Systems and methods for entity resolution using attributes from structured and unstructured data

1. A computer-implemented method, comprising:

obtaining unstructured text data including a plurality of references corresponding to entities, wherein the unstructured text data is not pre-arranged with a predefined data model or schema;

determining, from the unstructured text data, attributes associated with the entities;

obtaining structured data including predefined attributes associated with the entities;

comparing attributes associated with a first coreference unit with attributes associated with a second coreference unit, wherein the first coreference unit is a sub-entity representation having the attributes determined from the unstructured text data and the second coreference unit is a sub-entity representation having the predefined attributes; and

determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity.

2. The method of claim 1 , wherein the sub-entity representation is an aggregate of chains of coreferent mentions in unstructured text.

3. The method of claim 1 , wherein the structured data including the predefined attributes is obtained from a structured entity data store or is determined from a predefined lexicon.

4. The method of claim 1 , wherein the attributes comprise at least one of biographic data, descriptive data, and transactional data, wherein the biographic data is associated with immutable attributes of an entity, the descriptive data is associated with changeable attributes of an entity, and the transactional data is associated with a relationship between two entities.

5. The method of claim 1 , wherein determining, from the unstructured text data, the attributes associated with the entities comprises extracting the attributes from the unstructured text data based on identifying one or more predetermined language patterns.

6. The method of claim 1 , wherein comparing the attributes of the first coreference unit with the attributes of the second coreference unit comprises computing the similarity between the first coreference unit and the second coreference unit.

7. The method of claim 6 , wherein computing the similarity comprises:

forming, from features that include attributes determined from the unstructured text data, a first feature vector representation of the first coreference unit;

forming, from features that include attributes determined from the structured entity data, a second feature vector representation of the second coreference unit; and

computing a similarity between the first feature vector and the second feature vector.

8. The method of claim 1 , wherein determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity comprises increasing or decreasing a score representing the likelihood that the first coreference unit and the second coreference unit both correspond to the same entity.

9. The method of claim 1 , wherein determining whether the first coreference unit and the second coreference unit both correspond to the same entity comprises determining whether the attributes associated with the first coreference unit and the attributes associated with the second coreference unit both include a particular, unique entity identifier, and wherein the method further includes:

in response to determining that the attributes associated with the first coreference unit and the second coreference unit both include the unique entity identifier, resolving the first coreferent unit and the second coreference unit to a single entity having the unique entity identifier.

10. The method of claim 1 , further comprising:

in response to determining that an attribute from a first data source conflicts with an attribute from a second data source, resolving the conflict at least in part by selecting the attribute from the data source that has a higher source confidence than the other data source.

11. The method of claim 10 , wherein the source confidence is determined based on the accuracy of data previously obtained from the respective first data source or second data source.

12. The method of claim 1 , further comprising:

in response to determining that the first coreference unit and the second coreference unit both correspond to the same entity, resolving the first coreference unit and the second coreference unit to the same entity.

13. A system, comprising:

one or more processors; and

a memory device operatively coupled to the one or more processors and storing computer-executable instructions which, when executed by the one or more processors, cause the system to perform functions that include:

obtaining unstructured text data including a plurality of references corresponding to entities, wherein the unstructured text data is not pre-arranged with a predefined data model or schema;

determining, from the unstructured text data, attributes associated with the entities;

obtaining structured data including predefined attributes associated with the entities;

comparing attributes associated with a first coreference unit with attributes associated with a second coreference unit, wherein the first coreference unit is a sub-entity representation having the attributes determined from the unstructured text data and the second coreference unit is a sub-entity representation having the predefined attributes; and

determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity.

14. The system of claim 13 , wherein the sub-entity representation is an aggregate of chains of coreferent mentions in unstructured text.

15. The system of claim 13 , wherein the structured data including the predefined attributes is obtained from a structured entity data store or is determined from a predefined lexicon.

16. The system of claim 13 , wherein the attributes comprise at least one of biographic data, descriptive data, and transactional data, wherein the biographic data is associated with immutable attributes of an entity, the descriptive data is associated with changeable attributes of an entity, and the transactional data is associated with a relationship between two entities.

