WO2000072256A2 - Neuronales netz zum computergestützten wissensmanagement - Google Patents
Neuronales netz zum computergestützten wissensmanagement Download PDFInfo
- Publication number
- WO2000072256A2 WO2000072256A2 PCT/DE2000/001669 DE0001669W WO0072256A2 WO 2000072256 A2 WO2000072256 A2 WO 2000072256A2 DE 0001669 W DE0001669 W DE 0001669W WO 0072256 A2 WO0072256 A2 WO 0072256A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- optionally
- neural network
- knowledge
- computer
- assigned
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
Definitions
- the invention relates to a method for and a neural network for computer-aided knowledge management, based on a neural network formed by a computer, in particular for use in a decentralized computer-aided patent system that can be operated via the Internet.
- the neural network forms an artificial intelligence (AI) system by extending over an underlying knowledge base in the form of computer readable data.
- AI artificial intelligence
- the neural network itself is advantageously designed as a special type of vertically structured neural network similar to the harmony theory, in which each node or neuron is assigned a meaning as an element of the network. Each element is determined by several weighted connections (references) to different hierarchically higher elements.
- this invention supports all users of the patent system and factual documentation in the focusing, elaboration and priority-based filing of their development as an innovation.
- knowledge management includes the efficient management of all kinds of knowledge in document management, at the beginning of planning data, economic data, communication documents and publications in the form of text data, and later also in multimedia form using object-oriented technologies.
- the knowledge of suitable similarities is classified with the aim of reducing high redundancy right from the start.
- Automatic systems for classification mostly require text-based documents in order to be able to draw conclusions about their content with the help of computer-assisted text search engines, generally used text structuring for indexing, hierarchically classified thesaurus and linguistic laws.
- the latter form the basis of the self-classifying internet search engine that is common today.
- the associations found between the various documents can be used as input and training data for a neural network.
- the document DE341 1 168C2 discloses a hierarchical associative data structure, preferably for the storage of text, the document DE4108310C2 a processing system for a knowledge base in an expert system, in which the knowledge entered by a user about the degree of the original connection and the resultant result is stored Expert system is modified by a knowledge engineer using a knowledge-based processing system and is therefore maintained.
- the publication DE4124501C2 discloses an associative data structure in the form of a neural network and an associated method which leads to a clear classification using a metric between input data.
- the publication DE4400261C1 discloses, especially for the understanding of written texts, an associative data structure, which is constructed as an artificial neural network and contains a plurality of network nodes in successive layers, which contains especially unweighted node inputs and whose nodes are divided into two types, whereby one substantial redundancy reduction is achieved.
- DE19737939A1 shows a self-convergent associative data structure in the form of a neural network for computer-aided management of developments which form a Hilbert space with the special descriptions of the features determining the amount of knowledge as images of means in effects and vice versa.
- This associative data structure is suitable for the administration of all actually existing developments (hereinafter referred to as developments), in particular also for technical developments, which are the subject of patent law in the broader sense.
- patent system is to be understood as the entire knowledge management of technical developments, starting with the inventor or the decision-maker charged with a problem, through the service providers entrusted to them, up to sovereign institutions that are to be supported in their work in the factual area with computational technology .
- the recognized amount of knowledge describing the real world (hereinafter referred to as the amount of knowledge) is determined by developments that emerge hierarchically in terms of quantity.
- Knowledge management in this invention relates to such sets of knowledge.
- DE19737939A1 advantageously serves the person skilled in the art for a basic understanding of the structure, the functional principle and the use of this invention.
- the essence of the invention is based on the sets of knowledge assigned to one another in a hierarchical manner, which, due to a suitable definition of subsets in the form of an interaction pair, advantageously as ⁇ quantity
- the description of real developments or these descriptive sets of knowledge are based on the description of quantum mechanical states.
- the amount of knowledge that can thus be treated as a point set forms a vector space and has, in particular, favorable convergence properties for the computational application of the RITZ method.
- the convergence to the extremum statistically given over the elements in the applications with regard to a measure determined by the neural network leads to an associative data structure of minimal redundancy.
- the neural network is implemented computationally via a special dynamically managed associative data structure in the memory of a computer, for which very system-related programming of the method generating and managing this associative data structure is necessary; in particular, access to a linear memory area of sufficient size, the use and arithmetic of computer-specific pointers (hereinafter pointers) to parts of this memory area and a there is a sufficiently large stack for recursions via the associative data structure.
- pointers computer-specific pointers
- the associative data structure can be networked via networks.
- Each development of the associative data structure advantageously also includes a reference to a tabular content database, for example for a link to a source document from which the knowledge was generated.
- the input and output of the neural network to the user takes place via an intuitive dialog system, to which the neural network contained in a server is connected via communication links for data, particularly advantageously including the Internet.
- the neural network itself is advantageously generated exclusively by entering a standardized, strictly chronological stream, in that for each element, which is advantageously identified with an assigned unique time index, the interaction pair ⁇ quantity
- All the information required for output can advantageously be transferred back to the standardized, strictly chronological stream.
- the transmission to or the synchronization of several neural networks via communication links and by encrypting the stream ensures data security.
- neural networks mirrored on one side can read this stream in addition to their internal streams.
