SYSTEM AND METHOD FOR THE INDEXING AND RETRIEVAL OF SEMANTICALLY ANNOTATED DATA USING AN ONTOLOGY-BASED INFORMATION RETRIEVAL MODEL

• US 20160179945A1
• Filed: 12/19/2014
• Published: 06/23/2016
• Est. Priority Date: 12/19/2014
• Status: Active Grant

• Current Assignee: Universidad Nacional de Educación a Distancia
• Original Assignee: Universidad Nacional de Educación a Distancia
• Inventors: LASTRA DIAZ, Juan Jose, GARCIA SERRANO, Ana
• Granted Patent: US 10,509,814 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24522 Translation of natural language queries to structured queries
• CPC Class Codes: G06F 16/24573 using data annotations, e.g. user-defined metadata
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/284 Relational databases
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06F 16/367 Ontology
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

System and method for the indexing and retrieval of semantically annotated information units from a collection of semantically annotated indexed information units in response to a query using an ontology-based IR model. The retrieval method comprises: receiving a semantically annotated query with semantic annotations to individuals or classes within a determined populated base ontology; embedding, as a set of weighted-mentions to individuals or classes within the populated base ontology, the semantically annotated query in a semantic representation space of an ontology-based metric space IR model; obtaining the representation in the semantic representation space for every indexed information unit; computing the Hausdorff distance between the space representation of the query and the space representation of all the indexed information units of the collection; retrieving and ranking, the relevant information units based on the computed Hausdorff distance.

Claims

1. A computer-implemented method for retrieving semantically relevant information units from a collection of semantically annotated indexed information units in response to a query, the method comprising:

receiving, by a computer system, a semantically annotated query, the semantically annotated query including a set of semantic annotations to individuals or classes within a determined populated base ontology;

embedding, by the computer system, the semantically annotated query in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the indexed information units, the semantically annotated query being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

obtaining, by the computer system, the representation in the semantic representation space for every indexed information unit of the collection;

computing, by the computer system, the Hausdorff distance between the space representation of the query and the space representation of all the indexed information units of the collection;

retrieving and ranking, by the computer system, the relevant information units based on the computed Hausdorff distance.

2. The computer-implemented method of claim 1, further comprising:

receiving, by the computer system, an input query in natural language;

converting, by the computer system, the input query into the semantically annotated query with regard to the populated base ontology.

3. The computer-implemented method of claim 1, wherein a mention to a class in the semantically annotated query is considered as a reference to a full class, the mentioned full class subsuming all the descendant individuals and classes within the populated base ontology.

4. The computer-implemented method of claim 1, wherein the step of computing the Hausdorff distance comprises:

retrieving information-content values of the populated base ontology;

retrieving pre-computed pair-wise semantic distances between all the classes within the base ontology;

computing all pair-wise semantic distances between the weighted-mentions of the query and the weighted-mentions of the indexed information units of the collection, using the information-content values and the pre-computed pair-wise semantic distances according to the metric of the representation space.

5. The computer-implemented method of claim 4, wherein the information-content values are retrieved by accessing a populated ontology repository.

6. The computer-implemented method of claim 4, wherein the information-content values of the populated base ontology include IC values for each ontology node computed using an intrinsic node-based IC-computation method.

7. The computer-implemented method of claim 4, wherein the information-content values of the populated base ontology include edge-based IC weights for each ontology edge, computed using an intrinsic edge-based IC-computation method.

8. The computer-implemented method of claim 4, wherein the pre-computed pair-wise semantic distances are retrieved by accessing a file.

9. The computer-implemented method of claim 4, wherein the pair-wise semantic distances between two ontology nodes are computed as the shortest weighted-path of the populated base ontology considering all possible paths between said two nodes, and that the edges of the taxonomy can be traversed in any direction.

10. The computer-implemented method of claim 4, wherein the computation of pair-wise semantic distances between two ontology nodes includes the computation of a weighted distance value which is the sum of the edge weights for all the edges of the populated base ontology along the shortest path joining said two ontology nodes.

11. The computer-implemented method of claim 10, wherein the edge weight between two adjacent nodes of the populated base ontology is the information-content value of the joint probability between said two adjacent nodes, wherein the sum of all the joint probabilities for the subsumed concepts of any parent concept is equal to 1.