17. The system of claim 13 , wherein determining, from the unstructured text data, the attributes associated with the entities comprises extracting the attributes from the unstructured text data based on identifying one or more predetermined language patterns.

18. The system of claim 13 , wherein comparing the attributes of the first coreference unit with the attributes of the second coreference unit comprises computing the similarity between the first coreference unit and the second coreference unit, and wherein computing the similarity comprises:

forming, from features that include attributes determined from the unstructured text data, a first feature vector representation of the first coreference unit;

forming, from features that include attributes determined from the structured entity data, a second feature vector representation of the second coreference unit; and

computing a similarity between the first feature vector and the second feature vector.

19. The system of claim 13 , wherein determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity comprises increasing or decreasing a score representing the likelihood that the first coreference unit and the second coreference unit both correspond to the same entity.

20. The system of claim 13 , wherein the performed functions further include:

in response to determining that an attribute from a first data source conflicts with an attribute from a second data source, resolving the conflict at least in part by selecting the attribute from the data source that has a higher source confidence than the other data source, wherein the source confidence is determined based on the accuracy of data previously obtained from the respective first data source or second data source.

21. The system of claim 13 , wherein the performed functions further include:

in response to determining that the first coreference unit and the second coreference unit both correspond to the same entity, resolving the first coreference unit and the second coreference unit to the same entity.

22. A non-transitory computer-readable medium storing instructions which, when executed by one or more processors, cause a computer to perform functions that include:

obtaining unstructured text data including a plurality of references corresponding to entities, wherein the unstructured text data is not pre-arranged with a predefined data model or schema;

determining, from the unstructured text data, attributes associated with the entities;

obtaining structured data including predefined attributes associated with the entities;

comparing attributes associated with a first coreference unit with attributes associated with a second coreference unit, wherein the first coreference unit is a sub-entity representation having the attributes determined from the unstructured text data and the second coreference unit is a sub-entity representation having the predefined attributes; and

determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity.

23. The non-transitory computer-readable medium of claim 22 , wherein the sub-entity representation is an aggregate of chains of coreferent mentions in unstructured text.

24. The non-transitory computer-readable medium of claim 22 , wherein the structured data including the predefined attributes is obtained from a structured entity data store or is determined from a predefined lexicon.

25. The non-transitory computer-readable medium of claim 22 , wherein the attributes comprise at least one of biographic data, descriptive data, and transactional data, wherein the biographic data is associated with immutable attributes of an entity, the descriptive data is associated with changeable attributes of an entity, and the transactional data is associated with a relationship between two entities.

26. The non-transitory computer-readable medium of claim 25 , wherein the stored instructions further comprise instructions which, when executed by the one or more processors, cause the computer to perform further functions that include:

in response to determining that the first coreference unit and the second coreference unit both correspond to the same entity, resolving the first coreference unit and the second coreference unit to the same entity.

27. The non-transitory computer-readable medium of claim 22 , wherein determining, from the unstructured text data, the attributes associated with the entities comprises extracting the attributes from the unstructured text data based on identifying one or more predetermined language patterns.

28. The non-transitory computer-readable medium of claim 22 , wherein comparing the attributes of the first coreference unit with the attributes of the second coreference unit comprises computing the similarity between the first coreference unit and the second coreference unit, and wherein computing the similarity comprises:

forming, from features that include attributes determined from the unstructured text data, a first feature vector representation of the first coreference unit;

forming, from features that include attributes determined from the structured entity data, a second feature vector representation of the second coreference unit; and

computing a similarity between the first feature vector and the second feature vector.

29. The non-transitory computer-readable medium of claim 22 , wherein determining, based on the comparison, whether the first coreference unit and the second coreference unit both correspond to the same entity comprises increasing or decreasing a score representing the likelihood that the first coreference unit and the second coreference unit both correspond to the same entity.

30. The non-transitory computer-readable medium of claim 22 , wherein the stored instructions further comprise instructions which, when executed by the one or more processors, cause the computer to perform further functions that include:

in response to determining that an attribute from a first data source conflicts with an attribute from a second data source, resolving the conflict at least in part by selecting the attribute from the data source that has a higher source confidence than the other data source, wherein the source confidence is determined based on the accuracy of data previously obtained from the respective first data source or second data source.