Text summarizing method and system

• US 5,978,820 A
• Filed: 03/29/1996
• Issued: 11/02/1999
• Est. Priority Date: 03/31/1995
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Tsuji, Hiroshi, Mase, Hisao
• Primary Examiner(s): FETTING, ANTON W
• Application Number:
  US08/622,892
• Time in Patent Office: 1,313 Days
• Field of Search: 707/531, 707/508, 707/4, 707/5
• US Class Current: 715/210
• CPC Class Codes: G06F 16/345 Summarisation for human users
• CPC Class Codes: G06F 40/237 Lexical tools

Abstract

By storing data on text attributes comprising a plurality of attribute items, a text analysis rule, a plurality of text type identification rules, a plurality of text content/domain type identification rules, a plurality of summarization method setting tables, and a summarization rule in a storage, analysing an inputted text in an electronic form on the basis of the text analysis rule and the data on text attributes, generating a text analysis table, determining the text type of the inputted text in an electronic form on the basis of the text type identification rules and the content of the text analysis table, also determining the text content/domain type on the basis of the text content/domain type identification rules, selecting a summarization method setting table corresponding to a combination of the determined text type and text content/domain type, and summarizing the inputted text in an electronic form on the basis of the summarization method setting table and the summarization rule, a text summarizing method and system for preventing the correctness of text summarization from dropping due to a difference in the constitution, field, and content of a text.

Claims

1. A text summarizing method in a system comprising at least a processor and a storage and summarizing an inputted text comprising:

a step of setting a text analysis rule, a plurality of text type identification rules, a plurality of summarization method setting tables, and a summarization rule in said storage;

a step of extracting at least a numerical value of conditions defined relative to text structure elements, from said inputted text by said processor on the basis of said text analysis rule;

a step of determining the text type of said inputted text, based upon the extracted numerical value and said text type identification rules;

a step of selecting a summarization method setting table corresponding to said determined text type from said summarization method setting tables; and

,

a step of summarizing said inputted text on the basis of said selected summarization method setting table and said summarization rule.

2. A text summarizing method according to claim 1, further comprising:

a step of setting a plurality of text content/domain type identification rules in said storage;

a step of extracting a corresponding text content/domain type identification rule on the basis of said plurality of text content/domain type identification rules and said text analysis rule;

a step of determining the text type of said extracted identification rule as a text type of said inputted text; and

a step of selecting a summarization method setting table corresponding to a combination of said determined text type and said text content/domain type from said summarization method setting tables.

3. A text summarizing method according to claim 1, further comprising:

a step of informing of the text before summarization, said determined text type, and said selected text summarizing method by said output device according to an instruction of a user when said summarized text is informed of by said output device.

4. A text summarizing method according to claim 1, further comprising:

a step of presetting a default text type and a default text content/domain type and also presetting a summarization method setting table corresponding to a combination of a text type and a text content/domain type including one or both of said default text type and said default text content/domain type in said storage; and

a step of recognizing said default text type as a text type when it is instructed to omit said text type determination by said input device and recognizing said default text content/domain type as a text content/domain type when it is instructed to omit said text content/domain type determination by said input device.

5. A text summarizing method in a system comprising an input device, an output device, a processor, and a storage and preparing a summary by summarizing an inputted text in an electronic form, comprising:

a step of setting a word dictionary, grammar rules, data on text attributes including a plurality of attribute items, a text analysis rule, a plurality of text type identification rules, a plurality of text content/domain type identification rules, a plurality of summarization method setting tables, and a summarization rule in said storage;

a step of analysing said inputted text in an electronic form on the basis of said word dictionary, said grammar rules, said data on text attributes, and said text analysis rule and generating a text analysis table including a word table and a text table by said processor;

a step of extracting a text type identification rule conforming to the content of said text analysis table mostly by referring to said plurality of text type identification rules and determining the text type of said extracted identification rule as a text type of said inputted text in an electronic form;

a step of extracting a text content/domain type identification rule conforming to the content of said text analysis table mostly by referring to said plurality of text content/domain type identification rules and determining the text content/domain type of said extracted identification rule as a text content/domain type of said inputted text in an electronic form;

a step of selecting a summarization method setting table corresponding to a combination of said determined text type and said text content/domain type from said plurality of summarization method setting tables; and

a step of summarizing said inputted text in an electronic form on the basis of said selected summarization method setting table and said summarization rule.

6. A text summarizing method according to claim 5, further comprising:

a step of displaying said data on text attributes including a plurality of attribute items or a text type identification rule set in said storage already in said output device when said text type identification rule is set in said storage; and

a step of generating or updating a text type identification rule from an identifier indicating the text type inputted from said input device, the attribute value of each attribute item, and said attribute items.

7. A text summarizing method according to claim 5, further comprising:

a step of displaying said data on text attributes including a plurality of attribute items and data on text attributes including a plurality of attribute items for text content/domain type identification or a text type identification rule set in said storage already in said output device when said text content/domain type identification rule is set in said storage; and

a step of generating or updating a text content/domain type identification rule from an identifier indicating the text content/domain type inputted from said input device, the attribute value of each attribute item, and said attribute items.

