing a user profile for a user, the user profile based on information about the user, the user information including information derived from a set of documents, the set of documents comprising a plurality of documents selected from the set consisting of documents identified by search results from the search engine system, documents accessed by the user, documents linked to the documents identified by search results from the search engine system, and documents linked to the documents accessed by the user;
instructions for receiving a search query from the user;
instructions for identifying a set of search result documents that match the search query;
instructions for assigning a generic score to each of at least a plurality of the search result documents;
instructions for assigning a personalized score to each document of the plurality of search result documents in accordance with the generic score assigned to the document and the user profile; and
instructions for ranking at least the plurality of the search result documents according to their personalized scores.
28. The system of claim 27, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have been viewed by the user.
29. The system of claim 27, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have not been viewed by the user.
30. The system of claim 27, further including:
instructions for updating a term-based profile by identifying a set of terms from a document in the set of documents, and adding information about the identified set of terms to the term-based profile; and
instructions for updating a category-based profile by classifying the document into a plurality of categories, and adding information about the plurality of categories to the category-based profile.
31. The system of claim 30, further including:
instructions for updating a link-based profile by analyzing links in the document, and adding information derived from the analyzed links to the link-based profile.
32. The system of claim 31, wherein the link-based profile includes information about URLs or portions of URLs.
33. The system of claim 32, wherein the link-based profile comprises:
a plurality of URLs and a weight associated with each URL, wherein the weight is based on one or more factors selected from the group consisting of frequency with which the user visits the URL, time the user has spent viewing a document associated with the URL and quantity of the user's scrolling activity at the document; and
a plurality of hosts and a weight associated with each host, wherein the weight is based on frequency of the user's visits to the host.
34. The system of claim 33, wherein the URLs further include URLs that have not been visited by the user, but are related to the URLs that have been visited by the user and the weight of an unvisited URL depends on its distance to at least one related URLs that have been visited.
35. The system of claim 30, wherein a term in the term-based profile is an expression comprising at least one word and a weight.
36. The system of claim 35, wherein the weight is a weight associated with occurrences of the term in the set of documents.
37. The system of claim 35, wherein the weight of a term depends at least partially on the term's term frequency and inverse document frequency in said set of documents.
38. The system of claim 30, wherein a category in the category-based profile characterizes at least one aspect of documents in the category and the category is associated with a weight indicative of the category's importance relative to other categories.
39. The system of claim 38, wherein the at least one aspect of the documents in the category is selected from the group consisting of: document format, document type, document topic and document origin.
40. The system of claim 27, wherein the user profile is stored on a server of the search engine.
41. The system of claim 27, wherein the user profile is stored on a client associated with the user.
42. The system of claim 27, wherein the user is a group of users.
43. A computer program product for use in conjunction with a computer system, the computer program product comprising a computer readable storage medium and a computer program mechanism embedded therein, the computer program mechanism comprising:
instructions for accessing a user profile for a user based on information about the user, the user information including information derived from a set of documents, the set of documents comprising a plurality of documents selected from the set consisting of documents identified by search results from a search engine, documents accessed by the user, documents linked to the documents identified by search results from the search engine, and documents linked to the documents accessed by the user;
instructions for receiving a search query from the user;
instructions for identifying a set of search result documents that match the search query;
instructions for assigning a generic score to each of at least a plurality of the search result documents;
instructions for assigning a personalized score to each document of the plurality of search result documents in accordance with the generic scores assigned to the document and the user profile; and
instructions for ranking at least the plurality of the search result documents according to their personalized scores.
44. The computer program product of claim 43, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have been viewed by the user.
45. The computer program product of claim 43, wherein the set of documents include a plurality of documents that have been identified by search results from the search engine and that have not been viewed by the user.
46. The computer program product of claim 43, further including:
instructions for updating a term-based profile by identifying a set of terms from a document in the set of documents, and adding information about the identified set of terms to the term-based profile; and
instructions for updating a category-based profile by classifying the document into a plurality of categories, and adding information about the plurality of categories to the category-based profile.
47. The computer program product of claim 46, further including:
instructions for updating a link-based profile by analyzing links in the document, and adding information derived from the analyzed links to the link-based profile.
48. The computer program product of claim 46, wherein the link-based profile includes information about URLs or portions of URLs.
49. The computer program product of claim 48, wherein the link-based profile comprises:
a plurality of URLs and a weight associated with each URL, wherein the weight is based on one or more factors selected from the group consisting of frequency with which the user visits the URL, time the user has spent viewing a document associated with the URL and quantity of the user's scrolling activity at the document; and
a plurality of hosts and a weight associated with each host, wherein the weight is based on frequency of the user's visits to the host.
50. The computer program product of claim 49, wherein the URLs further include URLs that have not been visited by the user, but are related to the URLs that have been visited by the user and the weight of an unvisited URL depends on its distance to at least one related URLs that have been visited.
51. The computer program product of claim 46, wherein a term in the term-based profile is an expression comprising at least one word and a weight.
52. The computer program product of claim 48, wherein the weight is a weight associated with occurrences of the term in the set of documents.
53. The computer program product of claim 48, wherein the weight of a term depends at least partially on the term's term frequency and inverse document frequency in said set of documents.
54. The computer program product of claim 46, wherein a category in the category-based profile characterizes at least one aspect of documents in the category and the category is associated with a weight indicative of the category's importance relative to other categories.
55. The computer program product of claim 54, wherein the at least one aspect of the documents in the category is selected from the group consisting of: document format, document type, document topic and document origin.
56. The computer program product of claim 43, wherein the user profile is stored on a server of the search engine.
57. The computer program product of claim 43, wherein the user profile is stored on a client associated with the user.
58. The computer program product of claim 43, wherein the user is a group of users.
Description
FIELD OF THE INVENTION

The present invention relates generally to the field of a search engine in a computer network system, in particular to system and method of creating a user profile for a user of a search engine and using the user profile to customize search results in response to search queries submitted by the user.

BACKGROUND OF THE INVENTION

Search engines provide a powerful source of indexed documents from the Internet (or an intranet) that can be rapidly scanned in response to a search query submitted by a user. Such a query is usually very short (on average about two to three words). As the number of documents accessible via the Internet grows, the number of documents that match the query may also increase. However, not every document matching the query is equally important from the user's perspective. As a result, a user is easily overwhelmed by an enormous number of documents returned by a search engine, if the engine does not order the search results based on their relevance to the user's query.

One approach to improving the relevance of search results to a search query is to use the link structure of different web pages to compute global “importance” scores that can be used to influence the ranking of search results. This is sometimes referred to as the PageRank algorithm. A more detailed description of the PageRank algorithm can be found in the article “The Anatomy of a Large-Scale Hypertextual Search Engine” by S. Brin and L. Page, 7th International World Wide Web Conference, Brisbane, Australia and U.S. Pat. No. 6,285,999, both of which are hereby incorporated by reference as background information.

An important assumption in the PageRank algorithm is that there is a “random surfer” who starts his web surfing journey at a randomly picked web page and keeps clicking on the links embedded in the web pages, never hitting the “back” button. Eventually, when this random surfer gets bored of the journey, he may re-start a new journey by randomly picking another web page. The probability that the random surfer visits (i.e., views or downloads) a web page depends on the web page's page rank.

From an end user's perspective, a search engine using the PageRank algorithm treats a search query the same way no matter who submits the query, because the search engine does not ask the user to provide any information that can uniquely identify the user. The only factor that affects the search results is the search query itself, e.g., how many terms are in the query and in what order. The search results are a best fit for the interest of an abstract user, the “random surfer”, and they are not be adjusted to fit a specific user's preferences or interests.

In reality, a user like the random surfer never exists. Every user has his own preferences when he submits a query to a search engine. The quality of the search results returned by the engine has to be evaluated by its users' satisfaction. When a user's preferences can be well defined by the query itself, or when the user's preference is similar to the random surfer's preference with respect to a specific query, the user is more likely to be satisfied with the search results. However, if the user's preference is significantly biased by some personal factors that are not clearly reflected in a search query itself, or if the user's preference is quite different from the random user's preference, the search results from the same search engine may be less useful to the user, if not useless.

As suggested above, the journey of the random surfer tends to be random and neutral, without any obvious inclination towards a particular direction. When a search engine returns only a handful of search results that match a query, the order of the returned results is less significant because the requesting user may be able to afford the time to browse each of them to discover the items most relevant to himself. However, with billions of web pages connected to the Internet, a search engine often returns hundreds or even thousands of documents that match a search query. In this case, the ordering of the search results is very important. A user who has a preference different from that of the random surfer may not find what he is looking for in the first five to ten documents listed in the search results. When that happens, the user is usually left with two options: (1) either spending the time required to review more of the listed documents so as to locate the relevant documents; or (2) refining the search query so as to reduce the number of documents that match the query. Query refinement is often a non-trivial task, sometimes requiring more knowledge of the subject or more expertise with search engines than the user possesses, and sometimes requiring more time and effort than the user is willing to expend.

