Method and device for extracting entity relation based on deep learning, and server

1. A method for extracting entity relation based on deep learning, the method comprising:

preprocessing a text to be mined, to obtain a sentence with entities in the text to be mined;

determining an entity pair in the sentence with the entities, wherein the entity pair comprises at least two entities, and determining candidate relations between the entities in the entity pair; and

determining an entity relation between the entities in the entity pair from the candidate relations,

wherein the determining an entity relation between the entities in the entity pair from the candidate relations comprises:

calculating confidence levels of the candidate relations, and determining the entity relation from the candidate relations according to the confidence levels of the candidate relations.

2. The method of claim 1 , wherein preprocessing a text to be mined comprises:

punctuating the text to be mined; and

performing lexical analysis and syntactic analysis on a punctuated sentence obtained by punctuating, and identifying entities in the punctuated sentence, to obtain the sentence with the entities.

3. The method of claim 1 , wherein determining an entity pair in the sentence comprises:

determining all candidate entity pairs in the sentence; and

screening the candidate entity pairs according to a filtering condition, to obtain the entity pair in the sentence.

4. The method of claim 3 , wherein determining all candidate entity pairs in the sentence comprises:

identifying all entities in the sentence; and

forming the candidate entity pairs by selecting any two entities of all entities identified.

5. The method of claim 3 , wherein the filtering condition is determined according to at least one of:

a distance between the entities of the entity pair in a dependency tree,

the entities of the entity pair being core words of noun phrases in the sentence, and

parts of speech of the entities of the entity pair.

6. The method of claim 1 , wherein determining candidate relations between entities in the entity pair comprises:

extracting the candidate relations between the entities of the entity pair in the sentence according an extracting strategy.

7. The method of claim 6 , wherein the extracting strategy comprises:

extracting all nouns and verbs under the entity pair in a relation tree.

8. The method of claim 1 , wherein determining an entity relation between the entities in the entity pair from the candidate relations comprises:

applying the entity pair and the candidate relations of the entity pair to an entity relation determining model, to obtain output results corresponding to the respective candidate relations; and

in response to the output result of the entity relation determining model being a positive example, calculating the confidence level of the candidate relation, and determining the entity relation of the entity pair according to the confidence level.

9. A device for extracting entity relation based on deep learning, the device comprising:

a preprocessing module, configured for preprocessing a text to be mined, to obtain a sentence with entities in the text to be mined;

a determining module, configured for determining an entity pair in the sentence according to the entities, wherein the entity pair comprises at least two entities, and determining candidate relations between the entities in the entity pair; and

a processing module, configured for determining an entity relation between the entities in the entity pair from the candidate relations,

wherein the processing module is further configured for calculating confidence levels of the candidate relations from the candidate relations, and determining the entity relation according to the confidence levels of the candidate relations.

10. The device of claim 9 , wherein the preprocessing module comprises:

a punctuating unit, configured for punctuating the text to be mined; and

an analyzing unit configured for performing lexical analysis and syntactic analysis on a punctuated sentence obtained by punctuating, and identifying entities in the punctuated sentence, to obtain the sentence with the entities.

11. The device of claim 9 , wherein the determining module comprises:

a determining unit, configured for determining all candidate entity pairs in the sentence; and

a screening unit, configured for screening the candidate entity pairs according to a filtering condition, to obtain the entity pair in the sentence.

12. The device of claim 9 , wherein the determining module further comprises:

an extracting unit, configured for extracting the candidate relations between the entities of the entity pair in the sentence according an extracting strategy.

13. The device of claim 9 , wherein the processing module comprises:

an obtaining unit, configured for applying the entity pair and the candidate relations of the entity pair to an entity relation determining model, to obtain output results corresponding to the respective candidate relations; and

a calculating unit, configured for, in response to the output result of the entity relation determining model being a positive example, calculating the confidence level of the candidate relation, and determining the entity relation of the entity pair according to the confidence level.

14. A server, comprising:

one or more processors; and

a storage device for storing one or more programs;

wherein the one or more processors execute the one or more programs to implement the method of claim 1 .

15. A non-volatile computer-readable storage medium, in which a computer program is stored, wherein the program, when executed by a processor, implements the method of claim 1 .