System and method for coreference resolution
According to one aspect, a method for coreference resolution is provided. In one embodiment, the method includes receiving a segment of text that includes mentions corresponding to entities. A first feature vector is generated based on one or more features associated with a first mention, and a second feature vector is generated based on based on one or more features associated with a second mention. A measure of similarity between the first feature vector and second feature vector is computed and, based on the computed measure of similarity, it is determined if the first mention and the second mention both correspond to the same entity.
1. A computer-implemented method for coreference resolution, comprising:
receiving a segment of text comprising a plurality of mentions corresponding to entities;
generating a first feature vector based on at least one feature associated with a first mention;
generating a second feature vector based on at least one feature associated with a second mention;
computing a measure of similarity between the first feature vector and second feature vector; and
based on the computed measure of similarity, determining if the first mention and the second mention both correspond to the same entity.
2. The computer-implemented method of claim 1 , wherein the at least one feature associated with the first mention, and the at least one feature associated with the second mention, comprises at least one of assertions, associations, same-sentence features, same-document features, temporal features and geolocation features.
3. The computer-implemented method of claim 1 , wherein each of the entities is associated with a level of a hierarchical structure defining a plurality of entity levels.
4. The computer-implemented method of claim 3 , wherein at least one of computing the measure of similarity between the first feature vector and second feature vector and determining if the first mention and the second mention both correspond to the same entity comprises dynamically organizing mentions based on the hierarchical structure.
5. The computer-implemented method of claim 4 , wherein dynamically organizing the mentions comprises dimensional reduction.
6. The computer-implemented method of claim 5 , wherein the dimensional reduction comprises semantic hashing.
7. The computer-implemented method of claim 1 , wherein at least one of the features associated with the first mention and the second mention has an assigned weight that is higher or lower than an assigned weight of at least one of the other features.
8. The computer-implemented method of claim 1 , wherein the measure of similarity represents a degree or amount by which the first mention and second mention both correspond to the same entity.
9. The computer-implemented method of claim 8 , wherein determining if the first mention and the second mention both correspond to the same entity comprises
determining if the degree or amount by which the first mention and second mention both correspond to the same entity exceeds a predetermined threshold, and
upon determining that the degree or amount by which the first mention and second mention both correspond to the same entity exceeds the predetermined threshold, identifying the first mention and second mention as corresponding to the same entity.
10. A system for coreference resolution, comprising:
a processing unit;
a memory operatively coupled to the processing unit; and
a program module which executes in the processing unit from the memory and which, when executed by the processing unit, causes the computer system to
receive a segment of text comprising a plurality of mentions corresponding to entities;
generate a first feature vector based on at least one feature associated with at least one first mention;
generate a second feature vector based on at least one feature associated with at least one second mention;
compute a measure of similarity between the first feature vector and second feature vector; and
based on the computed measure of similarity, determine if the at least one first mention and the at least one second mention both correspond to the same entity.
11. The system of claim 10 , wherein the at least one feature associated with the at least one first mention, and the at least one feature associated with the at least one second mention, comprise at least one of assertion features, association features, same-sentence features, same-document features, temporal features, and geolocation features.
12. The system of claim 10 , wherein each of the entities is associated with a level of a hierarchical structure defining a plurality of entity levels.
13. The system of claim 12 , wherein at least one of computing the measure of similarity between the first feature vector and second feature vector and determining if the at least one first mention and the at least one second mention both correspond to the same entity comprises dynamically organizing the at least one first mention and at least one second mention based on the hierarchical structure.
14. The system of claim 13 , wherein dynamically organizing the at least one first mention and at least one second mention comprises aggregating the at least one first mention and at least one second mention upward from a lower level of the hierarchical structure to a higher level of the hierarchical structure.
15. The system of claim 14 , wherein dynamically organizing the at least one first mention and at least one second mention comprises dimensional reduction.
16. The system of claim 15 , wherein the dimensional reduction comprises semantic hashing.
17. The system of claim 10 , wherein the at least one first mention comprises a plurality of first mentions and the at least one second mention comprises a plurality of second mentions.
18. The system of claim 10 , wherein at least one of the features associated with the at least one first mention and the at least one second mention has an assigned weight that is higher or lower than an assigned weight of at least one of the other features.
19. The system of claim 10 , wherein the measure of similarity represents a degree or amount by which the at least one first mention and at least one second mention both correspond to the same entity.
20. The system of claim 19 , wherein determining if the at least one first mention and the at least one second mention both correspond to the same entity comprises
determining if the degree or amount by which the at least one first mention and the at least one second mention both correspond to the same entity exceeds a predetermined threshold, and
upon determining that the degree or amount by which the at least one first mention and the at least one second mention both correspond to the same entity exceeds the predetermined threshold, identifying the at least one first mention and at least one second mention as corresponding to the same entity.
21. The system of claim 10 , wherein the at least one feature associated with the at least one first mention and the at least one feature associated with the at least one second mention comprises at least one of parts of speech, titles, first characters, next characters, prefixes, and suffixes.
22. A computer-readable storage medium having computer-executable instructions stored thereon which, when executed by a computer, cause the computer to:
receive a segment of text comprising a plurality of mentions corresponding to entities;
generate at least one first feature vector based on at least one feature associated with a plurality of first mentions;
generate at least one second feature vector based on at least one feature associated with a plurality of second mentions;
compute a measure of similarity between the at least one first feature vector and at least one second feature vector; and
based on the computed measure of similarity, determine if at least one of the plurality of first mentions and at least one of the plurality of second mentions both correspond to the same entity.
23. The computer-readable storage medium of claim 22 , wherein the at least one feature associated with the plurality of first mentions, and the at least one feature associated with the plurality of second mentions, comprises at least one of assertions, associations, same-sentence features, same-document features, temporal features and geolocation features.
24. The computer-readable storage medium of claim 22 , wherein each of the entities is associated with a level of a hierarchical structure defining a plurality of entity levels.
25. The computer-readable storage medium of claim 24 , wherein at least one of computing the measure of similarity between the first feature vector and second feature vector and determining if the at least one first mention and the at least one second mention both correspond to the same entity comprises dynamically organizing the at least one first mention and at least one second mention based on the hierarchical structure.
26. The computer-readable storage medium of claim 25 , wherein dynamically organizing the at least one first mention and at least one second mention comprises dimensional reduction by semantic hashing.