- the development systems are based hierarchically on the basis of a root system, on the basis of set operations (OR. AND. NOT), whereby the development system that characterizes the development is based on the subset of all subsets ⁇ quantity
- Quality> of the reference systems is defined and the genus is determined via a quantity relation that contains it, e.g. via the subset of all union sets of quantity and quality or the subset of one of these sets of all reference systems.
- Quality> uses the verbal definition to define the development and thus precisely the amount of knowledge that maps the quantity to the quality and vice versa the quality to the quantity, i.e. forms a subset within the cross set of all quantities and all qualities. In the ideal limit, this mapping is mathematically unambiguous. This amount of knowledge assigned to the ideally defined characteristic is necessary to a maximum, since in the other case not all the quantities or qualities belonging to the genus meet the above condition.
- the neural network with the elements for the definition of knowledge sets is an associative data structure, which is developed and incrementally expanded by the input and modification of relationships between developments, whereby the developments already entered serve to define further developments in reverse.
- the data structure corresponds to the addition of entered elements with the addition of another element, which is defined in terms of the characteristics of the subset and in terms of weight by means of the usual addition of all weights of the reference elements that are evaluated by the references. Accordingly, this weight can be calculated recursively across all references.
- the negative element results from the assigned exclusion features as a NOT element with negative weight.
- the zero element is the ⁇ All
- the change in the status of a development results indirectly from an evaluation of its references and corresponds to the multiplication by a number which, in terms of the evaluation of a reference as the usual and in terms of the characteristics of the reference development as the number of Subset formation is defined with itself, whereby the interpretation of a realization probability for the use of this reference for a specific development is advantageously assumed and the realization probability 0 is to be interpreted as equally likely.
- Multiplication by one thus corresponds to multiplication by identity and one by zero by that with the zero element.
- the fractional number corresponds to an areal realization probability for a specific development.
- the developments and the references or the reference developments are via a subset relation in the form of an interaction pair in the form ⁇ quantity
- the subset relation of each element determines a development system ⁇ solution
- the elements contain or are linked to tabular databases, which contain the verbal definition, the index of the reference elements, the time index, the user index as well as optionally the short title, the short description and other unchangeable information assigned to the elements.
- the dialog system with the user essentially operates with the verbal interaction pairs and the associative data structure resulting from the related developments, for analysis and management in the neural network itself, essentially with the assigned coordinates, which result from the relationships between the developments and their evaluation by mapping the dynamic associative data structure of the neural network in the Hilbert space.
- the analysis and administration also takes into account knowledge sets that have no direct relationship to one another if their coordinates meet the analysis or administration conditions. This enables computer-aided management and analysis of all the amounts of knowledge captured by the neural network in the same and unambiguous manner.
- Each set of knowledge assigned to a related development is characterized by its element of the neural network by exactly one coordinate vector, which is advantageously assigned to this directly or linked and is therefore part of the element.
- the amount of knowledge can be computationally managed by the neural network in addition to the relationships to one another via their coordinates, which in particular form the basis for mutual analyzes.
- Methods used in vector calculation analysis whether within a sphere, within a solid angle range, projection onto certain vectors) are used as analysis methods, which are based in particular on the calculation of a dot product.
- the calculation of vectors or knowledge sets with certain properties is possible, for example. about orthogonality using the SCHMIDT method.
- filters can optionally hide or re-balance certain elements.
- Quality> based on his personal knowledge a concrete development of this amount of knowledge or not, whereby he at least unconsciously endeavors to extremize a certain measure resulting from the application.
- These assignments form the training input data of the neural network. They naturally only have a low information content and a high level of redundancy among themselves.
- the neural network changes dynamically with every input and evaluation with regard to its associative data structure. As a result, the amount of knowledge described is accumulated statistically independently by the neural network via a large number of inputs from different users and is "noise-reduced" by necessitating a development that is repeatedly assigned in the same way.
- the defining sets of knowledge are linked to one another by means of directed references to previously entered reference developments and thus limit the described amounts of knowledge of the development.
- a concretization of the amount of knowledge of the reference developments also leads to a concretization of the amount of knowledge of the development itself, which was initially only limited by the reference category.
- the underlying associative data structure as such converges to the necessary ideal classification of knowledge.
- This neural network is therefore also suitable as the basis for an automatic classification system that is always dynamically adapted to the state of knowledge by generating it from the main references of the network structure.
- the specification of the reference limits each defined interaction pair to the amount of a respective reference system, which optionally also defines the type of development.
- An ideal reference is equivalent to an ideal reference system which is uniquely determined with all interaction pairs, i.e. the quantity is clearly assigned to the quality and vice versa. This is usually not the case with the real references that occur in practice, but the probability that all references with a given included term make use of the amount of knowledge of the reference system with a given realization is at a maximum with ideal references.
- the associative data structure converges to the multitude of inputs and evaluations true fact, which is formed by ideal references or ideal systems and has no redundancy without ever reaching it.
- the references are advantageously weighted standardized and provided with a status, optionally based on fuzzy probability information, which optionally itself consists of several standardized references weighted according to probabilities. This enables different types of rating groups to be filtered and reorganized.
- the development or the element is advantageously provided with a short title.
- the neural network automatically generates a thesaurus for quantities and / or qualities, which structures these terms hierarchically and evaluates them according to the weight of the elements using them.
- these descriptions and definitions, which represent the knowledge base of the neural network are researched in the usual way text-oriented and by means of linguistically meaningful links in context for terms.