12. The computer-implemented method of claim 11, wherein the information-content value of the joint probability between two adjacent nodes is computed as the negative binary logarithm of the joint probability.

13. The computer-implemented method of claim 10, wherein the edge weight between two adjacent nodes of the populated base ontology is the information-content value of the child node minus the information-content value of the parent node.

14. The computer-implemented method of claim 1, wherein the information units are text documents, web pages, sentences, multimedia objects or any sort of data that can be represented as a collection of classes or individuals within an ontology.

15. The computer-implemented method of claim 1, wherein the ontological representation space defined by the ontology-based IR model satisfies the following structure-preserving axioms:

order invariance, metric invariance and inclusion invariance.

18. The system of claim 15, wherein a mention to a class in the semantically annotated query is considered by the processor as a reference to a full class, the mentioned full class subsuming all the descendant individuals and classes within the populated base ontology.

16. A semantic search system for retrieving semantically relevant information units from a collection of semantically annotated indexed information units in response to a query, the semantic search system comprising a processor and a memory coupled with and readable by the processor and storing a set of instructions which, when executed by the processor, causes the processor to:

receive a semantically annotated query, the semantically annotated query including a set of semantic annotations to individuals or classes within a determined populated base ontology;

embed the semantically annotated query in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the indexed information units, the semantically annotated query being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

obtain the representation in the semantic representation space for every indexed information unit of the collection;

compute the Hausdorff distance between the space representation of the query and the space representation of all the indexed information units of the collection;

retrieve and rank the relevant information units based on the computed Hausdorff distance.

17. The system of claim 16, wherein the processor is further configured to:

receive an input query in natural language;

convert the input query into the semantically annotated query with regard to the populated base ontology.

19. The system of claim 16, wherein the computing of the Hausdorff distance comprises:

retrieving information-content values of the populated base ontology;

retrieving pre-computed pair-wise semantic distances between all the classes within the base ontology;

computing all pair-wise semantic distances between the weighted-mentions of the query and the weighted-mentions of the indexed information units of the collection, using the information-content values and the pre-computed pair-wise semantic distances according to the metric of the representation space.

20. The system of claim 19, wherein the information-content values are retrieved by accessing a populated ontology repository.

21. The system of claim 19, wherein the information-content values of the populated base ontology include IC values for each ontology node computed using an intrinsic node-based IC-computation method.

22. The system of claim 19, wherein the information-content values of the populated base ontology include edge-based IC weights for each ontology edge, computed using an intrinsic edge-based IC-computation method.

23. The system of claim 19, wherein the pre-computed pair-wise semantic distances are retrieved by accessing a file.

24. The system of claim 19, wherein the pair-wise semantic distances between two ontology nodes are computed as the shortest weighted-path of the populated base ontology considering all possible paths between said two nodes, and that the edges of the taxonomy can be traversed in any direction.

25. The system of claim 19, wherein the computation of pair-wise semantic distances between two ontology nodes includes the computation of a weighted distance value which is the sum of the edge weights for all the edges of the populated base ontology along the shortest path joining said two ontology nodes.

26. The system of claim 25, wherein the edge weight between two adjacent nodes of the populated base ontology is the information-content value of the joint probability between said two adjacent nodes, wherein the sum of all the joint probabilities for the subsumed concepts of any parent concept is equal to 1.

27. The system of claim 26, wherein the information-content value of the joint probability between two adjacent nodes is computed as the negative binary logarithm of the joint probability.

28. The system of claim 25, wherein the edge weight between two adjacent nodes of the populated base ontology is the information-content value of the child node minus the information-content value of the parent node.

29. The system of claim 16, wherein the information units are text documents, web pages, sentences, multimedia objects or any sort of data that can be represented as a collection of classes or individuals within an ontology.

30. The system of claim 16, wherein the ontological representation space defined by the ontology-based IR model satisfies the following structure-preserving axioms:

order invariance, metric invariance and inclusion invariance.

31. A computer-readable memory for retrieving semantically relevant information units from a collection of semantically annotated indexed information units in response to a query, the computer-readable memory comprising a set of instructions stored therein which, when executed by a processor, causes the processor to:

receive a semantically annotated query, the semantically annotated query including a set of semantic annotations to individuals or classes within a determined populated base ontology;

embed the semantically annotated query in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the indexed information units, the semantically annotated query being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

obtain the representation in the semantic representation space for every indexed information unit of the collection;

compute the Hausdorff distance between the space representation of the query and the space representation of all the indexed information units of the collection;

retrieve and rank the relevant information units based on the computed Hausdorff distance.

