Methods and devices for customizing knowledge representation systems
Techniques for customizing knowledge representation systems including identifying, based on a plurality of concepts in a knowledge representation (KR), a group of one or more concepts relevant to user context information, and providing the identified group of one more concepts to a user. The KR may include a combination of modules. The modules may include a kernel and a customized module customized for the user. The kernel may accessible via a second KR.
1. A system for supporting machine-based inductive reasoning to predict new complex knowledge representations encoded as computer-readable data and stored on one or more tangible, non-transitory computer-readable storage media, the system comprising:
one or more processors configured to execute a synthesis engine, an analysis engine, an inference engine, and a statistical engine, wherein the synthesis engine, the inference engine, and the statistical engine operate together to perform synthesis of a predicted knowledge representation from a first knowledge representation, the first and predicted knowledge representations encoded as first and second computer-readable data structures, respectively, the first knowledge representation comprising a plurality of concepts and at least one relationship between the concepts, wherein the synthesis includes:
the synthesis engine synthesizing one or more probable concepts or probable concept relationships not explicitly encoded in the first knowledge representation, the one or more probable concepts or probable concept relationships being encoded in the predicted knowledge representation;
the inference engine applying one or more inference rules to reference data stored in the one or more tangible, non-transitory computer-readable storage media to infer a relationship between a first concept in the first knowledge representation and a second concept in the first knowledge representation or between the first concept in the first knowledge representation and a concept in the reference data, wherein the reference data is composed of one or more natural language documents, audio recordings, and audiovisual recordings; and
the statistical engine applying one or more probabilities associated with the plurality of concepts to the reference data;
wherein the analysis engine analyzes the predicted knowledge representation of the one or more probable concepts or probable concept relationships based on an output of at least one of the inference engine and the statistical engine;
and wherein one or more analyzed concepts or analyzed concept relationships in the predicted knowledge representation are stored as an output knowledge representation.
2. The system of claim 1 , wherein the synthesis engine is provided at least one of: context information, and a user model associated with one or more data consumers.
3. The system of claim 1 , wherein the synthesis engine is configured to synthesize a complex knowledge representation based at least in part on user preference information.
4. The system of claim 1 , wherein the synthesis engine applies constructive rules based on at least one of: formal concept analysis, faceted classification synthesis, dynamic taxonomies, atomic knowledge representation model (AKRM), and knowledge processing rules.
5. The system of claim 1 , wherein the inference engine applies at least one of: a logical inference rule, a linguistic inference rule, and a semantic inference rule.
6. The system of claim 1 , wherein the statistical engine is configured to generate a statistical graphical model based on probabilities associated with concepts and concept relationships in reference data.
7. The system of claim 1 , wherein the statistical engine applies a statistical inference technique to compute the semantic coherence or relevance between concepts and concept relationships.
8. The system of claim 1 , wherein the analysis engine applies an analytical method of at least one of: text analysis, entity and information extraction, information retrieval, data mining, classification, statistical clustering, linguistic analysis, facet analysis, and natural language processing.
9. The system of claim 1 , wherein the analysis engine is configured to deconstruct input complex knowledge representations into elemental data structures.
10. A computer-implemented method of supporting machine-based inductive reasoning to predict new complex knowledge representations encoded as computer-readable data and stored on one or more tangible, non-transitory computer-readable storage media, the method comprising:
storing a first knowledge representation encoded as a first computer-readable data structure and comprising a plurality of a concepts and at least one relationship between the plurality of concepts;
performing synthesis of a predicted knowledge representation from the first knowledge representation using a synthesis engine, an analysis engine, an inference engine, and a statistical engine, the predicted knowledge representation encoded as a second computer-readable data structure, the synthesis including:
synthesizing, by the synthesis engine, one or more probable concepts or probable concept relationships not explicitly encoded in the first knowledge representation, the one or more probable concepts or probable concept relationships being encoded in the predicted knowledge representation;
applying, by the inference engine, one or more inference rules to reference data stored in the one or more tangible, non-transitory computer-readable storage media to infer a relationship between a first concept in the first knowledge representation and a second concept in the first knowledge representation or between the first concept in the first knowledge representation and a concept in the reference data, wherein the reference data is composed of one or more natural language documents, audio recordings, and audiovisual recordings;
applying, by the analysis engine, one or more probabilities associated with the plurality of concepts to the reference data; and
wherein the analysis engine analyzes the predicted knowledge representation of the one or more probable concepts or probable concept relationships based on an output of at least one of the inference engine and the statistical engine;
and wherein one or more analyzed concepts and analyzed concept relationships in the predicted knowledge representation are stored as an output knowledge representation.
11. The method of claim 10 , wherein the synthesis engine is provided at least one of: context information, and a user model associated with one or more data consumers.
12. The method of claim 10 , wherein the synthesis engine is configured to synthesize a complex knowledge representation based at least in part on user preference information.
13. The method of claim 10 , wherein the synthesis engine applies constructive rules based on at least one of: formal concept analysis, faceted classification synthesis, dynamic taxonomies, atomic knowledge representation model (AKRM), and knowledge processing rules.
14. The method of claim 10 , wherein the inference engine applies at least one of: a logical inference rule, a linguistic inference rule, and a semantic inference rule.
15. The method of claim 10 , wherein the statistical engine is configured to generate a statistical graphical model based on probabilities associated with concepts and concept relationships in reference data.
16. The method of claim 10 , wherein the statistical engine applies a statistical inference technique to compute the semantic coherence or relevance between concepts and concept relationships.
17. The method of claim 10 , wherein the analysis engine applies an analytical method of at least one of: text analysis, entity and information extraction, information retrieval, data mining, classification, statistical clustering, linguistic analysis, facet analysis, and natural language processing.
18. The method of claim 10 , wherein the analysis engine is configured to deconstruct input complex knowledge representations into elemental data structures.
19. A non-transitory computer readable media encoding instructions for performing a method comprising:
storing a first knowledge representation encoded as a first computer-readable data structure and comprising a plurality of a concepts and at least one relationship between the plurality of concepts;
performing synthesis of a predicted knowledge representation from the first knowledge representation using a synthesis engine, an analysis engine, an inference engine, and a statistical engine, the predicted knowledge representation encoded as a second computer-readable data structure, the synthesis including:
synthesizing, by the synthesis engine, one or more probable concepts or probable concept relationships not explicitly encoded in the first knowledge representation, the one or more probable concepts or probable concept relationships being encoded in the predicted knowledge representation;
applying, by the inference engine, one or more inference rules to reference data stored in the one or more tangible, non-transitory computer-readable storage media to infer a relationship between a first concept in the first knowledge representation and a second concept in the first knowledge representation or between the first concept in the first knowledge representation and a concept in the reference data, wherein the reference data is composed of one or more natural language documents, audio recordings, and audiovisual recordings;
applying, by the analysis engine, one or more probabilities associated with the plurality of concepts to the reference data; and
wherein the analysis engine analyzes the predicted knowledge representation of the one or more probable concepts or probable concept relationships based on an output of at least one of the inference engine and the statistical engine;
and wherein one or more analyzed concepts and analyzed concept relationships in the predicted knowledge representation are stored as an output knowledge representation.
20. The non-transitory computer readable media of claim 19 , wherein the synthesis engine is provided at least one of: context information, and a user model associated with one or more data consumers.