- the NOT element designates precisely the knowledge quantity excluded from the knowledge set of the assigned element, which is required, for example, when forming elements with alternatives using the DE MORGAN formulas or with exclusion elements. It is favorable to generate the assigned NOT element immediately when generating an element. This can be distinguished favorably from the assigned element using a flag, for example the LSB of the time index.
- the references themselves are generated favorably by means of interval nesting to determine the memory area of the reference element assigned to the time index in the specific associative data structure and the one-way or two-way chaining of these by means of pointers.
- the backwards or forwards references optionally assigned to each individual element themselves are advantageously in turn managed via dynamic lists.
- the references themselves are advantageously provided with dynamically stored evaluations, which in turn are managed via dynamically linked lists.
- the evaluations are advantageously user-specific, as a result of which a similar and a statistically dependent multiple evaluation is advantageously avoided.
- a binary evaluation can be implemented cheaply in terms of storage technology via the LSB of the unique user index.
- a possible weighting can optionally be determined and taken into account via the user index.
- the time indices assigned during the entry or evaluation provide clear and verifiable assignments of the user to an entry, manipulations (time accumulations) for evaluation selectable and suppressible (time lock) as well as via an advantageously exponential. temporal decay function of the evaluations an evolutionary behavior of the associative data structure can be realized.
- the coordinate vector of each element can be determined recursively via the vector addition of all coordinate vectors of the reference elements that have been weighted. However, in order to optimize computing time, it is advantageous to manage it in a temporary dynamic memory area, the beginning or end of which is indicated by a coordinate pointer of the element.
- the coordinate vector which in turn is advantageously constructed as a dynamically linked list, it is again advantageous, in order to minimize the memory space required, to store only the coordinates required in each case and ordered according to their index.
- the Euclidean ones are suitable for the n-dimensional space, whereby each new development opens up a new dimension - these variables, which are important for analysis and administration, can be computationally simple be calculated. It is convenient to temporarily save the norm of each element with it.
- This neural network for knowledge management is included in particular in a central server, which is available to the majority of users free of charge via the Internet or optionally via the integration of a secure e-commerce system for information and financial services and corresponding factual entries or requests for knowledge served.
- the chronological stream used for input is generated by a user of the dialog system when the neural network is used, for example by defining the amount of knowledge to be entered via the input of related developments, making inquiries related to developments to the neural network, specifying new references or making evaluations . Since the associative data structure itself is always changed by the entered chronological stream, it is favorable to assign the time index of its last calculation to all optional temporary data areas of the various types of list elements.
- the calculation time index is compared with the respectively assigned time indices of these data areas. For individual calculation time indices that deviate from the current calculation time index, for example for special calculations in the past, it is advantageous to exclude the use of the temporary data areas, ie to recursively calculate the entire associative data structure.
- the process is based on a large number of entries or evaluations. which are entered into the associative data structure of the neural network by a large number of users, expanding and training, and thus modifying them via such inputs.
- the neural network itself ensures the convergence of the associative data structure via the implemented operators and carries out administrations and analyzes. In a figurative sense, the neural network compresses the input stream "noisy" due to non-factual information from various users to the true facts of low redundancy and ultimately forms an expert system, and the neural network analyzes developments in the Hilbert space of the elements, which are regarded as points and Have coordinates.
- the development is determined by references to one or more apparently obvious reference systems via interaction pairs and entered by the neural network as an element in the area of developments.
- the status of this reference is changed by a decision by a further user about the assignment of a new reference or the evaluation of the factual accuracy of the reference and thus generates another training input for the neural network.
- the references to the associative data structure from the neural network are successively adapted and the neural network is dynamized, which statistically corresponds to the ideal references converges and thus forms an expert system that has a decreasing redundancy.
- the associative data structure is always expanded somewhat and the neural network is trained.
- the latter is advantageously implemented via an implicit positive assessment of the amount of knowledge of a given reference development, since a reference represents a subjectively positive assessment of the development described by the interaction pair.
- explicit evaluations, preferably binary ones, and the specification of further references by third-party users are possible and favorable for the training of the neural network.
- special authorization e.g. due to a qualification or official appointment as a professor or examiner, an assessment can be carried out with an examiner-specific higher assessment weight (in the further exam).
- the entirety of the reference developments is advantageously checked to determine whether the development is contained in the individual sets of knowledge of the reference developments after subjective assessment by the examiner, that is, makes use of their teachings.
- the effect of evaluations of special countries, offices. Examiners, users, etc. can optionally be converted differently using special filters.
- Extended analysis options are favorable for service providers and separate neural networks for larger companies, which are connected to the central neural network via a communication link and optionally mirrored on one side.
- sovereign structures advantageously have special systems that are adapted to the respective special field of knowledge to describe the real world, for example, for patenting, macroeconomics, marketing and genetic engineering.
- gene sequences are read in with the assigned recognized bio functions, in macroeconomics individual farms and assigned fields of activity and in market analysis products and assigned customer needs.
- the extreme functional required as a measure for the convergence of these hierarchically diverging developments in accordance with RITZ's method is, as a law, corresponding to reality due to the absolute "level of invention” resulting from the curiosity of man about the inventive spirit the biological power of nature in the struggle for survival, given by the striving for financial wealth arising from the social order of man and the claims to life arising from the sluggishness of man.