For example, assume that a user submits to a search engine a search query having only one term “blackberry”. Without any other context, on the top of a list of documents returned by a PageRank-based search engine may be a link to www.blackberry.net, because this web page has the highest page rank. However, if the query requester is a person with interests in foods and cooking, it would be more useful to order the search results so as to include at the top of the returned results web pages with recipes or other food related text, pictures or the like. It would be desirable to have a search engine that is able to reorder its search results, or to otherwise customize the search results, so as to emphasize web pages that are most likely to be of interest to the person submitting the search query. Further, it would be desirable for such a system to require minimal input from individual users, operating largely or completely without explicit input from the user with regard to the user's preferences and interests. Finally, it would be desirable for such a system to meet users' requirements with respect to security and privacy.

SUMMARY

A search engine utilizes user profiles to customize search results. A user profile comprises multiple items that characterize a user's search preference. These items are extracted from various information sources, including previous search queries submitted by the user, links from or to the documents identified by the previous queries, sampled content from the identified documents as well as personal information implicitly or explicitly provided by the user.

When the search engine receives a search query from a user, it first identifies a set of documents that match the search query. Each document is associated with a generic rank based on the document's page rank, the text associated with the document, and the search query. The search engine also identifies the user's profile and correlates the user profile with each of the identified documents. The correlation between a document and the user profile produces a profile rank for the document, indicating the relevance of the document to the user. The search engine then combines the document's generic rank and profile rank into a personalized rank. Finally, the documents are ordered according to their personalized ranks.

In one embodiment, a user profile may comprise a plurality of sub-profiles, each sub-profile characterizing the user's interest from a different perspective. A term-based profile comprises a plurality of terms, each term carrying a weight indicative of its importance relative to other terms. A category-based profile comprises multiple categories, optionally organized into a hierarchical map. The user's search preferences may be associated with at least a subset of the multiple categories, each category having an associated weight indicating the user's interest in the documents falling into this category. There may be multiple category-based profiles for a user. In some embodiments, the sub-profiles include a link-based profile, which includes a plurality of links that are, directly or indirectly, related to identified documents, each link having a weight indicating the importance of the link. Links in the link-based profile may be further organized with respect to different hosts and domains.

The present invention, including user profile construction and search results re-ordering and/or scoring, can be implemented on either the client side or the server side of a client-server network environment.

BRIEF DESCRIPTION OF THE DRAWINGS

The aforementioned features and advantages of the invention as well as additional features and advantages thereof will be more clearly understood hereinafter as a result of a detailed description of preferred embodiments of the invention when taken in conjunction with the drawings.

FIG. 1 illustrates a client-server network environment.

FIG. 2 illustrates multiple sources of user information and their relationship to a user profile.

FIG. 3 is an exemplary data structure that may be used for storing term-based profiles for a plurality of users.

FIG. 4A is an exemplary category map that may be used for classifying a user's past search experience.

FIG. 4B is an exemplary data structure that may be used for storing category-based profiles for a plurality of users.

FIG. 5 is an exemplary data structure that may be used for storing link-based profiles for a plurality of users.

FIG. 6 is a flowchart illustrating paragraph sampling.

FIG. 7A is a flowchart illustrating context analysis.

FIG. 7B depicts a process of identifying important terms using context analysis.

FIG. 8 illustrates a plurality of exemplary data structures that may be used for storing information about documents after term-based, category-based and/or link-based analyses, respectively.

FIG. 9A is a flowchart illustrating a personalized web search process according to one embodiment.

FIG. 9B is a flowchart illustrating a personalized web search process according to another embodiment.

FIG. 10 is a block diagram of a personalized search engine.

Like reference numerals refer to corresponding parts throughout the several views of the drawings.

DESCRIPTION OF EMBODIMENTS

The embodiments discussed below include systems and methods that create a user profile based a user's past experience with a search engine and then use the user profile to rank search results in response to search queries provided by the user.

FIG. 1 provides an overview of a typical client-server network environment 100 in which the present invention may be implemented. A plurality of clients 102 are connected to a search engine system 107 through a network 105, e.g., the Internet. Search engine system 107 comprises one or more search engines 104. A search engine 104 is responsible for processing a search query submitted by a client 102, generating search results in accordance with the search query and returning the results to the client. Search engine system 107 may also comprise one or more content servers 106 and one or more user profile servers 108. A content server 106 stores a large number of indexed documents retrieved from different websites. Alternately, or in addition, the content server 106 stores an index of documents stored on various websites. In one embodiment, each indexed document is assigned a page rank according to the document's link structure. The page rank serves as a query independent measure of the document's importance. A search engine 104 communicates with one or more content servers 106 to select a plurality of documents in response to a specific search query. The search engine assigns a score to each document based on the document's page rank, the text associated with the document, and the search query.

A user profile server 108 stores a plurality of user profiles. Each profile includes information that uniquely identifies a user as well as his previous search experience and personal information, which can be used to refine search results in response to the search queries submitted by this user. Different approaches are available for user profile construction. For example, a user profile can be created by requiring a first-time user to fill in a form or answer a survey. This approach may be useful in certain applications such as opening a bank account. But it is hardly a favorable one in the context of a search engine. First, a user's interaction with a search engine is usually a dynamic process. As time goes on, the user's interests may change. This change may be reflected by the search queries submitted by the user, or by the user's handling of the search results, or both. The user's answers to questions on a form tend to become less useful over time, unless the user chooses to update his answers periodically. Unlike an occasional update of phone number in the case of an on-line bank account, frequent updates of a user profile in the case of a search engine significantly affect its user friendliness, which is an important consideration when a user chooses among the search engines currently available. Further, it is known that users are reluctant to provide explicit feedback, such as filling out of a form, as many users find it too burdensome. Thus, while some users may provide explicit feedback on their interests, it is desirable to have a procedure for implicitly obtaining information about the user's interests without requiring any explicit or new actions by the user.

It is has been observed that a search engine user's past search activities provide useful hints about the user's personal search preferences. FIG. 2 provides a list of sources of user information that are beneficial for user profile construction. For example, previously submitted search queries 201 are very helpful in profiling a user's interests. If a user has submitted multiple search queries related to diabetes, it is more likely than not that this is a topic of interest to the user. If the user subsequently submits a query including the term “organic food”, it can be reasonably inferred that he may be more interested in those organic foods that are helpful in fighting diabetes. Similarly, the universal resource locators (URL) 203 associated with the search results in response to the previous search queries and their corresponding anchor texts 205, especially for search result items that have been selected or “visited” by the user (e.g., downloaded or otherwise viewed by the user), are helpful in determining the user's preferences. When a first page contains a link to a second page, and the link has text associated with it (e.g., text neighboring the link), the text associated with the link is called “anchor text” with respect to the second page. Anchor text establishes a relationship between the text associated with a URL link in a document and another document to which the URL link points. The advantages of anchor text include that it often provides an accurate description of the document to which the URL link points, and it can be used to index documents that cannot be indexed by a text-based search engine, such as images or databases.

After receiving search results, the user may click on some of the URL links, thereby downloading the documents referenced by those links, so as to learn more details about those documents. Certain types of general information 207 can be associated with a set of user selected or use identified documents. For purposes of forming a user profile, the identified documents from which information is derived for inclusion in the user profile may include: documents identified by search results from the search engine, documents accessed (e.g., viewed or downloaded, for example using a browser application) by the user (including documents not identified in prior search results), documents linked to the documents identified by search results from the search engine, and documents linked to the documents accessed by the user, or any subset of such documents.

The general information 207 about the identified documents may answer questions such as, what is the format of the document? Is it in hypertext markup language (HTML), plain text, portable document format (PDF), or Microsoft Word? What is the topic of the document? Is it about science, health or business? This information is also helpful in profiling the user's interests. In addition, information about a user's activities 209 with respect to the user selected documents (sometimes herein call the identified documents), such as how long the user spent viewing the document, the amount of scrolling activity on the document, and whether the user has printed, saved or bookmarked the document, also suggests the importance of the document to the user as well as the user's preferences. In some embodiments, information about user activities 209 is used both when weighting the importance of information extracted or derived from the user identified documents. In some embodiments, information about user activities 209 is used to determine which of the user identified documents to use as the basis for deriving the user profile. For example, information 209 may be used to select only documents that received significant user activity (in accordance with predefined criteria) for generating the user profile, or information 209 may be used to exclude from the profiling process documents that the user viewed for less than a predefined threshold amount of time.