32. A computer-implemented method for indexing semantically annotated information units into a search system based on an ontology-based information retrieval model, the method comprising:

receiving, by a computer system, a semantically annotated information unit, the semantically annotated information unit including a set of semantic annotations to individuals or classes within a determined populated base ontology and the frequency of said semantic annotations within the information unit;

embedding, by the computer system, the semantically annotated information unit in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the information units, the semantically annotated information unit being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

storing the set of weighted-mentions of the indexed information unit in an information units repository;

storing the semantic annotations of the information unit in a populated ontology repository.

33. The computer-implemented method of claim 32, further comprising:

receiving, by the computer system, the information unit in natural language;

converting, by the computer system, the information unit into the semantically annotated information unit with regard to the populated base ontology.

34. The computer-implemented method of claim 32, wherein the step of storing the semantic annotations of the information unit in a populated ontology repository comprises:

inserting the new mentions to individuals and classes in the populated base ontology;

annotating the mentions to classes and individuals with inverted indexes to the indexed information units;

updating the information-content values for the registered individuals.

35. The computer-implemented method of claim 34, wherein the information-content values for the registered individuals are node-based IC values computed using an intrinsic node-based IC-computation method.

36. The computer-implemented method of claim 34, wherein the information-content values for the registered individuals are edge-based IC values computed using an intrinsic edge-based IC-computation method.

37. The computer-implemented method of claim 36, wherein the edge-based IC values are computed according to the updated joint probability between the individuals and their parent concepts.

38. The computer-implemented method of claim 32, wherein the ontological representation space defined by the ontology-based IR model satisfies the following structure-preserving axioms:

order invariance, metric invariance and inclusion invariance.

39. A system for indexing semantically annotated information units into a search system based on an ontology-based information retrieval model, the indexing system comprising a processor and a memory coupled with and readable by the processor and storing a set of instructions which, when executed by the processor, causes the processor to:

receive a semantically annotated information unit, the semantically annotated information unit including a set of semantic annotations to individuals or classes within a determined populated base ontology and the frequency of said semantic annotations within the information unit;

embed the semantically annotated information unit in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the information units, the semantically annotated information unit being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

store the set of weighted-mentions of the indexed information unit in an information units repository;

store the semantic annotations of the information unit in a populated ontology repository.

40. The system of claim 39, wherein the processor is further configured to:

receive the information unit in natural language;

convert the information unit into the semantically annotated information unit with regard to the populated base ontology.

41. The system of claim 39, wherein the storing the semantic annotations of the information unit in a populated ontology repository comprises:

inserting the new mentions to individuals and classes in the populated base ontology;

annotating the mentions to classes and individuals with inverted indexes to the indexed information units;

updating the information-content values for the registered individuals.

42. The system of claim 41, wherein the information-content values for the registered individuals are node-based IC values computed using an intrinsic node-based IC-computation method.

43. The system of claim 41, wherein the information-content values for the registered individuals are edge-based IC values computed using an intrinsic edge-based IC-computation method.

44. The system of claim 43, wherein the edge-based IC values are computed according to the updated joint probability between the individuals and their parent concepts.

45. The system of claim 39, wherein the ontological representation space defined by the ontology-based IR model satisfies the following structure-preserving axioms:

order invariance, metric invariance and inclusion invariance.

46. A computer-readable memory for indexing semantically annotated information units into a search system based on an ontology-based information retrieval model, the computer-readable memory comprising a set of instructions stored therein which, when executed by a processor, causes the processor to:

receive a semantically annotated information unit, the semantically annotated information unit including a set of semantic annotations to individuals or classes within a determined populated base ontology and the frequency of said semantic annotations within the information unit;

embed the semantically annotated information unit in a semantic representation space of an ontology-based IR model that uses a metric space for the representation of the information units, the semantically annotated information unit being embedded as a set of weighted-mentions to individuals or classes within the populated base ontology;

store the set of weighted-mentions of the indexed information unit in an information units repository;

store the semantic annotations of the information unit in a populated ontology repository.