System and method for topic extraction and opinion mining

Patent metadata

Field	Value
Publication number	US-10339184-B2
Application number	US-201615357018-A
Country	US
Kind code	B2
Filing date	Nov 21, 2016
Priority date	Sep 28, 2009
Publication date	Jul 2, 2019
Grant date	Jul 2, 2019

Abstract

Techniques for topic extraction and opinion mining are described. For example, a machine accesses a document from a document record of a database. The machine builds a syntax tree of a sentence of the document based on parsing the sentence. The machine assigns a polarity impact to one or more words of the sentence. The machine determines that a plurality of words in the sentence have conflicting polarity based on the syntax tree and the polarity impact of the plurality of words. The machine determines a classification of the sentence based on the determining that the plurality of words in the sentence have conflicting polarity. The machine generates a classification record of the classification in the database.

First claim

What is claimed is:

1. A system comprising:
one or more hardware processors; and
a machine-readable medium for storing instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising:
accessing a document from a document record of a database;
building a syntax tree of a sentence of the document based on parsing the sentence;
assigning a polarity impact to one or more words of the sentence based on matching one or more lexical patterns and the one or more words of the sentence, a lexical pattern being a token sequence, each token of the token sequence comprising a lemma field, a polarity tag field, and a part-of-speech tag field;
determining that a plurality of words in the sentence have conflicting polarity based on the syntax tree and the assigned polarity impact of the plurality of words;
determining a classification of the sentence based on the determining that the plurality of words in the sentence have conflicting polarity; and
generating a classification record of the classification in the database.

2. The system of claim 1 , wherein the operations further comprise identifying one or more key phrases in the document based on a filtering technique.

3. The system of claim 2 , wherein the operations further comprise identifying the sentence of the document based on the identified one or more key phrases, the sentence including the one or more key phrases.

4. The system of claim 1 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a polarity word having a dominant impact on a topic.

5. The system of claim 1 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a sum of polarities method.

6. The system of claim 1 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a syntactic distance between the one or more words and a topic key phrase in the syntactic tree.

7. The system of claim 1 , wherein the determining of the classification of the sentence includes selecting a polarity classification for the one or more words in the sentence, the selecting resulting in classified one or more words.

8. The system of claim 7 , wherein the determining of the classification of the sentence further includes applying one or more heuristic rules to the classified one or more words and the sentence.

9. The system of claim 1 , wherein the operations further comprise determining an opinion associated with a document that includes the sentence based on the classification of the sentence.

10. A method comprising:
accessing a document from a document record of a database;
building, using one or more hardware processors, a syntax tree of a sentence of the document based on parsing the sentence;
assigning a polarity impact to one or more words of the sentence based on matching one or more lexical patterns and the one or more words of the sentence, a lexical pattern being a token sequence, each token of the token sequence comprising a lemma field, a polarity tag field, and a part-of-speech tag field;
determining that a plurality of words in the sentence have conflicting polarity based on the syntax tree and the assigned polarity impact of the plurality of words;
determining a classification of the sentence based on the determining that the plurality of words in the sentence have conflicting polarity; and
generating a classification record of the classification in the database.

11. The method of claim 10 , further comprising:
identifying one or more key phrases in the document based on a filtering technique.

12. The method of claim 11 , further comprising:
identifying the sentence of the document based on the identified one or more key phrases, the sentence including the one or more key phrases.

13. The method of claim 10 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a polarity word having a dominant impact on a topic.

14. The method of claim 10 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a sum of polarities method.

15. The method of claim 10 , wherein the assigning of the polarity impact to the one or more words of the sentence is further based on a syntactic distance between the one or more words and a topic key phrase in the syntactic tree.

16. The method of claim 10 , wherein the determining of the classification of the sentence includes selecting a polarity classification for the one or more words in the sentence, the selecting resulting in classified one or more words.

17. The method of claim 16 , wherein the determining of the classification of the sentence further includes applying one or more heuristic rules to the classified one or more words and the sentence.

18. A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more hardware processors of a machine, cause the one or more hardware processors to perform operations comprising:
accessing a document from a document record of a database;
building a syntax tree of a sentence of the document based on parsing the sentence;
assigning a polarity impact to one or more words of the sentence based on matching one or more lexical patterns and the one or more words of the sentence, a lexical pattern being a token sequence, each token of the token sequence comprising a lemma field, a polarity tag field, and a part-of-speech tag field;
determining that a plurality of words in the sentence have conflicting polarity based on the syntax tree and the assigned polarity impact of the plurality of words;
determining a classification of the sentence based on the determining that the plurality of words in the sentence have conflicting polarity; and
generating a classification record of the classification in the database.

Assignees

• Ebay Inc

Inventors

• Sundaresan Neelakantan
• Zhang Yongzheng
• Baudin Catherine
• Shen Dan
• Huang Shen

Classifications

• G06F40/30

  Semantic analysis · CPC title

• G06F16/3344

  using natural language analysis · CPC title

• G06F16/93 Primary

  Document management systems · CPC title

• G06F16/5846

  using extracted text · CPC title

• G06F16/36

  Creation of semantic tools, e.g. ontology or thesauri · CPC title