Method for automatically generating a summarized text by a computer

• US 6,401,086 B1
• Filed: 09/16/1999
• Issued: 06/04/2002
• Est. Priority Date: 03/18/1997
• Status: Expired due to Fees

• Current Assignee: Siemens AG
• Original Assignee: Siemens AG
• Inventors: Brückner, Thomas
• Primary Examiner(s): Coby, Frantz
• Application Number:
  US09/381,180
• Time in Patent Office: 992 Days
• Field of Search: 707/3, 707/4, 707/5
• US Class Current: 1/1
• CPC Class Codes: G06F 16/345 Summarisation for human users
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

The method enables the sentence-based automatic summarization of a text on a computer. For this purpose, subject-related lexica are used that provide a relevance measure for each word contained therein. Each sentence of the text to be summarized is processed word-by-word, and for each word an individual word frequency, weighted with the relevance measure, is cumulated. For the summarization, the n sentences having the greatest probability that they belong to the summary are assembled, whereby n is a predeterminable reduction measure.

Claims

1. A method for automatic generation of a summary of a text by a computer, comprising the steps of:

calculating for each sentence a probability that the sentence belongs to the summary, including

determining a relevance measure for each word in the sentence from a lexicon that contains application-specific words with a predetermined relevance measure for each of these words, and

adding together the relevance measures for all words in the sentence to yield the probability that the sentence belongs to the summary,

sorting all sentences of the text according to the probabilities,

performing a predeterminable reduction measure,

displaying as the summary best sentences in a sequence given by the text.

2. A method according to claim 1, further comprising the steps of:

in addition to the relevance measure, determining an individual word frequency for each word, and

determining the probability that the sentence is contained in the summary by the following rule;

${\mathrm{WK}}_{\left(\mathrm{sentence}\right)}=\frac{1}{N}·<br><br>\underset{i=1}{\overset{N}{\sum <br><br>}}<br><br>\mathrm{tf}·<br><br>\mathrm{rlv}$

 

wherein

WK(sentence) is a probability that a sentence belongs to the summary,

N is a total number of words that occur in the sentence,

I is a count variable (I=1,2, . . . , N) for all the words in the sentence,

tf is a frequency of occurrence of the word under consideration in an entire text being summarized, and

rlv is a relevance measure for the word in the sentence.

3. A method according to claim 1, further comprising the step of:

allocating the text to at least one category using an application-specific lexicon.

4. A method according to claim 1, further comprising the steps of:

producing for each allocation of the text to a category an application-specific summary.