- the stream used for input can alternatively or advantageously additionally be generated in pre-processing from content databases for the input of a user via a dialog system, in which knowledge of related developments is hidden and which through its largely uniform structuring and specific formulation of this knowledge with sufficient significance enable conversion from their context into the format of the stream via a computer-assisted automatic projection.
- Patent documentation in text format should prove to be particularly suitable, whereby the stated state of the art, the task and the essence of the invention or the characterizing part of the protection claims with their typical formulations, as well as their number assignment to particularly low reference numbers, are of particular importance. It is advantageous to use the usual search options and the thesaurus about the knowledge base of the neural network during preprocessing in order to filter out the relevant significant passages of text. These should always have a high level of specification with regard to their terms. It is also favorable to provide such knowledge with a reference to the underlying content of a content database, for example the patent number, and to note the seniority awarded, which is particularly useful for analyzes in the patent system. Another source for the automatic generation of streams should be the patent classification and technical lexica.
- the neural network For use in l'a 'essence, in which knowledge or developments are managed, in patents. Utility models or techn. Publications are published or optionally represent industrial property rights, the neural network carries out at least supporting computer-aided testing, official administration, monitoring, maintenance, research or processing of world knowledge in a classification system and, as a factual expert system, is superior to the usual tabular databases on patents and is advantageously linked to them .
- the inventor as a user specifies references for the definition of his development defined as interaction pair and computer-aided administration in the associative data structure, to which subtasks are assigned, which the inventor finally or retrospectively determines from the pool of the original short description, which uses an interaction pair Determine the reference system and be used backwards to define further reference systems, whereby the scope of protection of the development system is determined from the subset of all reference systems.
- the neural network automatically calculates an "amount of invention” and a "novelty" between individual or all registered developments from the Hilbert space function and the time indices, taking into account the status of each individual reference.
- the associative data structure is incrementally expanded or modified by third-party information or official testing as additional input for the neural network, with the development of the current state of knowledge in the factual testing leading to a reduction in redundancy over time.
- the status and optionally the databases of the documents have a link to registers and / or a clear reference to content databases, with optional documents which refer to other development systems referring to dependent documents of the same seniority of the documents. Certificates therefore only participate indirectly in the dynamics of the associative data structure via the dynamics of their reference documents, and their references are frozen, which means that the currently concretized associative data structure can be used with the exclusion of a retrospective approach.
- a computationally defined factual "novelty” with respect to the resulting logical term of the defining terms of the entirety of the significant references, as well as a computationally defined factual “inventive step” via functions such as the dot product can optionally be calculated at a given time index compared to the associative data structure, e.g. as an orthogonal distance from the trend of all the nearest developments, without making a conclusive statement about these legal terms.
- a reference to the assigned property right or the official register with the legal status would also make sense.
- each development as an element of the associative data structure refers to the previously known reference developments that define the individual characteristics of the development via significantly weighted references
- the definition structure of a development can be broken down recursively. Due to an advantageous dynamic of the data structure, which converges due to the constant use of insufficient redundancy and thus a higher truth content, this definition structure largely follows the current state of knowledge.
- the definitions of first-order reference development (parents) should generally suffice to sufficiently disclose the level of knowledge used for a later user of the development as an innovation, but the definitions of second-order reference development (grandparents) and other orders can also be used. With that the Supports users in the preparation of a patent application with sufficient disclosure content.
- Such a provisional registration can be registered immediately with the relevant IPR authorities by priority using modern telecommunication means such as FAX and, if possible later, via data connections.
- a signature is not required immediately, but can be replaced by an electronic signature if necessary.
- an optional grace period for publication of the development entry in the data structure is provided for up to 18 months and up to this point in time it is securely encrypted and personalized to the user.
- This invention is advantageously integrated into a computer-assisted knowledge management system for research and analysis in the factual-technical field, in particular for patent documents, which can be accessed via the Internet / intranet and can optionally be paid for via e-commerce.
- the user with an intellectual implementation in the form of a concrete development operates this knowledge management system in the following rough steps:
- the knowledge management system offers the user a selection of previously known developments, with textually matching terms being emphasized.
- the knowledge management system gradually refines the selection and the user the definition.
- the knowledge management system uses AI research to determine the state of development in the knowledge space and mathematically calculates obvious and similar developments.
- the knowledge management system shows the user the closer development environment, makes it possible to hike over their references, the complete display of all text fields and optionally gives a link to the source documents of the respective developments, for example to patent documents.
- the user can estimate the innovation potential and, if necessary, change the development by redefining it.
- the user optionally enters the title and brief description of his entry as well as his personal identification and address in order to secure his authorship and to enable a provisional registration of property rights based on priority.
- the knowledge management system enters this entry individually encrypted in the database, whereby this entry only becomes publicly available after a grace period has expired. This is optionally followed by the computer-assisted generation of a provisional patent application and its immediate transmission to the IP office.
- the knowledge management system also enables computer-aided analysis of developments in relation to one another, as is customary in patenting, in particular support for analysis of property rights infringements, especially regarding developments from legally valid property rights.
- the development E is ultimately formed, for example, with three references B via: AND B1 OR B2 AND NOT B3, B2 having a source document XXX
- the user enters this development supplemented by his personal information NAME, ADDRESS and KS, KB with a publication grace period of 18 months ensured by personal encryption in the data structure.
