Information processing to search for related expressions

• US 10,380,151 B2
• Filed: 08/04/2014
• Issued: 08/13/2019
• Est. Priority Date: 08/13/2013
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Miyahara, Masanori, Araki, Kazunori, Kanemoto, Katsuyoshi, Nakahashi, Ryo, Yoshiyama, Kazuki, Takagi, Tomohiro
• Primary Examiner(s): Trujillo, James
• Assistant Examiner(s): Mueller, Kurt A
• Application Number:
  US14/450,403
• Publication Number:
  US 20150052140A1
• Time in Patent Office: 1,835 Days
• Field of Search: None
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification

Abstract

Disclosed is an information processing apparatus including an expression extraction unit, a feature extraction unit, a clustering unit, a related expression extraction unit, and an output unit. The expression extraction unit extracts a plurality of expressions from a plurality of documents. The feature extraction unit extracts feature amounts of the extracted respective expressions while distinguishing the expressions having the same notation. The clustering unit clusters the extracted respective expressions together while distinguishing the expressions having the same notation and calculates assignment degree vectors having assignment degrees of the respective expressions to two or more respective clusters as components. The related expression extraction unit extracts related expressions having the assignment degree vectors similar to those of a provided input expression while distinguishing the expressions having the same notation. The output unit outputs the related expressions and identification information for identifying the related expressions.

Claims

1. An information processing apparatus, comprising:

a processor configured to;

obtain an input expression;

extract, based on the input expression, a plurality of expressions and a plurality of expression feature amounts from a plurality of documents,

wherein the plurality of expression feature amounts is associated with the plurality of expressions,

wherein a first expression of the plurality of expressions is distinguished from a second expression of the plurality of expressions based on respective expression feature amounts of the first expression and the second expression, and

wherein the first expression and the second expression have a same notation;

cluster the plurality of expressions based on the plurality of expression feature amounts;

calculate a plurality of assignment degree vectors for the plurality of expressions based on an average value of first assignment degree vectors of the input expression,

wherein each of the plurality of assignment degree vectors comprises a plurality of assignment degrees of a respective expression of the plurality of expressions, and

wherein the assignment degree vectors include features common to the plurality of expressions;

determine related expression candidates, from the plurality of expressions, based on the input expression;

calculate a first score for each of the related expression candidates based on comparison between the related expression candidates and the input expression;

extract, from the plurality of expressions, a set of related expressions of a plurality of related expressions based on the calculated first score,

wherein the set of related expressions has second assignment degree vectors;

integrate the set of related expressions into at least one expression, wherein each of the set of related expressions has the same notation;

determine a combined score for the at least one expression based on an addition of scores of the set of related expressions;

determine combined assignment degree vectors for the at least one expression based on a weighted addition of the second assignment degree vectors of the set of related expressions;

extract, as a synonym of the input expression, an expression from the set of related expressions based on each of the combined score for the at least one expression and the combined assignment degree vectors for the at least one expression;

recommend at least one item based on the extracted expression of the set of related expressions; and

transmit information to a user device, wherein the information comprises the at least one item, the at least one expression, the related expression, the combined assignment degree vectors, and the combined score to the user device,

wherein the transmitted information is displayed on a display screen of the user device.

2. The information processing apparatus according to claim 1, wherein the processor is further configured to output the second assignment degree vectors and a second score that expresses similarity in the assignment degree vectors of the related expression and the assignment degree vectors of the input expression.

3. The information processing apparatus according to claim 2, wherein the processor is further configured to:

add a plurality of scores associated with the plurality of related expressions,

wherein each of the plurality of related expressions has the same notation, and

wherein the extracted expression of the plurality of related expressions has a highest score among the plurality of scores.

4. The information processing apparatus according to claim 3, wherein the processor is further configured to

determine at least two clusters corresponding to the synonym, based on whether components of the combined assignment degree vectors associated with the synonym is greater than a threshold.

5. The information processing apparatus according to claim 1,

wherein the processor is further configured to:

recommend an item to a user associated with the user device;

extract a related expression of a keyword related to the recommended item; and

present the related expression of the keyword to the user together with the recommended item.

6. The information processing apparatus according to claim 5, wherein the processor is further configured to present feature amounts of the related expression of the keyword to the user together with the recommended item.

7. The information processing apparatus according to claim 1,

wherein the assignment degree vectors of a related expression is similar to either a first vector obtained by an addition of assignment degree vectors for a plurality of input expressions or a second vector obtained by a multiplication of the assignment degree vectors for the plurality of input expressions.