8. A text summarizing method according to claim 5, wherein said text type or said text content/domain type inputted and designated from said input device is used in place of said determined text type or said determined text content/domain type.

9. A text summarizing method according to claim 5, further comprising:

a step of presetting a default text type and a default text content/domain type and also presetting a summarization method setting table corresponding to a combination of a text type and a text content/domain type including one or both of said default text type and said default text content/domain type in said storage; and

a step of recognizing said default text type as a text type when it is judged that the type cannot be determined as a result of said text type determination and recognizing said default text content/domain type as a text content/domain type when it is judged that the type cannot be determined as a result of said text content/domain type determination.

10. A text summarizing system comprising at least a processor and a storage and summarizing an inputted text comprising:

a storage for storing a text analysis rule, a plurality of text type identification rules, a plurality of summarization method setting tables, and a summarization rule;

means for extracting at least a numerical value of conditions defined relative to text structure elements, from said inputted text on the basis of said text analysis rule;

means for determining the text type of said inputted text, based upon the extracted numerical value and said text type identification rules;

means for selecting a summarization method setting table corresponding to said determined text type from said summarization method setting tables; and

,

means for summarizing said inputted text on the basis of said selected summarization method setting table and said summarization rule.

11. A text summarizing system according to claim 10, wherein said storage further stores a plurality of text content/domain type identification rules, said system further comprising:

means for extracting a corresponding text content/domain type identification rule on the basis of said plurality of text content/domain type identification rules and said text analysis rule;

means for determining the text content/domain type of said extracted identification rule as a text content/domain type of said inputted text; and

,

means for selecting a summarization method setting table corresponding to a combination of said determined text type and said determined text content/domain type from said summarization method setting tables.

12. A text summarizing system according to claim 10, further comprising:

means for informing of the text before summarization, said determined text type, and said selected text summarizing method by said output device according to an instruction of a user when said summarized text is informed of by said output device.

14. A text summarizing system according to claim 12, further comprising:

means for displaying said data on text attributes including a plurality of attribute items or a text type identification rule set in said storage already in said output device when said text type identification rule is set in said storage, generating or updating a text type identification rule from an identifier indicating the text type inputted from said input device, the attribute value of each attribute item, and said attribute items, and setting it in said storage.

15. A text summarizing system according to claim 12, further comprising:

means for displaying said data on text attributes including a plurality of attribute items and data on text attributes including a plurality of attribute items for text content/domain type identification or a text type identification rule set in said storage already in said output device when said text content/domain type identification rule is set in said storage, generating or updating a text content/domain type identification rule from an identifier indicating the text content/domain type inputted from said input device, the attribute value of each attribute item, and said attribute items, and setting it in said storage.

16. A text summarizing system according to claim 12, wherein said text type or said text content/domain type inputted and designated from said input device is used in place of said determined text type or said determined text content/domain type.

17. A text summarizing system according to claim 12, further comprising:

means for presetting a default text type and a default text content/domain type and also presetting a summarization method setting table corresponding to a combination of a text type and a text content/domain type including one or both of said default text type and said default text content/domain type in said storage;

wherein said text type determination module recognizes said default text type as a text type when said text type determination module judges that the type cannot be determined as a result of said text type determination and said text content/domain type determination module recognizes said default text content/domain type as a text content/domain type when said text content/domain type determination module judges that the type cannot be determined as a result of said text content/domain type determination.

18. A text summarizing system according to claim 12, further comprising:

means for presetting a default text type and a default text content/domain type and also presetting a summarization method setting table corresponding to a combination of a text type and a text content/domain type including one or both of said default text type and said default text content/domain type in said storage; and

means for recognizing said default text type as a text type when it is instructed to omit said text type determination by said input device and recognizing said default text content/domain type as a text content/domain type when it is instructed to omit said text content/domain type determination by said input device.

13. A text summarizing system for preparing a summary by summarizing an inputted text in an electronic form, comprising:

a storage for storing a word dictionary, grammar rules, data on text attributes including a plurality of attribute items, a text analysis rule, a plurality of text content/domain type identification rules, a plurality of summarization method setting tables, and a summarization rule;

a text analysis module for analyzing said inputted text in an electronic form on the basis of said word dictionary, said grammar rules, said data on text attributes, and said text analysis rule and generating a text analysis table including a word table and a text table;

a text type determination module for extracting a text type identification rule conforming to the content of said text analysis table mostly by referring to said plurality of text type identification rules and determining the text type of said extracted identification rule as a text type of said inputted text in an electronic form;

a text content/domain type determination module for extracting a text content/domain type identification rule conforming to the content of said text analysis table mostly by referring to said plurality of text content/domain type identification rules and determining the text content/domain type of said extracted identification rule as a text content/domain type of said inputted text in an electronic form;

a summarization method setting module for selecting a summarization method setting table corresponding to a combination of said determined text type and said text content domain type from said plurality of summarization method setting tables; and

,

a text summarization module for summarizing said inputted text in an electronic form on the basis of said selected summarization setting table and said summarization rule.