Finally, the content of the identified documents from previous search activities is a rich source of information about a user's interests and preferences. Key terms appearing in the identified documents and their frequencies with which they appear in the identified documents are not only useful for indexing the document, but are also a strong indication of the user's personal interests, especially when they are combined with other types of user information discussed above. In one embodiment, instead of the whole documents, sampled content 211 from the identified documents is extracted for the purpose of user profile construction, to save storage space and computational cost. In another embodiment, various information related to the identified documents may be classified to constitute category information 213 about the identified documents. More discussion about content sampling, the process of identifying key terms in an identified document and the usage of the category information is provided below.

Optionally, a user may choose to offer personal information 215, including demographic and geographic information associated with the user, such as the user's age or age range, educational level or range, income level or range, language preferences, marital status, geographic location (e.g., the city, state and country in which the user resides, and possibly also including additional information such as street address, zip code, and telephone area code), cultural background or preferences, or any subset of these. Compared with other types of personal information such as a user's favorite sports or movies that are often time varying, this personal information is more static and more difficult to infer from the user's search queries and search results, but maybe crucial in correctly interpreting certain queries submitted by the user. For example, if a user submits a query containing “Japanese restaurant”, it is very likely that he may be searching for a local Japanese restaurant for dinner. Without knowing the user's geographical location, it is hard to order the search results so as to bring to the top those items that are most relevant to the user's true intention. In certain cases, however, it is possible to infer this information. For example, users often select results associated with a specific region corresponding to where they live.

Creating a user profile 230 from the various sources of user information is a dynamic and complex process. In some embodiments, the process is divided into sub-processes. Each sub-process produces one type of user profile characterizing a user's interests or preferences from a particular perspective. They are:

    • a term-based profile 231—this profile represents a user's search preferences with a plurality of terms, where each term is given a weight indicating the importance of the term to the user;
    • a category-based profile 233—this profile correlates a user's search preferences with a set of categories, which may be organized in a hierarchal fashion, with each category being given a weight indicating the extent of correlation between the user's search preferences and the category; and
    • a link-based profile 235—this profile identifies a plurality of links that are directly or indirectly related to the user's search preferences, with each link being given a weight indicating the relevance between the user's search preferences and the link.

In some embodiments, the user profile 230 includes only a subset of these profiles 231, 233, 235, for example just one or two of these profiles. In one embodiment, the user profile 230 includes a term-based profile 231 and a category-based profile 233, but not a link-based profile 235.

In one embodiment, a user profile is created and stored on a server (e.g., user profile server 108) associated with a search engine. The advantage of such deployment is that the user profile can be easily accessed by multiple computers, and that since the profile is stored on a server associated with (or part of) the search engine 104, it can be easily used by the search engine 104 to personalize the search results. In another embodiment, the user profile can be created and stored on the user's computer, sometimes called the client in a network environment. Creating and storing a user profile on a user's computer not only reduces the computational and storage cost for the search engine's servers, but also satisfies some users' privacy requirements. In yet another embodiment, the user profile may be created and updated on the client, but stored on a search engine server. Such embodiment combines some of the benefits illustrated in the other two embodiments. A disadvantage of this arrangement is that it may increase the network traffic between clients and the search engine servers. It is understood by a person of ordinary skill in the art that the user profiles of the present invention can be implemented using client computers, server computers, or both.

FIG. 3 illustrates an exemplary data structure, a term-based profile table 300, that may be used for storing term-based profiles for a plurality of users. Table 300 includes a plurality of records 310, each record corresponding to a user's term-based profile. A term-based profile record 310 includes a plurality of columns including a USER_ID column 320 and multiple columns of (TERM, WEIGHT) pairs 340. The USER_ID column stores a value that uniquely identifies a user or a group of users sharing the same set of (TERM, WEIGHT) pairs, and each (TERM, WEIGHT) pair 340 includes a term, typically 1-3 words long, that is usually important to the user or the group of users and a weight associated with the term that quantifies the importance of the term. In one embodiment, the term may be represented as one or more n-grams. An n-gram is defined as a sequence of n tokens, where the tokens may be words. For example, the phrase “search engine” is an n-gram of length 2, and the word “search” is an n-gram of length 1.

N-grams can be used to represent textual objects as vectors. This makes it possible to apply geometric, statistical and other mathematical techniques, which are well defined for vectors, but not for objects in general. In the present invention, n-grams can be used to define a similarity measure between two terms based on the application of a mathematical function to the vector representations of the terms.

The weight of a term is not necessarily a positive value. If a term has a negative weight, it may suggest that the user prefers that his search results should not include this term and the magnitude of the negative weight indicates the strength of the user's preference for avoiding this term in the search results. By way of example, for a group of surfing fans at Santa Cruz, Calif., the term-based profile may include terms like “surfing club”, “surfing event” and “Santa Cruz” with positive weights. The terms like “Internet surfing” or “web surfing” may also be included in the profile. However, these terms are more likely to receive a negative weight since they are irrelevant and confusing with the authentic preference of the users sharing this term-based profile.

A term-based profile itemizes a user's preference using specific terms, each term having certain weight. If a document matches a term in a user's term-based profile, i.e., its content includes exactly this term, the term's weight will be assigned to the document; however, if a document does not match a term exactly, it will not receive any weight associated with this term. Such a requirement of relevance between a document and a user profile sometimes may be less flexible when dealing with various scenarios in which a fuzzy relevance between a user's preference and a document exists. For example, if a user's term-based profile includes terms like “Mozilla” and “browser”, a document containing no such terms, but other terms like “Galeon” or “Opera” will not receive any weight because they do not match any existing term in the profile, even though they are actually Internet browsers. To address the need for matching a user's interests without exact term matching, a user's profile may include a category-based profile.

FIG. 4A illustrates a hierarchal category map 400 according to the Open Directory Project (http://dmoz.org/). Starting from the root level of map 400, documents are organized under several major topics, such as “Art”, “News”, “Sports”, etc. These major topics are often too broad to delineate a user's specific interest. Therefore, they are further divided into sub- topics that are more specific. For example, topic “Art” may comprise sub-topics like “Movie”, “Music” and “Literature” and the sub-topic “Music” may further comprise sub-sub-topics like “Lyrics”, “News” and “Reviews”. Note that each topic is associated with a unique CATEGORY_ID like 1.1 for “Art”, 1.4.2.3 for “Talk Show” and 1.6.1 for “Basketball”.

A user's specific interests may be associated with multiple categories at various levels, each of which may have a weight indicating the degree of relevance between the category and the user's interest. In one embodiment, a category-based profile may be implemented using a Hash table data structure as shown in FIG. 4B. A category-based profile table 450 includes a table 455 that comprises a plurality of records 460, each record including a USER_ID and a pointer pointing to another data structure, such as table 460-1. Table 460-1 may include two columns, CATEGORY_ID column 470 and WEIGHT column 480. CATEGORY_ID column 470 contains a category's identification number as shown in FIG. 4A, suggesting that this category is relevant to the user's interests and the value in the WEIGHT column 480 indicates the degree of relevance of the category to the user's interests.

A user profile based upon the category map 400 is a topic-oriented implementation. The items in a category-based profile can also be organized in other ways. In one embodiment, a user's preference can be categorized based on the formats of the documents identified by the user, such as HTML, plain text, PDF, Microsoft Word, etc. Different formats may have different weights. In another embodiment, a user's preference can be categorized according to the types of the identified documents, e.g., an organization's homepage, a person's homepage, a research paper, or a news group posting, each type having an associated weight. Another type category that can be used to characterize a user's search preferences is document origin, for instance the country associated with each document's host. In yet another embodiment, the above-identified category-based profiles may co-exist, with each one reflecting one aspect of a user's preferences.

Besides term-based and category-based profiles, another type of user profile is referred to as a link-based profile. As discussed above, the PageRank algorithm is based on the link structure that connects various documents over the Internet. A document that has more links pointing to it is often assigned a higher page rank and therefore attracts more attention from a search engine. Link information related to a document identified by a user can also be used to infer the user's preferences. In one embodiment, a list of preferred URLs are identified for a user by analyzing the frequency of his access to those URLs. Each preferred URL may be further weighted according to the time spent by the user and the user's scrolling activity at the URL, and/or other user activities (209, FIG. 2) when visiting the document at the URL. In another embodiment, a list of preferred hosts are identified for a user by analyzing the user's frequency of accessing web pages of different hosts. When two preferred URLs are related to the same host the weights of the two URLs may be combined to determine a weight for the host. In another embodiment, a list of preferred domains are identified for a user by analyzing the user's frequency of accessing web pages of different domains. For example, for finance.yahoo.com, the host is “finance.yahoo.com” while the domain is “yahoo.com”.

