Category classification processing device and method

• US 9,552,415 B2
• Filed: 02/03/2012
• Issued: 01/24/2017
• Est. Priority Date: 03/18/2011
• Status: Expired due to Fees

• Current Assignee: Fujitsu Client Computing Limited (Lenovo Group Ltd.)
• Original Assignee: Fujitsu Limited
• Inventors: Nagano, Reiko, Inoue, Hajime
• Primary Examiner(s): Mahmoudi, Tony
• Assistant Examiner(s): Black, Linh
• Application Number:
  US13/365,904
• Publication Number:
  US 20120239657A1
• Time in Patent Office: 1,817 Days
• Field of Search: 707727-731, 707/750, 707/708, 707/771
• US Class Current: 1/1
• CPC Class Codes: G06F 16/353 into predefined classes

Abstract

A category classification processing device includes a search unit that stores, as a search keyword log assembly, Q&A examples, which are actually referred to by a client, together with keywords; and a category extracting unit that obtains keyword storage frequencies expressing a number of times each of the keywords, which are recorded together with the Q&A examples in the search keyword logs, is stored for each of the Q&A examples, extracts, as category candidates of each of the Q&A examples, an m number of top keywords (m being a positive integer) in a descending order of the keyword storage frequency, uses the extracted category candidates as categories, and associates the categories with the Q&A examples.

Claims

1. A category classification processing device comprising:

an example storage unit configured to store Q&

A (questions and answers) examples including pairs of questions and answers;

a log storage unit configured to store search keyword logs relevant to searching the Q&

A examples; and

a processor configured to perform a search process and a category extraction process,

wherein the search process includes

sending, to a terminal device, an input screen image used for inputting one or more keywords,

searching the example storage unit using the one or more keywords input to the input screen image to extract Q&

A examples based on the one or more keywords,

sending identification information items expressing the extracted Q&

A examples to the terminal device, and

storing, in the log storage unit, search keyword logs including the identification information items relevant to the Q&

A examples which are referred to at the terminal device using the identification information items, the identification information items in the search keyword logs being associated with the one or more keywords input to the input screen image during the searching, and

wherein the category extraction process includes

aggregating keyword storage frequencies expressing a number of times each of the keywords input to the input screen image during the searching and recorded together with the Q&

A examples in the search keyword logs, is stored for each of the Q&

A examples, and calculating keyword group storage frequencies expressing a number of times each of keyword groups is stored for each Q&

A example, each of the keyword groups including n keywords recorded together with the Q&

A examples in the search keyword log, where n is a positive integer,

extracting, as category candidates of each of the Q&

A examples, m top keywords in a descending order of the keyword storage frequencies, where m is a positive integer, and extracting, as tentative sub category candidates of each Q&

A example, keywords belonging to m top keyword groups in the descending order of the keyword group storage frequency,

using the extracted category candidates as categories, and using the extracted tentative sub category candidates as sub categories, and

associating the categories with the sub categories, and associating the sub categories with the Q&

A examples, to store the categories, the sub categories, and the Q&

A examples that are associated with each other in the example storage unit.

2. The category classification processing device according to claim 1, wherein the category extraction process includes

extracting, as a first word, a keyword of the tentative sub category candidate matching the category candidate,

extracting, as one or more second words, one or more keywords of the tentative sub category candidate other than the first word, and

using each of the one or more second words as the sub categories of the category when the category candidate matching the first word is used as the category.

3. The category classification processing device according to claim 1, further comprising:

a first file storage unit configured to store a word number setting file in which a value of n is described.

4. The category classification processing device according to claim 1, wherein, when the number of keywords recorded together with each Q&

A example exceeds n, the category extraction process extracts n keywords from the number of keywords exceeding n, and groups together the extracted n keywords to generate a plurality of the keyword groups.

5. The category classification processing device according to claim 1, further comprising:

a second file storing unit configured to store an extraction condition setting file in which a value of m is described.

6. The category classification processing device according to claim 1, wherein, when extracting the category candidates, the category extraction process sorts the keywords in the descending order of the storage frequency for each Q&

A example, specifies ranks to which the top m keywords belong, and extracts all of or some of the keywords belonging to the specified ranks as the category candidates.

7. The category classification processing device according to claim 6, wherein

when extracting the category candidates, the category extraction process determines, for each Q&

A example, whether a number of keywords belonging to the top rank is greater than or equal to m, and when m or more keywords belong to the top rank, extracts all of or some of the keywords belonging to the top rank as the category candidates, and

when less than m keywords belong to the top rank, the category extraction process determines whether a number of keywords belonging to a rank next to the top rank is greater than or equal to m, and when m or more keywords belong to the rank next to the top rank, extracts all of or some of the keywords belonging to the rank next to the top rank as the category candidates.