- the KL system uses mathematical algorithms to calculate a sufficient level of invention and confirms the novelty, so that an application to register an intellectual property right is recommended to the user, who agrees with the proposal.
- the computer then generates the following provisional application for industrial property rights and sends it via a communication link to an industrial property rights office, whereby the text blocks KURSIV are inserted for clarification:
- the invention denotes a KT.
- KTBl with KLBl, which is KABl.
- KTB2 with KLB2
- KAB2 is known from publication XXX.
- KTB3 with KLB3, which is KAB3.
- KTE KAE is appropriate.
- KLE according to the protection claim solves this task.
- the KTE is advantageously carried out as a KLBl in that DLBl causes this DABl.
- KBBl OR the KTE is carried out by DLB2 causing this DAB2.
- AND NOT the KTE is carried out by DLB3 causing this DAB3.
- KBB3. This makes the KTE KAE. CFU.
- KTE which is designed as KLBl OR KLB2 and NOT as KLB3, characterized in that it is designed as KLE.
- the invention designates a KTE.
- the KTE KAE is appropriate.
- provisional application for industrial property rights establishes a filing date, the priority of which can be claimed for a subsequent application that complies with customary practice.
- the successfully transmitted provisional application for industrial property rights is issued as a document-proof document, e.g. printed out.
- a neural network 1 for computer-aided knowledge management consists of elements 2 (usually referred to as neurons or nodes) which are related and weighted, which form a Hilbert space 3 due to their special properties and which are preferably assigned a coordinate vector 4 which is associated with the root element is advantageously selected [0].
- each individual element 2 of the corresponding detail in FIG. 1 is assigned a meaning which is designed as a special form of defining the development as a pair of interactions 5. and preferably describes a quantity 6 and a quality 7, and thus precisely defines the quantity of knowledge 8, which maps the quantity 6 to the quality 7 and the quality 7 to the quantity 6 via their subset.
- These sets of knowledge 8, 10 defined there are linked to one another via directed references 9 to reference developments already entered and thus limit the described set of knowledge 11 of the development relating to them.
- a concretization of the knowledge sets 8, 10 of the reference developments thereby also leads to a concretization of the knowledge set 11 of the development itself, initially limited only by the reference category 12.
- the stream 13 used for input is used in the form of a data stream, which contains the definition in the form of an interaction pair 5 and the references 9 to elements 2 that have already been entered in strict chronological order, for incremental expansion of the neural network, in that in the linear memory area 14 of the computer dynamically inserted via linked lists 15 managed associative data structure new structural elements 16 and linked via computer-specific pointers 17.
- an element 2 and a NOT element 18 form a pair 19 which points to the interaction pair 5 and advantageously contains a coordinate list 20 which contains the coordinates of the element 2, which correspond inverted to the coordinates of the NOT element.
- a back reference list 21 and a preheating reference list 22 are advantageously assigned to each element 2 and each NOT element 18, with the reference lists 21, 22 advantageously being assigned rating lists 23.
- a dialog system enables the input or the request for developments via the definition 25 of the verbal interaction pair and the specification of references via an intuitive surface 24.
- the neural network advantageously offers a quasi-simultaneous selection 26 of elements 2 of developments already recorded in the knowledge base with similar terms contained in the definitions 25, from which individual references to the terms used can be made via a switch 27.
- the offered selection 26 of developments one can be marked, for which the local environment 28 of the directly connected or the obvious elements 2 of the associative data structure is displayed, advantageously graphically, showing the distance between individual developments and their short titles.
- a sub-marking can be used to hike between the individual developments, the definition 25 and brief description 29 of these being displayed.
- the interaction pair and the references are signed and integrated into the associative data structure of the neural network via the stream.
- the dialog system in a similar or customary manner, intuitively enables the input or output of a login, a filter, a text-based search, a recursive input of more general developments for the pre-definition of terms, the input of additional references and ratings, the output of the Structural data of the registered, directly related developments, in particular coordinate details of the element for the analysis of the characteristics of the development, specification of their assigned mass, for example "inventions calculated with the aid of dot products.
- the associative data structure consists of elements (here documents and certificates), which are structured from older documents based on references to characteristics A. B, ... and are significantly evaluated for the initialization TO (31).
- a necessary final reference C to document II and the associated non-disclosed document (I) is entered.
- the document I ' represents a dependent document I' on the document I.
- T2 for example after the examination of a later registered document III, with regard to the document II for an equivalent characteristic C an older reference C than that of the reference document IV significant, which results in a devaluation of the old reference C and a changed associative data structure.
- KBE A common steel screw does not form rust if its surface is covered with an electroplated nickel coating.
- KLE nickel screw
- KBB 1 A screw has a positive locking screw
- KBB2 A screw can be screwed into an internal thread with its spiral valve
- DLB2 cylinder jacket is shaped like a spiral
- KAB2 can be screwed in
- KBB3 As a propeller, its rotation creates a movement of flooding media KLB3 propellers
- a screw with a threaded pin which is a rotationally positive connection, is also known.
- a screw with a spiral screw which can be screwed in from the publication XXX, is also known
- the anti-rust screw is stainless
- a screw with a threaded lock is essentially used, which is a positive locking OR
- Screw with spiral screw which can be screwed in AND NOT a drive screw with a propeller, which produces more flow than a cover screw by removing an electroplated coating
- Nickel on a threaded screw causes it to not rust
- the anti-rust screw is advantageously designed as a screw, in that the pin with an external thread ensures that this screwing movement is the sole degree of freedom.