FIG. 5 illustrates a link-based profile using a Hash table data structure. A link-based profile table 500 includes a table 510 that includes a plurality of records 520, each record including a USER_ID and a pointer pointing to another data structure, such as table 510-1. Table 510-1 may include two columns, LINK_ID column 530 and WEIGHT column 540. The identification number stored in the LINK_ID column 530 may be associated with a preferred URL or host. The actual URL/host/domain may be stored in the table instead of the LINK_ID, however it is preferable to store the LINK_ID to save storage space.

A preferred list of URLs and/or hosts includes URLs and/or hosts that have been directly identified by the user. The preferred list of URLs and/or host may furthermore extend to URLs and/or hosts indirectly identified by using methods such as collaborative filtering or bibliometric analysis, which are known to persons of ordinary skill in the art. In one embodiment, the indirectly identified URLs and/or host include URLs or hosts that have links to/from the directly identified URLs and/or hosts. These indirectly identified URLs and/or hosts are weighted by the distance between them and the associated URLs or hosts that are directly identified by the user. For example, when a directly identified URL or host has a weight of 1, URLs or hosts that are one link away may have a weight of 0.5, URLs or hosts that are two links away may have a weight of 0.25, etc. This procedure can be further refined by reducing the weight of links that are not related to the topic of the original URL or host, e.g., links to copyright pages or web browser software that can be used to view the documents associated with the user selected URL or host. Irrelevant Links can be identified based on their context or their distribution. For example, copyright links often use specific terms (e.g., copyright or “All rights reserved” are commonly used terms in the anchor text of a copyright link); and links to a website from many unrelated websites may suggest that this website is not topically related (e.g., links to the Internet Explorer website are often included in unrelated websites). The indirect links can also be classified according to a set of topics and links with very different topics may be excluded or be assigned a low weight.

The three types of user profiles discussed above are generally complimentary to one another since different profiles delineate a user's interests and preferences from different vantage points. However, this does not mean that one type of user profile, e.g., category-based profile, is incapable of playing a role that is typically played by another type of user profile. By way of example, a preferred URL or host in a link-based profile is often associated with a specific topic, e.g., finance.yahoo.com is a URL focusing on financial news. Therefore, what is achieved by a link-based profile that comprises a list of preferred URLs or hosts to characterize a user's preference may also be achievable, at least in part, by a category-based profile that has a set of categories that cover the same topics covered by preferred URLs or hosts.

It is a non-trivial operation to construct various types of user profiles that can be stored in the data structures shown in FIGS. 3-5 based on the user information listed in FIG. 2. Given a document identified (e.g., viewed) by a user, different terms in the document may have different importance in revealing the topic of the document. Some terms, e.g., the document's title, may be extremely important, while other terms may have little importance. For example, many documents contain navigational links, copyright statements, disclaimers and other text that may not be related to the topic of the document. How to efficiently select appropriate documents, content from those documents and terms from within the content is a challenging topic in computational linguistics. Additionally, it is preferred to minimize the volume of user information processed, so as make the process of user profile construction computationally efficient. Skipping less important terms in a document helps in accurately matching a document with a user's interest.

Paragraph sampling (described below with reference to FIG. 6) is a procedure for automatically extracting content from a document that may be relevant to a user. An important observation behind this procedure is that less relevant content in a document, such as navigational links, copyright statements, disclaimer, etc., tend to be relatively short segments of text. In one embodiment, paragraph sampling looks for the paragraphs of greatest length in a document, processing the paragraphs in order of decreasing length until the length of a paragraph is below a predefined threshold. The paragraph sampling procedure optionally selects up to a certain maximum amount of content from each processed paragraph. If few paragraphs of suitable length are found in a document, the procedure falls back to extracting text from other parts of the document, such as anchor text and ALT tags.

FIG. 6 is a flowchart illustrating the major steps of paragraph sampling. Paragraph sampling begins with the step 610 of removing predefined items, such as comments, JavaScript and style sheets, etc., from a document. These items are removed because they are usually related to visual aspects of the document when rendered on a browser and are unlikely to be relevant to the document's topic. Following that, the procedure may select the first N words (or M sentences) at step 620 from each paragraph whose length is greater than a threshold value, MinParagraphLength, as sampled content. In one embodiment, the values of N and M are chosen to be 100 and 5, respectively. Other values may be used in other embodiments.

In order to reduce the computational and storage load associated with the paragraph sampling procedure, the procedure may impose a maximum limit, e.g., 1000 words, on the sampled content from each document. In one embodiment, the paragraph sampling procedure first organizes all the paragraphs in a document in length decreasing order, and then starts the sampling process with a paragraph of maximum length. It is noted that the beginning and end of a paragraph depend on the appearance of the paragraph in a browser, not on the presence of uninterrupted a text string in the HTML representation of the paragraph. For this reason, certain HTML commands, such as commands for inline links and for bold text, are ignored when determining paragraph boundaries. In some embodiments, the paragraph sampling procedure screens the first N words (or M sentences) so as to filter out those sentences including boilerplate terms like “Terms of Service” or “Best viewed”, because such sentences are usually deemed irrelevant to the document's topic.

Before sampling a paragraph whose length is above the threshold value, the procedure may stop sampling content from the document if the number of words in the sampled content has reached the maximum word limit. If the maximum word limit has not been reached after processing all paragraphs of length greater than the threshold, optional steps 630, 640, 650 and 670 are performed. In particular, the procedure adds the document title (630), the non-inline HREF links (640), the ALT tags (650) and the meta tags (670) to the sampled content until it reaches the maximum word limit.

Once the documents identified by a user have been scanned, the sampled content can be used for identifying a list of most important (or unimportant) terms through context analysis. Context analysis attempts to learn context terms that predict the most important (or unimportant) terms in a set of identified documents. Specifically, it looks for prefix patterns, postfix patterns, and a combination of both. For example, an expression “x's home page” may identify the term “x” as an important term for a user and therefore the postfix pattern “* home page” can be used to predict the location of an important term in a document, where the asterisk “* ” represents any term that fits this postfix pattern. In general, the patterns identified by context analysis usually consist of m terms before an important (or unimportant) term and n terms after the important (or unimportant) term, where both m and n are greater than or equal to 0 and at least one of them is greater than 0. Typically, m and n are less than 5, and when non-zero are preferably between 1 and 3. Depending on its appearance frequency, a pattern may have an associated weight that indicates how important (or unimportant) the term recognized by the pattern is expected to be.

According to one embodiment of the present invention (FIG. 7A), context analysis has two distinct phases, a training phase 701 and an operational phase 703. The training phase 701 receives and utilizes a list of predefined important terms 712, an optional list of predefined unimportant terms 714, and a set of training documents (step 710). In some embodiments, the list of predefined unimportant terms is not used. The source of the lists 712, 714 is not critical. In some embodiments, these lists 712, 714 are generated by extracting words or terms from a set of documents (e.g., a set of several thousand web pages of high page rank) in accordance with a set of rules, and then editing them to remove terms that in the opinion of the editor do not belong in the lists. The source of the training documents is also not critical. In some embodiments, the training documents comprise a randomly or pseudo-randomly selected set of documents already known to the search engine. In other embodiments, the training documents are selected from a database of documents in the search engine in accordance with predefined criteria.

During the training phase 701, the training documents are processed (step 720), using the lists of predefined important and unimportant terms, so as to identify a plurality of context patterns (e.g., prefix patterns, postfix patterns, and prefix-postfix patterns) and to associate a weight with each identified context pattern. During the operational phase 703, the context patterns are applied to documents identified by the user (step 730) to identify a set of important terms (step 740) that characterize the user's specific interests and preferences. Learning and delineating a user's interests and preferences is usually an ongoing process. Therefore, the operational phase 703 may be repeated to update the set of important terms that have been captured previously. This may be done each time a user accesses a document, according to a predetermined schedule, at times determined in accordance with specified criteria, or otherwise from time to time. Similarly, the training phase 701 may also be repeated to discover new sets of context patterns and to recalibrate the weights associated with the identified context patterns.

