Synonym identification based on categorical contexts
View Patent ↗Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training recognition canonical representations corresponding to named-entity phrases in a second natural language based on translating a set of allowable expressions with canonical representations from a first natural language, which may be generated by expanding a context-free grammar for the allowable expressions for the first natural language.
1. A computer-implemented method comprising:
identifying a category from a knowledge base, the category including a plurality of entities;
identifying a set of queries including terms related to the entities included in the category from a query stream;
calculating a specificity value for the category, based at least on the entities included in the category and the set of queries, comprising:
determining a probability distribution for co-occurrence of the terms in the set of queries with the entities associated with the category;
determining a background probability distribution for the occurrence of the terms in the set of queries; and
comparing the probability distribution of the co-occurrence of the terms in the set of queries with the entities associated with the category and the background probability distribution to obtain the specificity value for the category; and
classifying the category as a coherent category, wherein a coherent category is a category whose specificity value satisfies a threshold.
2. The method of claim 1 , wherein calculating the specificity value for the category uses a Kullback-Leibler (KL) divergence between the probability distribution for co-occurrence of the terms and the background probability distribution.
3. The method of claim 1 , wherein the terms comprise individual words.
4. The method of claim 1 , wherein the terms comprise individual words and aggregated sub-phrases.
5. The method of claim 1 , further comprising determining that the category is a useful category when it is a coherent category and the category is associated with a number of entities that satisfies a threshold.
6. The method of claim 1 , further comprising determining that the category is a useful category when it is a coherent category and the set of queries contains a number of queries that satisfies a threshold.
7. A system comprising:
one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
identifying a category from a knowledge base, the category including a plurality of entities;
identifying a set of queries including terms related to the entities included in the category from a query stream;
calculating a specificity value for the category, based at least on the entities included in the category and the set of queries, comprising:
determining a probability distribution for co-occurrence of the terms in the set of queries with the entities associated with the category;
determining a background probability distribution for the occurrence of the terms in the set of queries; and
comparing the probability distribution of the co-occurrence of the terms in the set of queries with the entities associated with the category and the background probability distribution to obtain the specificity value for the category; and
classifying the category as a coherent category, wherein a coherent category is a category whose specificity value satisfies a threshold.
8. The system of claim 7 , wherein calculating the specificity value for the category uses a Kullback-Leibler (KL) divergence between the probability distribution for co-occurrence of the terms and the background probability distribution.
9. The system of claim 7 , wherein the terms comprise individual words.
10. The system of claim 7 , wherein the terms comprise individual words and aggregated sub-phrases.
11. The system of claim 7 , further comprising determining that the category is a useful category when it is a coherent category and the category is associated with a number of entities that satisfies a threshold.
12. The system of claim 7 , further comprising determining that the category is a useful category when it is a coherent category and the set of queries contains a number of queries that satisfies a threshold.
13. A non-transitory computer-readable storage medium storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to perform operations comprising:
identifying a category from a knowledge base, the category including a plurality of entities;
identifying a set of queries including terms related to the entities included in the category from a query stream;
calculating a specificity value for the category, based at least on the entities included in the category and the set of queries, comprising:
determining a probability distribution for co-occurrence of the terms in the set of queries with the entities associated with the category;
determining a background probability distribution for the occurrence of the terms in the set of queries; and
comparing the probability distribution of the co-occurrence of the terms in the set of queries with the entities associated with the category and the background probability distribution to obtain the specificity value for the category; and
classifying the category as a coherent category, wherein a coherent category is a category whose specificity value satisfies a threshold.
14. The medium of claim 13 , wherein calculating the specificity value for the category uses a Kullback-Leibler (KL) divergence between the probability distribution for co-occurrence of the terms and the background probability distribution.
15. The medium of claim 13 , wherein the terms comprise individual words.
16. The medium of claim 13 , wherein the terms comprise individual words and aggregated sub-phrases.
17. The medium of claim 13 , further comprising determining that the category is a useful category when it is a coherent category and the category is associated with a number of entities that satisfies a threshold.
18. The medium of claim 13 , further comprising determining that the category is a useful category when it is a coherent category and the set of queries contains a number of queries that satisfies a threshold.