- a screw has a twist-locking thread thread OR
- the anti-rust screw is carried out in the cylinder jacket is spiral Molded causes this possibility of rotation with axial translation.
- a screw can be screwed into an internal thread with its spiral valve AND NOT the anti-rust screw is carried out by screw-shaped drive means for fluid media causing this rotation to produce movement of a fluid.
- a drive screw produces a propeller with its rotation a movement of fluid media This means that the anti-rust screw is rust-proof
- a common steel screw does not rust if its surface is provided with a galvanically applied nickel coating
- Antrostatic screw which is designed as a screw or as a screw and not as a drive screw, characterized in that it is designed as a nickel screw
- the invention relates to an anti-rust screw.
- the anti-rust screw is rustproof.
- a conventional steel screw does not form rust if its surface is provided with a galvanically applied nickel coating
Abstract
Description
Claims
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP00943636A EP1222625B1 (de) | 1999-05-24 | 2000-05-24 | Neuronales netz zum computergestützten wissensmanagement |
AU58041/00A AU5804100A (en) | 1999-05-24 | 2000-05-24 | Neuronal network for computer-assisted knowledge management |
DE50012539T DE50012539D1 (de) | 1999-05-24 | 2000-05-24 | Neuronales netz zum computergestützten wissensmanagement |
US09/979,684 US7092857B1 (en) | 1999-05-24 | 2000-05-24 | Neural network for computer-aided knowledge management |
DE10081401T DE10081401D2 (de) | 1999-05-24 | 2000-05-24 | Neuronales Netz zum Computergestützten Wissensmanagement |
Applications Claiming Priority (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE19923622A DE19923622A1 (de) | 1998-08-31 | 1999-05-24 | Neuronales Netz zum rechnergestützten Wissenmanagement |
DE19923622.4 | 1999-05-24 | ||
EP99116789A EP0984364A3 (de) | 1998-08-31 | 1999-08-31 | Neuronales Netz zum computergestützten Wissensmanagement |
EP99116789.1 | 1999-08-31 | ||
DE19964094A DE19964094A1 (de) | 1999-12-31 | 1999-12-31 | Neuronales Netz zur Unterstützung von Innovationen |
DE19964094.7 | 1999-12-31 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2000072256A2 true WO2000072256A2 (de) | 2000-11-30 |
WO2000072256A3 WO2000072256A3 (de) | 2002-04-04 |
Family
ID=27219157
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/DE2000/001669 WO2000072256A2 (de) | 1999-05-24 | 2000-05-24 | Neuronales netz zum computergestützten wissensmanagement |
Country Status (6)
Country | Link |
---|---|
US (1) | US7092857B1 (de) |
EP (1) | EP1222625B1 (de) |
AT (1) | ATE322723T1 (de) |
AU (1) | AU5804100A (de) |
DE (2) | DE10081401D2 (de) |
WO (1) | WO2000072256A2 (de) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AT501028A1 (de) * | 2001-09-27 | 2006-05-15 | Siemens Ag Oesterreich | Verfahren und vorrichtung zur bewertung von börsekursen |
DE10291392B4 (de) * | 2001-02-15 | 2007-08-16 | Metalife Ag | Verfahren, System und Datenträger zur Erzeugung von Korrelationen und/oder Interaktionen und/oder Wissen aus einer Vielzahl von durchsuchten Datensätzen |
US8992227B2 (en) | 2001-04-28 | 2015-03-31 | Manchester Metropolitan University | Methods and apparatus for analysing the behaviour of a subject |
EP2909756A1 (de) | 2012-10-19 | 2015-08-26 | Patent Analytics Holding Pty Ltd | System und verfahren zur präsentation und visuellen navigation durch netzwerkdatensätze |
CN108431832A (zh) * | 2015-12-10 | 2018-08-21 | 渊慧科技有限公司 | 利用外部存储器扩增神经网络 |
Families Citing this family (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8005740B2 (en) | 2002-06-03 | 2011-08-23 | Research Affiliates, Llc | Using accounting data based indexing to create a portfolio of financial objects |
US8374951B2 (en) | 2002-04-10 | 2013-02-12 | Research Affiliates, Llc | System, method, and computer program product for managing a virtual portfolio of financial objects |
US7587352B2 (en) | 2002-04-10 | 2009-09-08 | Research Affiliates, Llc | Method and apparatus for managing a virtual portfolio of investment objects |
US7792719B2 (en) * | 2004-02-04 | 2010-09-07 | Research Affiliates, Llc | Valuation indifferent non-capitalization weighted index and portfolio |
US7747502B2 (en) | 2002-06-03 | 2010-06-29 | Research Affiliates, Llc | Using accounting data based indexing to create a portfolio of assets |
US8374937B2 (en) | 2002-04-10 | 2013-02-12 | Research Affiliates, Llc | Non-capitalization weighted indexing system, method and computer program product |
US7620577B2 (en) * | 2002-06-03 | 2009-11-17 | Research Affiliates, Llc | Non-capitalization weighted indexing system, method and computer program product |
US8589276B2 (en) | 2002-06-03 | 2013-11-19 | Research Afiliates, LLC | Using accounting data based indexing to create a portfolio of financial objects |