Below is a segment of pseudo code that exemplifies the training phase:

For each document in the set {
 For each improtant term in the document {
  For m = 0 to MaxPrefix {
   For n = 0 to MaxPostfix {
    Extract the m words before the important
    term and the n words after the important
    term as s;
    Add 1 to ImportantContext (m,n,s);
   }
  }
 }
 For each unimportant term in the document {
  For m = 0 to MaxPrefix {
   For n = 0 to MaxPostfix {
    Extract the m words before the
    unimportant term and the n words after
    the unimportant term as s;
    Add 1 to UnimportantContext (m,n,s);
   }
  }
 }
}
For m = 0 to MaxPrefix {
 For n = 0 to MaxPostfix {
  For each vaalue of s {
   Set the weight for s to a function of
   ImportantContext (m,n,s), and
   UnimportantContext (m,n,s);
  }
 }
}

In the pseudo code above, the expression s refers to a prefix pattern (n=0), a postfix pattern (m=0) or a combination of both (m>0 & n>0). Each occurrence of a specific pattern is registered at one of the two multi-dimensional arrays, ImportantContext(m,n,s) or UnimportantContext(m,n,s). The weight of a prefix, postfix or combination pattern is set higher if this pattern identifies more important terms and fewer unimportant terms and vice versa. Note that it is possible that a same pattern may be associated with both important and unimportant terms. For example, the postfix expression “* operating system” may be used in the training documents 716 in conjunction with terms in the list of predefined important terms 712 and also used in conjunction with terms in the list of predefined unimportant terms 714. In this situation, the weight associated with the postfix pattern “* operating system” (represented by the expression Weight(1,0,“operating system”)) will take into account the number of times the postfix expression is used in conjunction with terms in the list of predefined important terms as well as the number of times the postfix expression is used in conjunction with terms in the list of predefined unimportant terms. One possible formula to determine the weight of a context pattern s is:
Weight(m,n,s)=Log(ImportantContext(m,n,s)+1)−Log(UnimportantContext(m,n,s)+1).
Other weight determination formulas may be used in other embodiments.

In the second phase of the context analysis process, the weighted context patterns are used to identify important terms in one or more documents identified by the user. Referring to FIG. 7B, in the first phase a computer system receives training data 750 and creates a set of context patterns 760, each context pattern having an associated weight. The computer system then applies the set of context patterns 760 to a document 780. In FIG. 7B, previously identified context patterns found within the document 780 are highlighted. Terms 790 associated with the context patterns are identified and each such term receives a weight based on the weights associated with the context patterns. For example, the term “Foobar” appears in the document twice, in association with two different patterns, the prefix pattern “Welcome to *” and the postfix pattern “* builds”, and the weight 1.2 assigned to “Foobar” is the sum of the two patterns' weights, 0.7 and 0.5. The other identified term “cars” has a weight of 0.8 because the matching prefix pattern “world's best *” has a weight of 0.8. In some embodiments the weight for each term is computed using a log transform, where the final weight is equal to log(initial weight+1). It is possible that the two terms “Foobar” and “cars” may not be in the training data 750 and may have never been encountered by the user before. Nevertheless, the context analysis method described above identifies these terms and adds them to the user's term-based profile. Thus, context analysis can be used to discover terms associated with a user's interests and preferences even when those terms are not included in a predefined database of terms.

As noted, the output of context analysis can be used directly in constructing a user's term-based profile. Additionally, it may be useful in building other types of user profiles, such as a user's category-based profile. For example, a set of weighted terms can be analyzed and classified into a plurality of categories covering different topics, and those categories can be added to a user's category-based profile.

After executing the context analysis on a set of documents identified by or for a user, the resulting set of terms and weights may occupy a larger amount of storage than allocated for each user's term-based profile. Also, the set of terms and corresponding weights may include some terms with weights much, much smaller than other terms within the set. Therefore, in some embodiments, at the conclusion of the context analysis, the set of terms and weights is pruned by removing terms having the lowest weights (A) so that the total amount of storage occupied by the term-based profile meets predefined limits, and/or (B) so as to remove terms whose weights are so low, or terms that correspond to older items, as defined by predefined criteria, that the terms are deemed to be not indicative of the user's search preferences and interests. In some embodiments, similar pruning criteria and techniques are also applied to the category-based profile and/or the link-based profile.

In some embodiments, a user's profile is updated each time the user performs a search and selects at least one document from the search results to download or view. In some embodiments, the search engine builds a list of documents identified by the user (e.g., by selecting the documents from search results) over time, and at predefined times (e.g., when the list reaches a predefined length, or a predefined amount of time has elapsed), performs a profile update. When performing an update, new profile data is generated, and the new profile data is merged with the previously generated profile data for the user. In some embodiments, the new profile data is assigned higher importance than the previously generated profile data, thereby enabling the system to quickly adjust a user's profile in accordance with changes in the user's search preferences and interests. For example, the weights of items in the previously generated profile data may be automatically scaled downward prior to merging with the new profile data. In one embodiment, there is a date associated with each item in the profile, and the information in the profile is weighted based on its age, with older items receiving a lower weight than when they were new. In other embodiments, the new profile data is not assigned high importance than the previously generated profile data.

The paragraph sampling and context analysis methods may be used independently or in combination. When used in combination, the output of the paragraph sampling is used as input to the context analysis method.

It is further noted that the above-described methods used for creating user profiles, e.g., paragraph sampling and context analysis, may be also leveraged for determining the relevance of a candidate document to a user's preference. Indeed, the primary mission of a search engine is to identify a series of documents that are most relevant to a user's preference based on the search queries submitted by the user as well as the user's user profile. FIG. 8 illustrates several exemplary data structures that can be used to store information about a document's relevance to a user profile from multiple perspectives. For each candidate document, each identified by a respective DOC_ID, term-based document information table 810 includes multiple pairs of terms and their weights, category-based document information table 830 includes a plurality of categories and associated weights, and link-based document information table 850 includes a set of links and corresponding weights.

The rightmost column of each of the three tables (810, 830 and 850) stores the rank (i.e., a computed score) of a document when the document is evaluated using one specific type of user profile. A user profile rank can be determined by combining the weights of the items associated with a document. For instance, a category-based or topic-based profile rank may be computed as follows. A user may prefer documents about science with a weight of 0.6, while he dislikes documents about business with a weight of −0.2. Thus, when a science document matches a search query, it will be weighted higher than a business document. In general, the document topic classification may not be exclusive. A candidate document may be classified as being a science document with probability of 0.8 and a business document with probability of 0.4. A link-based profile rank may be computed based on the relative weights allocated to a user's URL, host, domain, etc., preferences in the link-based profile. In one embodiment, term-based profile rank can be determined using known techniques, such as the term frequency-inverse document frequency (TF-IDF). The term frequency of a term is a function of the number of times the term appears in a document. The inverse document frequency is an inverse function of the number of documents in which the term appears within a collection of documents. For example, very common terms like “the” occur in many documents and consequently as assigned a relatively low inverse document frequency.

When a search engine generates search results in response to a search query, a candidate document D that satisfies the query is assigned a query score, QueryScore, in accordance with the search query. This query score is then modulated by document D's page rank, PageRank, to generate a generic score, GenericScore, that is expressed as
GenericScore=QueryScore*PageRank.

This generic score may not appropriately reflect document D's importance to a particular user U if the user's interests or preferences are dramatically different from that of the random surfer. The relevance of document D to user U can be accurately characterized by a set of profile ranks, based on the correlation between document D's content and user U's term-based profile, herein called the TermScore, the correlation between one or more categories associated with document D and user U's category-based profile, herein called the CategoryScore, and the correlation between the URL and/or host of document D and user U's link-based profile, herein called the LinkScore. Therefore, document D may be assigned a personalized rank that is a function of both the document's generic score and the user profile scores. In one embodiment, this personalized score can be expressed as:
PersonalizedScore=GenericScore*(TermScore+CategoryScore+LinkScore).

FIGS. 9A and 9B represent two embodiments, both implemented in a client-server network environment such as the network environment 100 shown in FIG. 1. In the embodiment shown in FIG. 9A, the search engine 104 receives a search query from a client 102 at step 910 that is submitted by a particular user. In response, the search engine 104 may optionally generate a query strategy at step 915 (e.g., the search query is normalized so as to be in proper form for further processing, and/or the search query may be modified in accordance with predefined criteria so as to automatically broaden or narrow the scope of the search query). At step 920, the search engine 104 submits the search query (or the query strategy, if one is generated) to the content server 106. The content server identifies a list of documents that match the search query at step 920, each document having a generic score that depends on the document's page rank and the search query. In general, all the three operations (steps 910, 915 and 920) are conducted by the search engine system 107, which is on the server side of the network environment 100. There are two options on where to implement the operations following these first three steps.

