Objective-function based sentiment

• US 8,949,211 B2
• Filed: 01/31/2011
• Issued: 02/03/2015
• Est. Priority Date: 01/31/2011
• Status: Active Grant

• Current Assignee: Valtrus Innovations Limited (f/k/a Dolya Holdco 9 Limited) (Key Patent Innovations Limited)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Lu, Yue, Castellanos, Maria G, Dayal, Umeshwar
• Primary Examiner(s): Alam, Hosain
• Assistant Examiner(s): HARPER, ELIYAH STONE
• Application Number:
  US13/018,412
• Publication Number:
  US 20120197903A1
• Time in Patent Office: 1,464 Days
• Field of Search: 707/999.102, 707/999.101, 707/705
• US Class Current: 707/705
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

A system and article are disclosed for objective-function based sentiment. In one example, the system includes a set of domain information, and a computer programmed with executable instructions which operate a set of modules. The modules include a sentiment polarization module for identifying a domain-aspect opinion-word pair within a set of domain data, and assigning a sentiment polarity score to the domain-aspect opinion-word pair based on an objective function which includes sentiment data from the domain information.

Claims

1. A system, comprising:

a set of domain information; and

a computer; and

a sentiment polarization module executable in the computer to;

identify a first domain-aspect opinion-word pair from plural domain-aspect opinion-word pairs within domain data pertaining to a given domain; and

assign a sentiment polarity score to the first domain-aspect opinion-word pair based on an objective function which includes sentiment data from the domain information, the objective function including;

a first constraint specifying that the sentiment polarity score is positively correlated with an overall sentiment rating associated with the domain data; and

a second constraint specifying that the sentiment polarity score assigned to the first domain-aspect opinion-word pair is related to a sentiment polarity score assigned to a second domain-aspect opinion-word pair from the plural domain-aspect opinion-word pairs, depending upon whether an opinion word in the first domain-aspect opinion-word pair is a synonym or antonym of an opinion word in the second domain-aspect opinion-word pair.

2. The system of claim 1:

wherein the domain information includes;

a set of domain-independent information including sentiment data assigned to opinion-words independent of the given domain; and

a set of domain-dependent information including sentiment data assigned to opinion-words dependent upon the given domain; and

wherein the assigning further includes assigning the sentiment polarity score based on the domain-independent information and the domain-dependent information.

3. The system of claim 1:

wherein the objective function is optimized using a set of constraints including the first constraint and the second constraint.

4. The system of claim 3:

wherein the set of constraints further includes a constraint specifying that the sentiment polarity score is close to another sentiment polarity score assigned by a domain-independent sentiment lexicon associated with the domain information.

5. The system of claim 1:

wherein the objective function includes weighting parameters that are set to different values for different trust levels of the first and second constraints, and the sentiment polarity score assigned to the first domain-aspect opinion-word pair is based on the weighting parameters.

6. An article comprising at least one non-transitory computer-readable storage medium storing instructions that upon execution cause a computer system to:

identify domain-aspect opinion-word pairs within domain data pertaining to a given domain; and

assign a first sentiment polarity score to a first of the domain-aspect opinion-word pairs based on an objective function which includes sentiment data from a set of domain-independent information and a set of domain-dependent information, the set of domain-independent information including sentiment data assigned to opinion-words independent of the given domain, and the set of domain-dependent information including sentiment data assigned to opinion-words dependent upon the given domain, and the objective function includes a first constraint specifying that the first sentiment polarity score assigned to the first domain-aspect opinion-word pair is related to a second sentiment polarity score assigned to a second of the domain-aspect opinion-word pairs, dependent upon whether an opinion word in the first domain-aspect opinion-word pair is a synonym or antonym of an opinion word in the second domain-aspect opinion-word pair.

7. The article of claim 6:

wherein the domain-independent information is at least one from a group including;

a domain-independent sentiment lexicon;

a dictionary;

a synonym dictionary;

an antonym dictionary;

a thesaurus;

a set of language clues;

an “

and”

language clue;

a “

but”

language clue; and

a “

negation”

language clue.

8. The article of claim 6:

wherein the domain-dependent information is at least one from a group including;

a domain-dependent sentiment lexicon;

domain-aspects;

word-clusters;

an overall sentiment ratings; and

a document-level overall sentiment rating.

9. The article of claim 6, wherein the identifying comprises:

parsing the domain data into a set of sentence clauses;

tagging each clause with a domain-aspect having a word-cluster which overlaps with words in the clause; and

identifying opinion words within the clauses.

12. The article of claim 6, wherein the objective function further includes a second constraint specifying that the first sentiment polarity score is positively correlated with an overall sentiment rating associated with the domain data.

13. The article of claim 12, wherein the objective function is includes weighting parameters that are set to different values for different trust levels of the first and second constraints, and the first sentiment polarity score assigned to the first domain-aspect opinion-word pair is based on the weighting parameters.

10. A sentiment polarity system comprising:

a set of domain information;

at least one processor; and

a sentiment polarization module executable on the at least one processor to;

identify a first of plural domain-aspect opinion-word pairs within first domain data pertaining to a given domain;

assign a first sentiment polarity score to the first domain-aspect opinion-word pair based on an objective function which includes sentiment data from the set of domain information, wherein the objective function includes a first constraint specifying that the first sentiment polarity score assigned to the first domain-aspect opinion-word pair is related to a second sentiment polarity score assigned to a second of the plural domain-aspect opinion-word pairs, depending upon whether an opinion word in the first domain-aspect opinion-word pair is a synonym or antonym of an opinion word in the second domain-aspect opinion-word pair;

populate a domain-dependent sentiment lexicon with the first domain-aspect opinion-word pair and the first sentiment polarity score; and

assign a sentiment polarity score to a domain-aspect opinion-word pair within second domain data based on the domain-dependent sentiment lexicon.

11. The system of claim 10, wherein the objective function further includes a second constraint specifying that the first sentiment polarity score is positively correlated with an overall sentiment rating associated with a document in the first domain data.

14. The system of claim 11, wherein the objective function is includes weighting parameters that are set to different values for different trust levels of the first and second constraints, and the first sentiment polarity score assigned to the first domain-aspect opinion-word pair is based on the weighting parameters.