AU2003258556A1 (en) * | 2003-08-01 | 2005-03-16 | Centrum Fur Ertragsoptimierung Aktiengesellschaft | Measuring method and pattern recognition machine for identifying a vector characteristic of business management of a subject of knowledge and method and machine for automatically characterizing a subject of knowledge from the point of view of business management |
US9015093B1 (en) | 2010-10-26 | 2015-04-21 | Michael Lamport Commons | Intelligent control with hierarchical stacked neural networks |
US8775341B1 (en) | 2010-10-26 | 2014-07-08 | Michael Lamport Commons | Intelligent control with hierarchical stacked neural networks |
WO2013028935A1 (en) * | 2011-08-23 | 2013-02-28 | Research Affiliates, Llc | Using accounting data based indexing to create a portfolio of financial objects |
CN103064008B (zh) * | 2012-12-26 | 2016-01-20 | 浙江大学 | 一种基于希尔伯特-黄变换的非线性模拟电路软故障诊断方法 |
US9785886B1 (en) | 2017-04-17 | 2017-10-10 | SparkCognition, Inc. | Cooperative execution of a genetic algorithm with an efficient training algorithm for data-driven model creation |
US10963790B2 (en) | 2017-04-28 | 2021-03-30 | SparkCognition, Inc. | Pre-processing for data-driven model creation |
US10410121B2 (en) * | 2017-10-25 | 2019-09-10 | SparkCognition, Inc. | Adjusting automated neural network generation based on evaluation of candidate neural networks |
US20190138890A1 (en) * | 2017-11-08 | 2019-05-09 | Ping Liang | Expandable and real-time recofigurable hardware for neural networks and logic reasoning |
CN108763445B (zh) * | 2018-05-25 | 2019-09-17 | 厦门智融合科技有限公司 | 专利知识库的构建方法、装置、计算机设备和存储介质 |
US10685286B1 (en) | 2019-07-30 | 2020-06-16 | SparkCognition, Inc. | Automated neural network generation using fitness estimation |
US20230019194A1 (en) * | 2021-07-16 | 2023-01-19 | Dell Products, L.P. | Deep Learning in a Virtual Reality Environment |
CN116938644A (zh) * | 2022-03-31 | 2023-10-24 | 华为技术有限公司 | 一种信号处理方法和装置 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0378115A2 (de) * | 1989-01-06 | 1990-07-18 | Hitachi, Ltd. | Neuronaler Rechner |
US4967369A (en) * | 1988-06-07 | 1990-10-30 | Jourjine Alexander N | Variable capacity transmission medium |
EP0493292A2 (de) * | 1990-12-11 | 1992-07-01 | International Business Machines Corporation | Verfahren und Gerät mit Vorausschau für vorhersagendes Wählen mit einem neuronalen Netzwerk |
US5410716A (en) * | 1989-03-28 | 1995-04-25 | Mitterauer; Bernhard | Computation system for the simulation of the cerebral cortex |
US5659666A (en) * | 1994-10-13 | 1997-08-19 | Thaler; Stephen L. | Device for the autonomous generation of useful information |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0468229A3 (en) * | 1990-07-27 | 1994-01-26 | Hnc Inc | A neural network with expert system functionality |
US6400996B1 (en) * | 1999-02-01 | 2002-06-04 | Steven M. Hoffberg | Adaptive pattern recognition based control system and method |
DE19715503A1 (de) * | 1997-04-14 | 1998-10-15 | Siemens Ag | Integriertes Rechner- und Kommunikationssystem für den Anlagenbereich |
GB2344446A (en) * | 1998-12-02 | 2000-06-07 | Mars Inc | Classifying currency items |
-
2000
- 2000-05-24 DE DE10081401T patent/DE10081401D2/de not_active Expired - Lifetime
- 2000-05-24 US US09/979,684 patent/US7092857B1/en not_active Expired - Fee Related
- 2000-05-24 EP EP00943636A patent/EP1222625B1/de not_active Expired - Lifetime
- 2000-05-24 AT AT00943636T patent/ATE322723T1/de not_active IP Right Cessation
- 2000-05-24 DE DE50012539T patent/DE50012539D1/de not_active Expired - Lifetime
- 2000-05-24 AU AU58041/00A patent/AU5804100A/en not_active Abandoned
- 2000-05-24 WO PCT/DE2000/001669 patent/WO2000072256A2/de active IP Right Grant
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4967369A (en) * | 1988-06-07 | 1990-10-30 | Jourjine Alexander N | Variable capacity transmission medium |
EP0378115A2 (de) * | 1989-01-06 | 1990-07-18 | Hitachi, Ltd. | Neuronaler Rechner |
US5410716A (en) * | 1989-03-28 | 1995-04-25 | Mitterauer; Bernhard | Computation system for the simulation of the cerebral cortex |
EP0493292A2 (de) * | 1990-12-11 | 1992-07-01 | International Business Machines Corporation | Verfahren und Gerät mit Vorausschau für vorhersagendes Wählen mit einem neuronalen Netzwerk |
US5659666A (en) * | 1994-10-13 | 1997-08-19 | Thaler; Stephen L. | Device for the autonomous generation of useful information |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE10291392B4 (de) * | 2001-02-15 | 2007-08-16 | Metalife Ag | Verfahren, System und Datenträger zur Erzeugung von Korrelationen und/oder Interaktionen und/oder Wissen aus einer Vielzahl von durchsuchten Datensätzen |
