Reply generation apparatus and method and program for causing computer to function as reply generating apparatus

• US 8,271,485 B2
• Filed: 04/05/2007
• Issued: 09/18/2012
• Est. Priority Date: 04/05/2006
• Status: Active Grant

• Current Assignee: I-Business Center Corporation
• Original Assignee: I-Business Center Corporation
• Inventors: Shiraishi, Tomihisa, Iwasaki, Atsuhiro
• Primary Examiner(s): Chen, Susan
• Application Number:
  US12/225,922
• Publication Number:
  US 20090094204A1
• Time in Patent Office: 1,993 Days
• Field of Search: 707/1, 707/715, 704/257, 705/1, 709/223
• US Class Current: 707/731
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

A reply conforming to applied search criteria is generated using data obtained by a search based upon the search criteria. A reply generating server includes a storage unit storing a plurality of items of basic data; a search unit for searching and retrieving basic data, which is in line with search criteria that has been input by the user of a terminal device, from the storage unit; a reply generating unit which, in a case where a plurality of items of basic data in line with the search criteria exist, generates anew a reply conforming to the search criteria by utilizing the plurality of items of basic data; and a reply transmitting unit for transmitting the reply to the terminal device.

Claims

1. A reply generating apparatus, comprising:

a storage unit of a computer in which a plurality of text aggregates of previous replies to queries have been stored, wherein a text aggregate of said text aggregates represents a text that is divided into text segments, each of which has a coherent meaning and each of which comprises a text having a time-wise relevancy logical flow, each of the text segments being classified under at least one of a plurality of category items;

a search unit for searching said storage unit in accordance with an applied query;

a deciding unit for deciding a core category item that has a highest frequency of appearance among the category items, wherein the core category item is decided from the plurality of category items that correspond to the text segments that constitute each of the text aggregates that are found by the search unit in accordance with the applied query;

a counter to calculate an adjacent-category count numbers for each category item of the plurality of category items, a number of other category items adjacent to each category item including the core category item, based on the text aggregates that are found by the search unit in accordance with the applied query, wherein the adjacent-category count is calculated with regard to each of the category items that correspond to the text segments that constitute said each of the text aggregates that are found by the search unit in accordance with the applied query;

a related category item deciding unit for deciding related category items to be adjacent to the core category item from the plurality of category items other than the core category item, based upon the adjacent-category count numbers calculated by said counter; and

a logical framework of a reply generating unit for generating a new logical framework of a reply to the applied query, wherein the new logical framework of the reply adopts the core category item as a core and places the related category items adjacent to the core category item to express a reply along lines of the new logical framework of a reply to an end user,

wherein the category items include a fixed-format category item under which a text segment having a stereotypical meaning is classified, and

wherein said deciding unit decides the core category item from category items other than the fixed-format category item.

2. The apparatus according to claim 1, wherein on a basis of the adjacent-category count numbers, said related category item deciding unit decides a related category item, which is further adjacent to the core category item, from the category items other than the category item decided upon as the core category item and the category item decided upon as the related category item adjacent to the core category item.

3. The apparatus according to claim 2, further comprising an expression deciding unit for deciding upon at least one of the text segments, which have been stored in the storage unit and classified under the core category item, as a text segment that expresses the core category item, and deciding upon at least one of text segments, which have been stored in the storage unit and classified under the related category items, as a text segment that expresses the related category items.

4. The apparatus according to claim 1, further comprising an expression deciding unit for deciding upon at least one of the text segments, which have been stored in the storage unit and classified under the core category item, as a text segment that expresses the core category item, and deciding upon at least one of the text segments, which have been stored in the storage unit and classified under the related category items, as a text segment that expresses the related category items.

5. The apparatus according to claim 1,

wherein said reply generating unit prioritizes each of a plurality of combinations of the core category item and the related category items based upon the adjacent-category count numbers regarding the core category item and the related category items that exist in a combination, and, in a case where a related category item of a part of a low-priority combination matches a related category item of a part of a high-priority combination, excludes the low-priority combination from the reply generated in the reply generating unit.