In some embodiments that employ a server-side implementation, the user's identification number is embedded in the search query. Based on the user's identification number, the user profile server 108 identifies the user's user profile at step 925. Starting from step 930, the user profile server 108 or the search engine 104 analyzes each document identified at step 920 to determine its relevance to the user's profile, creates a profile score for the identified document at step 935 and then assigns the document a personalized score that is a function of the document's generic and profile scores at step 940. At step 942, the user profile server 108 or the search engine 104 checks whether this the last one in the list of identified documents. If no, the system processes the next document in the list. Otherwise, the list of documents are re-ordered according to their personalized scores and then sent to the corresponding client from which the user submitted the search query.

Embodiments using a client-side implementation are similar to the server-side implementation, except that after step 920, the identified documents are sent to the corresponding client from which the user submitted the query. This client stores the user's user profile and it is responsible for re-ordering the documents based upon the user profile. Therefore, this client-side implementation may reduce the server's workload. Further, since there is no privacy concern with the client-side implementation, a user may be more willing to provide private information to customize the search results. However, a significant limitation to the client-side implementation is that only a limited number of documents, e.g., the top 50 documents (as determined using the generic rank), may be sent to a client for re-ordering due to limited network bandwidth. In contrast, the server-side implementation may be able to apply a user's profile to a much larger number of documents, e.g., 1000, that match the search query. Therefore, the client-side implementation may deprive a user access to those documents having relatively low generic ranks, but significantly high personalized ranks.

FIG. 9B illustrates another embodiment. Unlike the embodiment depicted in FIG. 9A, where the search query is not personalized before submitting the search query to the search engine 104, a generic query strategy is adjusted (step 965) according to the user's user profile to create a personalized query strategy. For example, relevant terms from the user profile may be added to the search query with associated weights. The creation of the personalized query strategy can be performed either on the client side or on the server side of the system. This embodiment avoids the network bandwidth restriction facing the previous embodiment. Finally, the search engine 104 submits the personalized query strategy to the content server 106 (step 970), and therefore the search results returned by the content server have already been ordered by the documents' personalized ranks (step 975).

The profiles of a group of users with related interests may be combined together to form a group profile, or a single profile may be formed based on the documents identified by the users in the group. For instance, several family members may use the same computer to submit search queries to a search engine. If the computer is tagged with a single user identifier by the search engine, the “user” will be the entire family of users, and the user profile will be represent a combination or mixture of the search preferences of the various family members. An individual user in the group may optionally have a separate user profile that differentiates this user from other group members. In operation, the search results for a user in the group are ranked according to the group profile, or according to the group profile and the user's user profile when the user also has a separate user profile.

It is possible that a user may switch his interests so dramatically that his new interests and preferences bear little resemblance to his user profile, or a user may be temporarily interested in a new topic. In this case, personalized search results produced according to the embodiments depicted in FIGS. 9A and 9B may be less favorable than search results ranked in accordance with the generic ranks of the documents in the search results. Additionally, the search results provided to a user may not include new websites among the top listed documents because the user's profile tends to increase the weight of older websites which the user has visited (i.e., older websites from which the user has viewed or downloaded web pages) in the past.

To reduce the impact caused by a change in a user's preferences and interests, the personalized search results may be merged with the generic search results. In one embodiment, the generic search results and personalized search results are interleaved, with the odd positions (e.g., 1, 3, 5, etc.) of a search results list reserved for generic search results and the even positions (e.g., 2, 4, 6, etc.) reserved for personalized search results, or vice versa. Preferably, the items in the generic search results will not duplicate the items listed in the personalized search results, and vice versa. More generally, generic search results are intermixed or interleaved with personalized search results, so that the items in the search results presented to the user include both generic and personalized search results.

In another embodiment, the personalized ranks and generic ranks are further weighted by a user profile's confidence level. The confidence level takes into account factors such as how much information has been acquired about the user, how close the current search query matches the user's profile, how old the user profile is, etc. If only a very short history of the user is available, the user's profile may be assigned a correspondingly low confidence value. The final score of an identified document can be determined as:
FinalScore=ProfileScore*ProfileConfidence+GenericScore*(1−ProfileConfidence).
When intermixing generic and personalized results, the fraction of personalized results may be adjusted based on the profile confidence, for example using only one personalized result when the confidence is low.

Sometimes, multiple users may share a machine, e.g., in a public library. These users may have different interests and preferences. In one embodiment, a user may explicitly login to the service so the system knows his identity. Alternatively, different users can be automatically recognized based on the items they access or other characteristics of their access patterns. For example, different users may move the mouse in different ways, type differently, and use different applications and features of those applications. Based on a corpus of events on a client and/or server, it is possible to create a model for identifying users, and for then using that identification to select an appropriate “user” profile. In such circumstances, the “user” may actually be a group of people having somewhat similar computer usage patterns, interests and the like.

Referring to FIG. 10, a personalized search engine system 1000 typically includes one or more processing units (CPU's) 1002, one or more network or other communications interfaces 1010, memory 1012, and one or more communication buses 1014 for interconnecting these components. The system 1000 may optionally include a user interface 1004, for instance a display 1006 and a keyboard 1008. Memory 1012 may include high speed random access memory and may also include non-volatile memory, such as one or more magnetic disk storage devices. Memory 1012 may include mass storage that is remotely located from the central processing unit(s) 1002. The memory 1012 preferably stores:

    • an operating system 1016 that includes procedures for handling various basic system services and for performing hardware dependent tasks;
    • a network communication module 1018 that is used for connecting the system 1000 to other servers or computers via one or more communication networks (wired or wireless), such as the Internet, other wide area networks, local area networks, metropolitan area networks, and so on;
    • a system initialization module 1020 that initializes other modules and data structures stored in memory 1012 required for the appropriate operation of system 1000;
    • a search engine 1022 for processing a search query, identifying and ordering search results according to the search query and a user's profile;
    • a user profile engine 1030 for gathering and processing user information, such as the user information identified in FIG. 2, and creating and updating a user's user profile that characterizes the user's search preferences and interests; and
    • data structures 1040, 1060 and 1080 for storing a plurality of user profiles.

The search engine 1022 may further comprise:

    • a generic rank module (or instructions) 1024 for processing a search query submitted by a user, identifying a list of documents matching the query and assigning each identified document a generic rank without reference to user specific information;
    • a user profile rank module (or instructions) 1026 for correlating each of a plurality of documents identified by the generic rank module 1024 with the user's user profile and assigning the document a profile rank indicating the relevance of the document to the user's search preferences and interests; and
    • a rank mixing module (or instructions) 1028 for combining the generic rank and the profile rank of an identified document into a personalized rank and re-ordering the list of documents according to their personalized ranks.
      In some embodiments, these modules 1024, 1026, 1028 may be implemented within a single procedure or in a set of procedures that reside within a single software module.

The user profile engine 1030 may further comprise:

    • a user information collection module 1032 for collecting and assorting various user information listed in FIG. 2;
    • a document content extraction module 1034 for selecting and extracting content from the documents identified by the user, to identify content relevant to the user's interests, using techniques such as paragraph sampling (as discussed above); and
    • a context analysis module 1036 for analyzing the content extracted by the document extraction module 1034 so as to identify terms that characterize a user's search preferences.

Each data structure hosting a user profile may further comprise:

    • a data structure 1042, 1062 or 1082 for storing a term-based user profile;
    • a data structure 1044, 1064 or 1084 for storing a category-based user profile; and
    • a data structure 1046, 1066 or 1086 for storing a link-based user profile.

The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated.

Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US7165119 *Oct 14, 2003Jan 16, 2007America Online, Inc.Search enhancement system and method having rankings, explicitly specified by the user, based upon applicability and validity of search parameters in regard to a subject matter
US7219073Aug 2, 2000May 15, 2007Brandnamestores.ComMethod for extracting information utilizing a user-context-based search engine
US7321889 *Sep 9, 2005Jan 22, 2008Suggestica, Inc.Authoring and managing personalized searchable link collections
US7333997Aug 12, 2003Feb 19, 2008Viziant CorporationKnowledge discovery method with utility functions and feedback loops
US7389282Nov 2, 2005Jun 17, 2008Viziant CorporationSystem and method for predictive analysis and predictive analysis markup language
US7421419Apr 12, 2006Sep 2, 2008Viziant CorporationSystem and method for evidence accumulation and hypothesis generation
US7467232Jun 20, 2006Dec 16, 2008Aol LlcSearch enhancement system and method having rankings, explicitly specified by the user, based upon applicability and validity of search parameters in regard to a subject matter
US7469276 *Dec 27, 2004Dec 23, 2008International Business Machines CorporationService offering for the delivery of information with continuing improvement
US7472119 *Jun 30, 2005Dec 30, 2008Microsoft CorporationPrioritizing search results by client search satisfaction
US7493301Sep 9, 2005Feb 17, 2009Suggestica, Inc.Creating and sharing collections of links for conducting a search directed by a hierarchy-free set of topics, and a user interface therefor
US7502783Sep 9, 2005Mar 10, 2009Suggestica, Inc.User interface for conducting a search directed by a hierarchy-free set of topics
US7536408Jul 26, 2004May 19, 2009Google Inc.Phrase-based indexing in an information retrieval system
US7562068 *Jun 30, 2004Jul 14, 2009Microsoft CorporationSystem and method for ranking search results based on tracked user preferences
US7567959Jan 25, 2005Jul 28, 2009Google Inc.Multiple index based information retrieval system
US7577718Jul 31, 2006Aug 18, 2009Microsoft CorporationAdaptive dissemination of personalized and contextually relevant information
US7580921Jul 26, 2004Aug 25, 2009Google Inc.Phrase identification in an information retrieval system
US7580929 *Jul 26, 2004Aug 25, 2009Google Inc.Phrase-based personalization of searches in an information retrieval system
US7584175Jul 26, 2004Sep 1, 2009Google Inc.Phrase-based generation of document descriptions
US7599914Jul 26, 2004Oct 6, 2009Google Inc.Phrase-based searching in an information retrieval system
US7599916 *Apr 20, 2005Oct 6, 2009Microsoft CorporationSystem and method for personalized search
US7606794 *Jun 28, 2005Oct 20, 2009Yahoo! Inc.Active Abstracts
US7624101 *Jan 31, 2006Nov 24, 2009Google Inc.Enhanced search results
US7640232Oct 14, 2003Dec 29, 2009Aol LlcSearch enhancement system with information from a selected source
US7680786 *Oct 30, 2006Mar 16, 2010Yahoo! Inc.Optimization of targeted advertisements based on user profile information
US7685192 *Jun 30, 2006Mar 23, 2010Amazon Technologies, Inc.Method and system for displaying interest space user communities
US7685199 *Jul 31, 2006Mar 23, 2010Microsoft CorporationPresenting information related to topics extracted from event classes
US7739277 *Sep 30, 2004Jun 15, 2010Microsoft CorporationSystem and method for incorporating anchor text into ranking search results
US7739408Dec 12, 2008Jun 15, 2010Aol Inc.System and method for general search parameters having quantized relevance values that are associated with a user
US7752201May 10, 2007Jul 6, 2010Microsoft CorporationRecommendation of related electronic assets based on user search behavior
US7761464 *Jun 19, 2006Jul 20, 2010Microsoft CorporationDiversifying search results for improved search and personalization
US7774340 *Jun 30, 2004Aug 10, 2010Microsoft CorporationMethod and system for calculating document importance using document classifications
US7779001 *Oct 29, 2004Aug 17, 2010Microsoft CorporationWeb page ranking with hierarchical considerations
US7783636Sep 28, 2006Aug 24, 2010Microsoft CorporationPersonalized information retrieval search with backoff
US7810027 *May 1, 2006Oct 5, 2010Bendik Mary MDocument management systems and methods
US7827170Aug 28, 2007Nov 2, 2010Google Inc.Systems and methods for demoting personalized search results based on personal information
US7827175 *Jun 10, 2004Nov 2, 2010International Business Machines CorporationFramework reactive search facility
US7836411Jun 10, 2004Nov 16, 2010International Business Machines CorporationSearch framework metadata
US7849079Jul 31, 2006Dec 7, 2010Microsoft CorporationTemporal ranking of search results
US7881981May 7, 2007Feb 1, 2011Yoogli, Inc.Methods and computer readable media for determining a macro-context based on a micro-context of a user search
US7920849Jul 13, 2007Apr 5, 2011Pop AdrianMethod and system for providing advertisements/messages based on wireless data communication technology
US7925649Dec 30, 2005Apr 12, 2011Google Inc.Method, system, and graphical user interface for alerting a computer user to new results for a prior search
US7930298 *Dec 27, 2007Apr 19, 2011Institute For Information IndustrySystem and method for generating 'snapshot's of learning objects
US7933975Aug 28, 2008Apr 26, 2011International Business Machines CorporationService offering for the delivery of information with continuing improvement
US7962462 *May 31, 2005Jun 14, 2011Google Inc.Deriving and using document and site quality signals from search query streams
US7966309Jan 17, 2007Jun 21, 2011Google Inc.Providing relevance-ordered categories of information
US7966321Jan 17, 2007Jun 21, 2011Google Inc.Presentation of local results
US7966324May 30, 2006Jun 21, 2011Microsoft CorporationPersonalizing a search results page based on search history
US7987185 *Dec 29, 2006Jul 26, 2011Google Inc.Ranking custom search results
US8005822Jan 17, 2007Aug 23, 2011Google Inc.Location in search queries
US8015184 *Apr 11, 2005Sep 6, 2011Yahoo! Inc.Method and apparatus for a search-enabled remote control device
US8019777 *Jun 7, 2010Sep 13, 2011Nexify, Inc.Digital content personalization method and system
US8027964 *Jul 13, 2007Sep 27, 2011Medio Systems, Inc.Personalized query completion suggestion
US8037042 *May 10, 2007Oct 11, 2011Microsoft CorporationAutomated analysis of user search behavior
US8060520 *Mar 11, 2010Nov 15, 2011Yahoo! Inc.Optimization of targeted advertisements based on user profile information
US8078607Mar 30, 2006Dec 13, 2011Google Inc.Generating website profiles based on queries from webistes and user activities on the search results
US8086605 *Jun 28, 2006Dec 27, 2011Yahoo! Inc.Search engine with augmented relevance ranking by community participation
US8095408Oct 11, 2005Jan 10, 2012Sharethis, Inc.System and method for facilitating network connectivity based on user characteristics
US8095565May 5, 2006Jan 10, 2012Microsoft CorporationMetadata driven user interface
US8095582 *Jan 4, 2010Jan 10, 2012Surf Canyon IncorporatedDynamic search engine results employing user behavior
US8103659 *Jun 6, 2006Jan 24, 2012A9.Com, Inc.Perspective-based item navigation
US8108362 *Feb 8, 2007Jan 31, 2012Microsoft CorporationSecure content descriptions
US8108383 *Nov 23, 2009Jan 31, 2012Google Inc.Enhanced search results
US8140515Oct 28, 2009Mar 20, 2012Cbs Interactive Inc.Personalization engine for building a user profile
US8145703Nov 16, 2007Mar 27, 2012Iac Search & Media, Inc.User interface and method in a local search system with related search results
US8180771Jul 18, 2008May 15, 2012Iac Search & Media, Inc.Search activity eraser
US8185514 *May 4, 2009May 22, 2012Topix LlcUser-interface feature and technique for providing users of a network site links that have been determined to be of interest to the user
US8185523 *Mar 17, 2006May 22, 2012Search Engine Technologies, LlcSearch engine that applies feedback from users to improve search results
US8204797Oct 11, 2007Jun 19, 2012William Paul WankerCustomizable electronic commerce comparison system and method
US8209337 *Nov 19, 2008Jun 26, 2012Core Logic, Inc.Content recommendation apparatus and method using tag cloud
US8214346Jan 30, 2009Jul 3, 2012Cbs Interactive Inc.Personalization engine for classifying unstructured documents
US8229862Jun 17, 2011Jul 24, 2012Google Inc.Methods and systems for rating associated members in a network
US8234263 *Jan 29, 2010Jul 31, 2012Cbs Interactive Inc.Personalization engine for building a dynamic classification dictionary
US8244721 *Feb 13, 2008Aug 14, 2012Microsoft CorporationUsing related users data to enhance web search