US8992227B2 (en) | 2001-04-28 | 2015-03-31 | Manchester Metropolitan University | Methods and apparatus for analysing the behaviour of a subject |
AT501028A1 (de) * | 2001-09-27 | 2006-05-15 | Siemens Ag Oesterreich | Verfahren und vorrichtung zur bewertung von börsekursen |
AT501028B1 (de) * | 2001-09-27 | 2008-08-15 | Siemens Ag Oesterreich | Verfahren und vorrichtung zur bewertung von börsekursen |
EP2909756A1 (de) | 2012-10-19 | 2015-08-26 | Patent Analytics Holding Pty Ltd | System und verfahren zur präsentation und visuellen navigation durch netzwerkdatensätze |
US9767160B2 (en) | 2012-10-19 | 2017-09-19 | Patent Analytics Holding Pty Ltd | System and method for presentation and visual navigation of network data sets |
US10565215B2 (en) | 2012-10-19 | 2020-02-18 | Patent Analytics Holding Pty Ltd | System and method for presentation and visual navigation of network data sets |
CN108431832A (zh) * | 2015-12-10 | 2018-08-21 | 渊慧科技有限公司 | 利用外部存储器扩增神经网络 |
Also Published As
Publication number | Publication date |
---|---|
DE10081401D2 (de) | 2002-12-05 |
US7092857B1 (en) | 2006-08-15 |
ATE322723T1 (de) | 2006-04-15 |
EP1222625A2 (de) | 2002-07-17 |
EP1222625B1 (de) | 2006-04-05 |
AU5804100A (en) | 2000-12-12 |
DE50012539D1 (de) | 2006-05-18 |
WO2000072256A3 (de) | 2002-04-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2000072256A2 (de) | Neuronales netz zum computergestützten wissensmanagement | |
EP1783633B1 (de) | Suchmaschine für eine ortsbezogene Suche | |
DE112018005725T5 (de) | Daten-deidentifikation auf der grundlage eines erkennens von zulässigen konfigurationen für daten-deidentifikationsprozesse | |
DE10356399B4 (de) | Datenverarbeitungssystem | |
DE60032258T2 (de) | Bestimmen ob eine variable numerisch oder nicht numerisch ist | |
EP1264253B1 (de) | Verfahren und anordnung zur modellierung eines systems | |
DE19963123B4 (de) | Analytisches Informationssystem | |
DE102021127398A1 (de) | Beziehungserkennung und -quantifizierung | |
WO2002061615A2 (de) | Rechnersystem | |
WO2021190715A1 (de) | Computerimplementiertes verfahren und verteiltes speichersystem zum bereitstellen vertrauenswürdiger datenobjekte | |
EP0984364A2 (de) | Neuronales Netz zum computergestützten Wissensmanagement | |
EP1094405A2 (de) | Verfahren zum Erzeugen einer dynamischen Auswahlmaske für den Abruf von Daten aus einer Datenbank | |
WO2001071577A2 (de) | Verfahren zum auffinden von objekten | |
Yu | Improving Knowledge Accessibility on the Web-from Knowledge Base Augmentation to Search as Learning | |
EP1376402A2 (de) | System zur Visualisierung eines Portfolios | |
EP1098273A2 (de) | Elektronisches Kartenvorverkaufssystem | |
DE19940990C2 (de) | Netzwerk | |
EP0833263A2 (de) | Datenbank, Verfahren zum Ableiten einer Datenbank und Verfahren zur Abfrage einer Datenbank | |
DE19964094A1 (de) | Neuronales Netz zur Unterstützung von Innovationen | |
DE102019200474A1 (de) | Eingabesystem und Verfahren zur Sicherstellung einheitlicher Begriffskonventionen in einer für mehrere Benutzer zugänglichen Anwendung | |
WO2004095313A1 (de) | Datenverarbeitungssystem für benutzerfreundliche datenbank-recherchen | |
Haffner | Request prediction and hyperlink proposals: methodologies and mathematics behind web applications | |
DE102017007331A1 (de) | Verfahren zur Mensch-Maschine-Interaktion in Bewertungssysemen | |
DE102022124990A1 (de) | Datenbank eines Rechners | |
EP3716578A1 (de) | Verfahren und eine vorrichtung zum ansteuern eines technischen geräts mit einem optimalen modell |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A2 Designated state(s): AE AL AM AT AU AZ BA BB BG BR BY CA CH CN CR CU CZ DE DK DM EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT TZ UA UG US UZ VN YU ZA ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A2 Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 09979684 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2000943636 Country of ref document: EP |
|
AK | Designated states |
Kind code of ref document: A3 Designated state(s): AE AL AM AT AU AZ BA BB BG BR BY CA CH CN CR CU CZ DE DK DM EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT TZ UA UG US UZ VN YU ZA ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A3 Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG |
|
WWP | Wipo information: published in national office |
Ref document number: 2000943636 Country of ref document: EP |
|
REF | Corresponds to |
Ref document number: 10081401 Country of ref document: DE Date of ref document: 20021205 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 10081401 Country of ref document: DE |
|
NENP | Non-entry into the national phase |
Ref country code: JP |
|
WWG | Wipo information: grant in national office |
Ref document number: 2000943636 Country of ref document: EP |