SYSTEM AND METHOD FOR SUGGESTION MINING

• US 20130096909A1
• Filed: 10/13/2011
• Published: 04/18/2013
• Est. Priority Date: 10/13/2011
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Brun, Caroline, Hagege, Caroline
• Granted Patent: US 8,738,363 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/295 Named entity recognition

Abstract

A system and method for extraction of suggestions for improvement form a corpus of documents, such as customer reviews, are disclosed. A structured terminology provided or a topic includes a set of semantic classes, each including a set of terms. A thesaurus of terms relating to suggestions of improvement is provided. Text elements of text strings in the documents which are instances of terms in the structured terminology are labeled with the corresponding semantic class and text elements which are instances of terms in the thesaurus are also labeled. A set of patterns is applied to the labeled text strings to identify suggestions of improvement expressions. The patterns define syntactic relations between text elements, some of which are required to be instances of one of the terms in a particular semantic class or thesaurus. A set of suggestions for improvements is output based on the identified suggestions of improvement expressions.

Claims

1. A method for extraction of suggestions for improvement comprising:

providing a structured terminology for a topic, the structured terminology including a set of semantic classes, each of a plurality of the semantic classes including a finite set of terms;

providing a thesaurus of terms relating to suggestions of improvement;

receiving a corpus of text documents, each document comprising a text string in a natural language;

labeling text elements in the text strings which are instances of terms in the structured terminology with the corresponding semantic class;

labeling text elements in the text strings which are instances of terms in the thesaurus;

with a processor, applying a set of patterns to the labeled text strings to identify suggestions of improvement expressions, the patterns each defining a syntactic relation between text elements, the patterns including;

for each of the semantic classes in the set, at least one pattern which specifies a syntactic relation in which one of the text elements in the relation is labeled as an instance of the semantic class, and

wherein at least one of the patterns specifies a syntactic relation in which one of the text elements in the relation is labeled as an instance of one of the terms in the thesaurus; and

outputting a set of suggestions for improvements based on the identified suggestions of improvement expressions.

2. The method of claim 1, further comprising natural language processing the text strings to extract syntactic relations between text elements in the text strings, the syntactic relations including the syntactic relations applied in the patterns.

3. The method of claim 1, wherein the topic comprises a product or service and the structured terminology includes terms related to the product or service.

4. The method of claim 3, wherein the corpus of text documents comprises reviews of the product or service.

5. The method of claim 3, wherein the product comprises an electromechanical device.

6. The method of claim 3, wherein the structured terminology includes a set of classes selected from the group consisting of:

a. a class of physical parts of the product;

b. a class of manufacturers of the product;

c. a class of trade names of the product;

d. a class of types of the product;

e. a class of physical attributes of the product;

f. a class of functional attributes of the product;

g. a class of commercial attributes of the product; and

combinations thereof.

7. The method of claim 6, wherein at least three of the classes are selected from the group.

8. The method of claim 1, wherein each of the classes in the structured terminology is in at least one of the patterns as a subject or object in a syntactic relation.

9. The method of claim 1, wherein the thesaurus of terms relating to suggestions of improvement includes a set of terms relating to a belief concept and a set of terms relating to a lack concept.

10. The method of claim 9, wherein the set of terms relating to the belief concept includes terms selected from the group consisting of figure, think, believe, suppose, assume, consider, guess, suspect, speculate, imagine, presume, deny, disclaim, and question.

11. The method of claim 9, wherein the set of terms relating to the lack concept include at least one of:

a verb selected from the group consisting of;

miss and lack; and

a noun selected from the consisting of;

absence;

shortage, deficiency;

want;

dearth;

need;

insufficiency;

and combinations thereof.

12. The method of claim 1, wherein the corpus of text documents relate to a type of product and the patterns include a plurality of patterns selected from the group consisting of:

a) a pattern of the form;

13. The method of claim 1, wherein the structured terminology includes at least three classes.

14. The method of claim 1, wherein a plurality of the classes each include at least five terms.

15. The method of claim 1, wherein a majority of the terms in the structured terminology are nouns or noun phrases.

16. The method of claim 1, wherein all the terms in a majority of the classes in the structured terminology are nouns or noun phrases.

17. The method of claim 1, wherein a plurality of the patterns each specify a syntactic relation in which one of the text elements in the relation is labeled as an instance of one of the terms in the thesaurus.

18. The method of claim 1, wherein the method further includes extracting opinions from the text documents.

19. A computer program product comprising a non-transitory recording medium which stores instructions which when executed by a computer, causes the computer to perform the method of claim 1.

20. A suggestions for improvement extraction system comprising memory which stores instructions for performing the method of claim 1 and a processor in communication with the memory for executing the instructions.

21. A system for extraction of suggestions for improvement comprising:

memory which stores;

a structured terminology for a topic, the structured terminology including a set of semantic classes, each of a plurality of the semantic classes including a finite set of terms,

a thesaurus of terms relating to suggestions of improvement, and

a set of patterns for identify suggestions of improvement expressions, in input text, the patterns each defining a syntactic relation between two text elements, the patterns including, for each of the semantic classes in the set of semantic classes, at least one pattern which specifies a syntactic relation in which one of the text elements in the relation is labeled as an instance of the semantic class, and wherein at least one of the patterns specifies a syntactic relation in which one of the text elements in the relation is labeled as an instance of a term in the thesaurus;

a parser configured for labeling text elements in input text strings, which are instances of terms in the structured terminology, with the corresponding semantic class and for labeling text elements in the text strings which are instances of terms in the thesaurus;

a suggestion review component for extracting suggestions for improvement expressions by applying the set of patterns to the input text strings and outputting suggestions for improvement based on the extracted expressions; and

a processor for implementing the parser and the suggestion review component.

22. A method for forming a system for extraction of suggestions for improvement comprising:

generating a structured terminology for a topic, the structured terminology including a set of semantic classes, each of a plurality of the semantic classes including a finite set of terms related to the respective semantic class, the terms in the structured terminology including nouns or noun phrases;

generating a thesaurus of terms relating to suggestions of improvement, the terms in the thesaurus including verbs;

deriving a set of patterns to be applied to text strings to identify suggestions of improvement expressions, the patterns each defining a syntactic relation between two text elements, the patterns including;

for each of the semantic classes in the set, at least one pattern which specifies a syntactic relation in which one of the text elements in the syntactic relation is labeled as an instance of the semantic class, and

wherein at least one of the patterns specifies a syntactic relation in which one of the text elements in the relation is labeled as an instance of a term in the thesaurus; and

with a processor, testing the patterns on a corpus of text documents to evaluate the performance of the system.

23. The method of claim 22, wherein the method further comprises refining the patterns if the performance is not at a satisfactory level.