6. The apparatus according to claim 1, wherein the category items corresponding to a respective plurality of the text segments are arranged in an order.

7. The apparatus according to claim 1, wherein in accordance with the applied query, the text aggregate is retrieved from the plurality of text aggregates that is stored in the storage unit.

8. The apparatus according to claim 1, wherein the related category item deciding unit decides a category item that has a highest degree of association with the query among the category items as the core category item.

9. The apparatus according to claim 1, wherein the related category item deciding unit decides from the category items that correspond to the text segments that constitute said each of the text aggregates that are found, a category item of a largest number as the core category item.

10. The apparatus according to claim 1, wherein the related category item deciding unit decides from the plurality of category items a category item having a largest number of adjacent other category items as the core category item.

11. A method of generating a reply, said method comprising:

storing a plurality of text aggregates of previous replies to queries in a storage unit, wherein a text aggregate of said text aggregates represents a text that is divided into text segments, each of which has a coherent meaning and each of which comprises a text having a time-wise relevancy logical flow, each of the text segments being classified under at least one of a plurality of category items;

searching the storage unit in accordance with an applied query;

deciding a core category item that has a highest frequency of appearance among the category items, wherein the core category item is decided from the plurality of category items that correspond to the text segments that constitute each of the text aggregates that are found by a search unit in accordance with the applied query;

calculating an adjacent-category count numbers for each category item of the plurality of category items, a number of other category items adjacent to each category item including the core category item, based on the text aggregates that are found by the search unit in accordance with the applied query, wherein the adjacent-category count is calculated with regard to each of the category items that correspond to the text segments that constitute said each of the text aggregates that are found by the search unit in accordance with the applied query;

deciding related category items to be adjacent to the core category item from the plurality of category items other than the core category item, based upon the adjacent-category count numbers calculated; and

generating a new logical framework of a reply to the applied query, wherein the new logical framework of the reply adopts the core category item as a core and places the related category items adjacent to the core category item to express a reply along lines of the new logical framework of a reply to an end user,

wherein the category items include a fixed-format category item under which a text segment having a stereotypical meaning is classified, and

wherein in said deciding the core category item is decided from category items other than the fixed-format category item.

12. A non-transitory computer-readable storage medium encoded with a computer program, the computer program including instructions for a method of generating a reply, said method comprising:

storing, in a computer having a storage unit a plurality of text aggregates of previous replies to queries, wherein a text aggregate of said text aggregates represents a text that is divided into text segments, each of which has a coherent meaning and each of which comprises a text having a time-wise relevancy logical flow and each of the text segments is classified under at least one of a plurality of category items;

searching the storage unit in accordance with an applied query;

deciding a core category item that has a highest frequency of appearance among the category items, wherein the core category item is decided from the plurality of category items that correspond to the text segments that constitute each of the text aggregates that are found by a search unit in accordance with the applied query;

calculating an adjacent-category count numbers for each category item of the plurality of category items, a number of other category items adjacent to each category item including the core category item, based on the text aggregates that are found by the search unit in accordance with the applied query, wherein the adjacent-category count is calculated with regard to each of the category items that correspond to the text segments that constitute said each of the text aggregates that are found by the search unit in accordance with the applied query;

deciding related category items to be adjacent to the core category item from the plurality of category items other than the core category item, based upon the adjacent-category count numbers calculated; and

generating a new logical framework of a reply to the applied query, wherein the new logical framework of the reply adopts the core category item as a core and places the related category items adjacent to the core category item to express a reply along lines of the new logical framework of a reply to an end user,

wherein the category items include a fixed-format category item under which a text segment having a stereotypical meaning is classified, and

wherein in said deciding the core category item is decided from category items other than the fixed-format category item.