NLP-based systems and methods for providing quotations

• US 9,092,416 B2
• Filed: 01/31/2014
• Issued: 07/28/2015
• Est. Priority Date: 03/30/2010
• Status: Active Grant

• Current Assignee: Fiver LLC
• Original Assignee: VCVC III LLC (Vulcan, Inc.)
• Inventors: Koperski, Krzysztof, Dhillon, Navdeep S., Liang, Jisheng
• Primary Examiner(s): VO, HUYEN X
• Application Number:
  US14/169,972
• Publication Number:
  US 20140149110A1
• Time in Patent Office: 543 Days
• Field of Search: 704 1- 10, 704/250, 704/251, 704/255, 704/257, 704/270, 704/270.1, 715/723
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30

Abstract

Techniques for providing quotations obtained from text documents using natural language processing techniques are described. Some embodiments provide a content recommendation system (“CRS”) configured to provide quotations by extracting quotations from a corpus text documents, and providing access to the extracted quotations in response to search requests received from users. The CRS may extract quotations by using natural language processing-based techniques to identify one or more entities, such as people, places, objects, concepts, or the like, that are referenced by the extracted quotations. The CRS may then store the extracted quotations along with identified entities, such as quotation speakers and subjects, for later access via search requests.

Claims

1. A method in a content recommendation system, the method comprising:

under control of a computing system having a computer processor, using the computer processor to provide quotation information by automatically,

under control of the computer processor, extracting a quotation from a text document in a corpus of text documents indexed by the computing system;

under control of the computer processor, identifying one or more entities that are referenced by the text document, each of the determined entities being electronically represented by the content recommendation system, wherein identifying the one or more entities includes linking together multiple mentions of a same entity across the text document, the linking including resolving pronoun coreference to identify an entity that is a speaker of the quotation;

under control of the computer processor, attributing the quotation to the speaker of the quotation by storing data that associates the quotation with the identified entity that is the speaker of the quotation; and

under control of the computer processor, providing the quotation by transmitting text that represents the quotation and the attributed speaker.

2. The method of claim 1 wherein extracting the quotation includes dividing the text document into linguistic units including at least one of:

a sentence, a phrase, a clause, and/or a paragraph.

3. The method of claim 1 wherein extracting the quotation includes performing parts-of-speech tagging on sentences of the text document.

4. The method of claim 1 wherein extracting the quotation includes performing lexical analysis to identify phrases in the text document.

5. The method of claim 1 wherein extracting the quotation includes determining grammatical roles including subjects, verbs, and objects of sentences of the text document.

6. The method of claim 1 wherein extracting the quotation includes detecting punctuation indicating a beginning or end of the quotation.

7. The method of claim 1 wherein extracting the quotation includes detecting a quotation verb by determining whether a verb in a sentence of the text document is one of a predetermined list of quotation verbs.

8. The method of claim 1 wherein the linking further includes resolving references and abbreviations, and resolving definite noun or pronoun anaphora.

9. The method of claim 1 wherein identifying one or more entities includes assigning a type or facet to each of the one or more entities.

10. The method of claim 1 wherein identifying one or more entities includes recognizing and disambiguating a document reference to an entity, such that the reference is linked to a representation of the entity stored in an entity repository.

11. The method of claim 1 wherein providing the quotation includes presenting at least one of:

a quotation;

an attribution;

a context including text surrounding a quotation;

or document metadata including title, publication date, and/or author.

12. The method of claim 1 wherein extracting the quotation includes determining a speaker, verb, quote triple for the extracted quotation.

13. The method of claim 1 wherein extracting the quotation includes identifying a speaker for a quotation by identifying a subject of a sentence in the text document, and identifying one or more modifiers associated with the subject.

14. The method of claim 1 wherein extracting the quotation includes identifying a verb for the quotation by identifying a verb of a sentence in the text document, and identifying prepositional modifiers of the verb.

15. The method of claim 1, further comprising indexing the extracted quotation.

16. The method of claim 15 wherein indexing the extracted quotation includes storing a speaker-verb-quote triple in an inverted index managed by the content recommendation system, the inverted index including quotations extracted from other text documents of the corpus of text documents.

17. The method of claim 15 wherein indexing the extracted quotation includes storing at least one of:

a reference to a subject entity;

a facet associated with a subject entity;

subject modifiers;

an action;

an action modifier; and

/or one or more references to object entities and/or facets that are referenced within the quotation.

18. The method of claim 1 further comprising receiving a quotation search request that includes a query, and determining whether the quotation matches the query.

19. The method of claim 18 wherein the query specifies a speaker by including at least one of:

an entity reference, a facet reference, and/or any speaker.

20. The method of claim 19 wherein the query specifies one or more modifiers of the speaker.

21. The method of claim 18 wherein the query specifies a subject by including at least one of:

an entity reference, a facet reference, one or more keyterms, and/or any subject.

22. The method of claim 18 wherein the query includes Boolean combinations of multiple speakers or subjects.

23. The method of claim 18 wherein providing the quotation includes:

determining multiple quotations that match the query; and

ranking the multiple quotations based on one or more factors including;

publication date, number of documents including the quotation, and/or credibility of quotation source.

24. A computing system configured to recommend content, comprising:

a memory;

a module stored on the memory that is configured, when executed, to;

extract a quotation from a text document in a corpus of text documents;

identify one or more entities that are referenced by the text document, each of the determined entities being electronically represented by the content recommendation system, wherein identifying the one or more entities includes linking together multiple mentions of a same entity across the text document, the linking including resolving pronoun coreference to identify an entity that is a speaker of the quotation;

attributing the quotation to the speaker of the quotation by storing data that associates the quotation with the identified entity that is the speaker of the quotation; and

provide the quotation by transmitting text that represents the quotation and the attributed speaker.

25. The computing system of claim 24 wherein the module includes software instructions for execution in the memory of the computing system.

26. The computing system of claim 24 wherein the module is a content recommendation system.

27. The computing system of claim 24 wherein the module is configured to recommend content items to at least one of a desktop computing system, a personal digital assistant, a smart phone, a laptop computer, a mobile application, and/or a third-party application.

28. A non-transitory computer-readable medium including:

contents that, when executed, cause a computing system to recommend content, by performing a method comprising;

extracting a quotation from a text document in a corpus of text documents;

identifying one or more entities that are referenced by the text document, each of the determined entities being electronically represented by the content recommendation system, wherein identifying the one or more entities includes linking together multiple mentions of a same entity across the text document, the linking including resolving pronoun coreference to identify an entity that is a speaker of the quotation;

attributing the quotation to the speaker of the quotation by storing data that associates the quotation with the identified entity that is the speaker of the quotation; and

providing the quotation by transmitting text that represents the quotation and the attributed speaker.

29. The computer-readable medium of claim 28 wherein the computer-readable medium is at least one of a memory in a computing device or a data transmission medium transmitting a generated signal containing the contents.

30. The computer-readable medium of claim 28 wherein the contents are instructions that when executed cause the computing system to perform the method.