8. The information processing apparatus according to claim 1,

wherein the transmitted information expresses a document, of the plurality of documents, from which a related expression is extracted and a position in the document.

9. The information processing apparatus according to claim 1,

wherein the feature amounts of the plurality of expressions are based on the plurality of documents from which the plurality of expressions are extracted.

10. The information processing apparatus according to claim 1, wherein the processor is further configured to:

recommend an item to a user associated with the user device;

extract a related expression of a keyword included in at least one of expression feature amounts of the user or expression feature amounts of the item; and

add the related expression of the keyword to at least one of the expression feature amounts of the user or the expression feature amounts of the item.

11. The information processing apparatus according to claim 10, wherein the processor is further configured to add expression feature amounts of the related expression of the keyword to at least one of the expression feature amounts of the user or the expression feature amounts of the item.

12. An information processing method, comprising:

extracting, based on an input expression, a plurality of expressions and a plurality of expression feature amounts from a plurality of documents,

wherein the plurality of expression feature amounts is associated with the plurality of expressions,

wherein a first expression of the plurality of expressions is distinguished from a second expression of the plurality of expressions based on respective expression feature amounts of the first expression and the second expression, and

wherein the first expression and the second expression have a same notation;

clustering the plurality of expressions based on the plurality of expression feature amounts;

calculating a plurality of assignment degree vectors for the plurality of expressions based on an average value of first assignment degree vectors of the input expression,

wherein each of the plurality of assignment degree vectors comprises a plurality of assignment degrees of a respective expression of the plurality of expressions, and

wherein the assignment degree vectors include features common to the plurality of expressions;

determining related expression candidates, from the plurality of expressions, based on the input expression;

calculating a score for each of the related expression candidates based on comparison between the related expression candidates and the input expression;

extracting, from the plurality of expressions, a set of related expressions of a plurality of related expressions based on the calculated score,

wherein the set of related expressions has second assignment degree vectors;

integrating the set of related expressions into at least one expression, wherein each of the set of related expressions has the same notation;

determining a combined score for the at least one expression based on an addition of scores of the set of related expressions;

determining combined assignment degree vectors for the at least one expression based on a weighted addition of the second assignment degree vectors of the set of related expressions;

extracting, as a synonym of the input expression, an expression from the set of related expressions based on each of the combined score for the at least one expression and the combined assignment degree vectors for the at least one expression;

recommending at least one item based on the extracted expression of the set of related expressions; and

transmitting information to a user device, wherein the information comprises the at least one item, the at least one expression, the related expression, the combined assignment degree vectors, and the combined score,

wherein the transmitted information is displayed on a display screen of the user device.

13. A non-transitory computer-readable medium having stored thereon computer executable instructions that, when executed by a processor, cause a computer to execute operations, the operations comprising:

extracting, based on an input expression, a plurality of expressions and a plurality of expression feature amounts from a plurality of documents,

wherein the plurality of expression feature amounts are associated with the plurality of expressions,

wherein a first expression of the plurality of expressions is distinguished from a second expression of the plurality of expressions based on respective expression feature amounts of the first expression and the second expression, and

wherein the first expression and the second expression have a same notation;

clustering the plurality of expressions based on the plurality of expression feature amounts;

calculating a plurality of assignment degree vectors for the plurality of expressions based on an average value of first assignment degree vectors of the input expression,

wherein each of the plurality of assignment degree vectors comprises a plurality of assignment degrees of a respective expression of the plurality of expressions; and

wherein the assignment degree vectors include features common to the plurality of expressions;

determining related expression candidates, from the plurality of expressions, based on the input expression;

calculating a score for each of the related expression candidates based on comparison between the related expression candidates and the input expression;

extracting, from the plurality of expressions, a set of related expressions of a plurality of expressions based on the calculated score,

wherein the set of related expressions has second assignment degree vectors;

integrating the set of related expressions into at least one expression, wherein each of the set of related expressions has the same notation;

determining a combined score for the at least one expression based on an addition of scores of the set of related expressions;

determining combined assignment degree vectors for the at least one expression based on a weighted addition of the second assignment degree vectors of the set of related expressions;

extracting, as a synonym of the input expression, an expression from the set of related expressions based on each of the combined score for the at least one expression and the combined assignment degree vectors for the at least one expression;

recommending at least one item based on the extracted expression of the set of related expressions; and

transmitting information to a user device, wherein the information comprises the at least one item, the at least one expressions, the related expression, the combined assignment degree vectors, the combined score,

wherein the transmitted information is displayed on a display screen of the user device.