8. The category classification processing device according to claim 6, wherein, when extracting the category candidates, the category extraction process extracts all of the keywords recorded together with each Q&

A example as the category candidates, when the number of the keywords recorded together with each Q&

A example is less than m.

9. The category classification processing device according to claim 1, wherein the search process includes

searching the categories and the Q&

A examples, which are associated with each other by the category extraction process, by using the one or more keywords input to the input screen image,

extracting the Q&

A examples associated with the categories, and

sending the identification information expressing the extracted Q&

A examples to the terminal device.

10. A category classification processing method comprising:

sending, to a terminal device, an input screen image used for inputting one or more keywords;

searching, an example storage unit that stores Q&

A examples including pairs of questions and answers, by using the one or more keywords input to the input screen image to extract Q&

A (questions and answers) examples based on the one or more keywords;

sending identification information items expressing the extracted Q&

A examples to the terminal device;

storing, in a log storage unit that stores search keyword logs relevant to searching the Q&

A examples, search keyword logs including the identification information items relevant to the Q&

A examples which are referred to at the terminal device using the identification information items, the identification information items in the search keyword logs being associated with the one or more keywords input to the input screen image during the searching;

aggregating keyword storage frequencies expressing a number of times each of the keywords input to the input screen image during the searching and recorded together with the Q&

A examples in the search keyword logs, is stored for each of the Q&

A examples, and calculating keyword group storage frequencies expressing a number of times each of keyword groups is stored for each Q&

A example, each of the keyword groups including n keywords recorded together with the Q&

A examples in the search keyword log, where n is a positive integer;

extracting, as category candidates of each of the Q&

A examples, m top keywords in a descending order of the keyword storage frequencies, where m is a positive integer, and extracting, as tentative sub category candidates of each Q&

A example, keywords belonging to m top keyword groups in the descending order of the keyword group storage frequency;

using the extracted category candidates as categories, and using the extracted tentative sub category candidates as sub categories; and

associating the categories with the sub categories, and associating the sub categories with the Q&

A examples, to store the categories, the sub categories, and the Q&

A examples that are associated with each other in the example storage unit.

12. The category classification processing method according to claim 10, wherein the extracting includes

extracting, as a first word, a keyword of the tentative sub category candidate matching the category candidate, and

extracting, as one or more second words, one or more keywords of the tentative sub category candidate other than the first word, and

wherein the using includes using each of the one or more second words as the sub categories of the category when the category candidate matching the first word is used as the category.

11. A non-transitory computer-readable recording medium having stored therein a category classification processing program for causing a computer to execute a process comprising:

sending, to a terminal device, an input screen image used for inputting one or more keywords;

searching, an example storage unit that stores Q&

A examples including pairs of questions and answers, by using the one or more keywords input to the input screen image to extract Q&

A (questions and answers) examples based on the one or more keywords;

sending identification information items expressing the extracted Q&

A examples to the terminal device;

storing, in a log storage unit that stores search keyword logs relevant to searching the Q&

A examples, search keyword logs including the identification information items relevant to the Q&

A examples which are referred to at the terminal device using the identification information items, the identification information items in the search keyword logs being associated with the one or more keywords input to the input screen image during the searching;

aggregating keyword storage frequencies expressing a number of times each of the keywords input to the input screen image during the searching and recorded together with the Q&

A examples in the search keyword logs, is stored for each of the Q&

A examples, and calculating keyword group storage frequencies expressing a number of times each of keyword groups is stored for each Q&

A example, each of the keyword groups including n keywords recorded together with the Q&

A examples in the search keyword log, where n is a positive integer;

extracting, as category candidates of each of the Q&

A examples, m top keywords in a descending order of the keyword storage frequencies, where m is a positive integer, and extracting, as tentative sub category candidates of each Q&

A example, keywords belonging to m top keyword groups in the descending order of the keyword group storage frequency;

using the extracted category candidates as categories, and using the extracted tentative sub category candidates as sub categories; and

associating the categories with the sub categories, and associating the sub categories with the Q&

A examples, to store the categories, the sub categories, and the Q&

A examples that are associated with each other in the example storage unit.

13. The non-transitory computer-readable recording medium as claimed in claim 11, wherein the extracting includes

extracting, as a first word, a keyword of the tentative sub category candidate matching the category candidate, and

extracting, as one or more second words, one or more keywords of the tentative sub category candidate other than the first word, and

wherein the using includes using each of the one or more second words as the sub categories of the category when the category candidate matching the first word is used as the category.