APPARATUS AND METHOD FOR GENERATING DIGEST ACCORDING TO HIERARCHICAL STRUCTURE OF TOPIC

• US 20020184267A1
• Filed: 10/21/1998
• Published: 12/05/2002
• Est. Priority Date: 03/20/1998
• Status: Active Grant

• Current Assignee: Recognition Interface, Inc.
• Original Assignee: Fujitsu Limited
• Inventors: NAKAO, YOSHIO
• Granted Patent: US 6,638,317 B2
• Time in Patent Office: Days
• US Class Current: 715/515
• CPC Class Codes: G06F 16/345 Summarisation for human users
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

A digest generator apparatus calculates a lexical cohesion degree at each position in a document using a plurality of windows having different sizes, and calculates the candidate section of a topic boundary for each topic level corresponding to the size of a window. Then, by unifying the candidate section of different levels, the digest generator apparatus detects the topic boundary for each level. Then, based on the relation between a summarization-target topic passage and a long topic passage containing the summarization-target topic passage the digest generator apparatus extracts key sentences and generates a digest.

Claims

1. A digest generator apparatus, comprising:

structure detecting means for detecting a hierarchical structure of topics in a given document;

extracting means for extracting a keyword regarding each detected topic;

selecting means for selecting a key sentence from each topic passage depending on a use condition of said keyword, and generating a digest using the key sentence; and

.

outputting means for outputting said digest.

2. The digest generator apparatus according to claim 1, wherein

said structure detecting means calculates a lexical cohesion degree in a vicinity area of each position in said document, detects a topic boundary based on the cohesion degree, and by repeating detection of the topic boundary while gradually reducing the size of the vicinity area, detects the hierarchical structure of topics ranging from a long topic passage to a short topic passage.

3. The digest generator apparatus according to claim 2, wherein

said structure detecting means calculates said cohesion degree in the vicinity area of each position based on a similarity of a vocabulary contained in two windows set up before and after each position, and repeats the detection of said topic boundary while gradually reducing a window width.

4. The digest generator apparatus according to claim 2, wherein

said structure detecting means comprises candidate section detecting means for handling a moving average of said cohesion degree as both a forward cohesion force at the beginning of the moving average zone and a backward cohesion force at the end of the moving average zone, and recognizing a vicinity area of a position where the forward cohesion force and the backward cohesion force balance each other as a candidate section of a topic boundary, and identifies the topic boundary using the candidate section.

5. The digest generator apparatus according to claim 2, further comprising:

major part specifying means for removing document portion having a lower cohesion degree and extracting document portion having a higher cohesion degree as a major part, and wherein

said selecting means selects said key sentence from a topic passage corresponding to the major part.

6. The digest generator apparatus according to claim 1, wherein

said extracting means evaluates whether or not a term used in a range of said topic passage is characteristic of the topic passage, and extracts said keyword based on an evaluation result.

7. The digest generator apparatus according to claim 6, wherein

said extracting means obtains said evaluation result using both a use frequency of an evaluation target term in said topic passage and a use frequency of the evaluation target term in a long topic passage containing the topic passage.

8. The digest generator apparatus according to claim 1, wherein

said extracting means extracts a local keyword from a summarization-target topic passage, and extracts a global keyword from a long topic passage containing the summarization-target topic passage;

and

said selecting means selects said key sentence from the summarization-target topic passage based on a use condition of the local keyword and the global keyword.

9. The digest generator apparatus according to claim 1, further comprising:

determining means for determining a size of said topic passage according to a size of said digest.

10. A digest generator apparatus, comprising:

extracting means for evaluating whether or not a word is characteristic of a process target topic passage in a given document using both a use frequency of the word in the process target topic passage and a use frequency of the word in a long topic passage containing the process target topic passage, and extracting a keyword from the process target topic passage based on an evaluation result;

generating means for generating a digest according to a use condition of said keyword; and

outputting means for outputting said digest.

11. A digest generator apparatus, comprising:

extracting means for extracting a local keyword from a summarization-target topic passage, and extracting a global keyword from a long topic passage containing the summarization-target topic passage;

generating means for generating a digest based on a use condition of said local keyword and global keyword; and

outputting means for outputting said digest.

12. A digest generator apparatus, comprising:

means for calculating a lexical cohesion degree in a vicinity area of each position in a given document;

major part specifying means for removing document portion having a lower cohesion degree from a process target, and extracting document portion having a higher cohesion degree as a major part;

generating means for generating a digest using said major part; and

outputting means for outputting said digest.

13. A computer-readable storage medium for storing a program which enables a computer to execute the steps of:

detecting a hierarchical structure of topics in a given document;

extracting a keyword regarding each detected topic;

7

selecting a key sentence from each topic passage according to a use condition of said keyword; and

generating a digest using said key sentence.

14. A computer-readable storage medium for storing a program which enables a computer to execute the steps of:

calculating a use frequency of a word in a process target topic passage in a given document;

calculating a use frequency of said word in a long topic passage containing said process target topic passage;

evaluating whether or not said word is characteristic of said process target topic passage using the two obtained use frequencies;

extracting a keyword from said process target topic passage based on an evaluation result; and

generating a digest according to a use condition of said keyword.

15. A computer-readable storage medium for storing a program which enables a computer to execute the steps of:

extracting a local keyword from a summarization-target topic passage;

extracting a global keyword from a long topic passage containing said summarization-target topic passage; and

generating a digest based on a use condition of said local keyword and global keyword.

16. A computer-readable storage medium for storing a program which enables a computer to execute the steps of:

calculating a lexical cohesion degree in a vicinity area of each position in a given document;

removing document portion having a lower cohesion degree from a process target, and extracting document portion having a higher cohesion degree as a major part; and

generating a digest using said major part.

17. A method of generating a digest, comprising the steps of:

detecting a hierarchical structure of topics in a given document;

extracting a keyword regarding each detected topic;

selecting a key sentence from each topic passage according to a use condition of said keyword; and

generating a digest using said key sentence.