US8244737Jun 18, 2007Aug 14, 2012Microsoft CorporationRanking documents based on a series of document graphs
US8244766Apr 13, 2010Aug 14, 2012Microsoft CorporationApplying a model of a persona to search results
US8271485 *Apr 5, 2007Sep 18, 2012I-Business Center CorporationReply generation apparatus and method and program for causing computer to function as reply generating apparatus
US8316021 *Jun 30, 2010Nov 20, 2012Emergency 24, Inc.Methods and systems for enhanced placement search engine based on user usage
US8326861 *Jun 23, 2010Dec 4, 2012Google Inc.Personalized term importance evaluation in queries
US8341020Oct 25, 2010Dec 25, 2012Steve MorsaMatch engine marketing
US8346749 *Jun 27, 2008Jan 1, 2013Microsoft CorporationBalancing the costs of sharing private data with the utility of enhanced personalization of online services
US8359312 *Mar 16, 2009Jan 22, 2013Amiram GrynbergMethods for generating a personalized list of documents associated with a search query
US8359319 *Aug 21, 2008Jan 22, 2013Sudhir PendseTool for personalized search
US8364670 *Dec 28, 2005Jan 29, 2013Dt Labs, LlcSystem, method and apparatus for electronically searching for an item
US8370348 *Dec 6, 2011Feb 5, 2013Google Inc.Magazine edition recommendations
US8412698Apr 7, 2005Apr 2, 2013Yahoo! Inc.Customizable filters for personalized search
US8412702 *Mar 12, 2008Apr 2, 2013Yahoo! Inc.System, method, and/or apparatus for reordering search results
US8438165May 12, 2011May 7, 2013Microsoft CorporationInterest tracking using shared search queries and interactions
US8442973 *May 1, 2007May 14, 2013Surf Canyon, Inc.Real time implicit user modeling for personalized search
US8484098May 4, 2011Jul 9, 2013Michael BilottaSystem for information delivery facilitating partner rating of users and user ratings of partners
US8484186Sep 16, 2011Jul 9, 2013Consumerinfo.Com, Inc.Personalized people finder
US8484219 *Sep 22, 2010Jul 9, 2013Sony Computer Entertainment America LlcDeveloping a knowledge base associated with a user that facilitates evolution of an intelligent user interface
US8489515 *May 8, 2009Jul 16, 2013Comcast Interactive Media, LLC.Social network based recommendation method and system
US8489591Jun 2, 2011Jul 16, 2013Google Inc.Presentation of local results
US8510262 *May 21, 2008Aug 13, 2013Microsoft CorporationPromoting websites based on location
US8521734 *Jul 28, 2011Aug 27, 2013Yahoo! Inc.Search engine with augmented relevance ranking by community participation
US8521735 *Feb 27, 2012Aug 27, 2013Google Inc.Anonymous personalized recommendation method
US8538940 *Jun 17, 2008Sep 17, 2013International Business Machines CorporationIdentification of shared resources
US8538970Dec 30, 2004Sep 17, 2013Google Inc.Personalizing search results
US8543570Jan 20, 2012Sep 24, 2013Surf Canyon IncorporatedAdaptive user interface for real-time search relevance feedback
US8549011Jun 3, 2011Oct 1, 2013Microsoft CorporationIdentifying and recommending experts using shared queries and interactions
US8595228 *Jan 9, 2009Nov 26, 2013Google Inc.Preferred sites
US8606781 *Aug 9, 2005Dec 10, 2013Palo Alto Research Center IncorporatedSystems and methods for personalized search
US8612851Dec 6, 2011Dec 17, 2013Google Inc.Edition grid layout
US8620915 *Aug 28, 2007Dec 31, 2013Google Inc.Systems and methods for promoting personalized search results based on personal information
US8635216 *Sep 30, 2004Jan 21, 2014Avaya Inc.Enhancing network information retrieval according to a user search profile
US20070185858 *Aug 2, 2006Aug 9, 2007Yunshan LuSystems for and methods of finding relevant documents by analyzing tags
US20070233671 *Feb 14, 2007Oct 4, 2007Oztekin Bilgehan UGroup Customized Search
US20080114751 *May 1, 2007May 15, 2008Surf Canyon IncorporatedReal time implicit user modeling for personalized search
US20090030923 *Jun 17, 2008Jan 29, 2009International Business Machines CorporationIdentification of shared resources
US20090063475 *Aug 21, 2008Mar 5, 2009Sudhir PendseTool for personalized search
US20090094204 *Apr 5, 2007Apr 9, 2009Tomihisa ShiraishiReply generation apparatus and method and program for causing computer to function as reply generating apparatus
US20090138329 *Nov 26, 2007May 28, 2009William Paul WankerApplication of query weights input to an electronic commerce information system to target advertising
US20090234834 *Mar 12, 2008Sep 17, 2009Yahoo! Inc.System, method, and/or apparatus for reordering search results
US20100094859 *Jun 17, 2008Apr 15, 2010Gupta Puneet KFolksonomy-Enhanced Enterprise-Centric Collaboration and Knowledge Management System
US20100299140 *May 20, 2010Nov 25, 2010Cycorp, Inc.Identifying and routing of documents of potential interest to subscribers using interest determination rules
US20100299317 *Dec 20, 2007Nov 25, 2010Victor David UyMethod of displaying a subjective score with search engine results
US20100318533 *Jun 10, 2009Dec 16, 2010Yahoo! Inc.Enriched document representations using aggregated anchor text
US20110035375 *Aug 6, 2009Feb 10, 2011Ron BekkermanBuilding user profiles for website personalization
US20110179025 *Jan 21, 2010Jul 21, 2011Kryptonite Systems IncSocial and contextual searching for enterprise business applications
US20110295612 *May 31, 2011Dec 1, 2011Thierry Donneau-GolencerMethod and apparatus for user modelization
US20120066587 *Jul 3, 2009Mar 15, 2012Bao-Yao ZhouApparatus and Method for Text Extraction
US20120072424 *Sep 22, 2010Mar 22, 2012George WeisingDeveloping a Knowledge Base Associated with a User That Facilitates Evolution of an Intelligent User Interface
US20120078710 *Dec 8, 2011Mar 29, 2012Mark CramerDynamic search engine results employing user behavior
US20120117449 *Nov 8, 2010May 10, 2012Microsoft CorporationCreating and Modifying an Image Wiki Page
US20120233161 *Jul 28, 2011Sep 13, 2012Yahoo! Inc.Search engine with augmented relevance ranking by community participation
US20130191391 *Jul 23, 2012Jul 25, 2013Cbs Interactive, Inc.Personalization engine for building a dynamic classification dictionary
US20130226946 *Feb 27, 2012Aug 29, 2013Google Inc.Anonymous personalized recommendation method
EP1717720A1 *Apr 28, 2006Nov 2, 2006Palo Alto Research Center IncorporatedDiscovery of related documents based on a proximal neighborhood measure
EP1808786A1 *Jan 12, 2006Jul 18, 2007Yoogli, Inc.User context based search engine
EP1860575A1May 16, 2006Nov 28, 2007Hurra Communications GmbHMethod for evaluating information to be represented on a network page
EP1862916A1 *Jun 1, 2006Dec 5, 2007Microsoft CorporationIndexing Documents for Information Retrieval based on additional feedback fields
EP1974297A2 *Dec 20, 2006Oct 1, 2008Microsoft CorporationDynamic search with implicit user intention mining
EP1987484A2 *Feb 16, 2007Nov 5, 2008Hillcrest Laboratories, Inc.Systems and methods for placing advertisements
EP2105846A1 *Mar 28, 2008Sep 30, 2009Sony CorporationMethod of recommending content items
EP2336905A1 *Aug 26, 2009Jun 22, 2011Huawei Technologies Co., Ltd.A searching method and system
EP2704080A1 *Apr 24, 2008Mar 5, 2014KIT Digital Inc.Recommendation systems and methods
WO2006026612A2 *Aug 30, 2005Mar 9, 2006Chirag ChamanMethod and system for a personalized search engine
WO2006110480A1 *Apr 6, 2006Oct 19, 2006Yahoo IncCustomizable filters for personalized search
WO2007080413A1 *Jan 12, 2007Jul 19, 2007Vodafone PlcSearch platform
WO2007106269A1 *Feb 8, 2007Sep 20, 2007Microsoft CorpMining web search user behavior to enhance web search relevance
WO2007130914A2 *Apr 30, 2007Nov 15, 2007Surf Canyon IncDynamic search engine results employing user behavior
WO2008089346A1 *Jan 17, 2008Jul 24, 2008Google IncProviding relevance-ordered categories of information
WO2009030990A1 *Feb 6, 2008Mar 12, 2009Chin San Sathya WongMethod and system of interacting with a server, and method and system for generating and presenting search results
WO2009117216A1 *Feb 23, 2009Sep 24, 2009Yahoo, Inc.Personalizing sponsored search advertising layout using user behavior history
WO2012097124A1 *Jan 12, 2012Jul 19, 2012Alibaba Group Holding LimitedRanking of query results based on individuals' needs
WO2013013217A1 *Jul 20, 2012Jan 24, 2013Microsoft CorporationModeling search in a social graph
WO2013095901A1 *Dec 4, 2012Jun 27, 2013General Instrument CorporationA hierarchical behavioral profile
Classifications
U.S. Classification1/1, 707/E17.109, 707/999.003
International ClassificationG06F17/30
Cooperative ClassificationG06F17/3053, G06F17/30867
European ClassificationG06F17/30W1F
Legal Events
DateCodeEventDescription
Sep 8, 2004ASAssignment
Owner name: GOOGLE INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:LAWRENCE, STEPHEN R.;REEL/FRAME:015094/0672
Effective date: 20030930