JPS61248160A - Document information registering system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a file system for storing academic papers, etc., and in particular automates the input of keywords for searching.
Suitable for automatically generating summary sentences as search auxiliary information,
Regarding document information registration method.

[Background of the invention]

When accumulating and storing coded text that has been manually written or created using a word processor for the purpose of printing or archiving, conventionally it has been necessary to input search keywords separately from the text at the same time as entering the main text. there were. Furthermore, when searching, it is not possible to select a desired text sufficiently using keywords alone, so a summary of a single text is often used as supplementary information for the search, but this summary also needs to be created and entered separately from the main text. However, there was a drawback that data registration was complicated.

When registering text information as an image, the inventors automatically create search auxiliary information (guidance image 1 image index) corresponding to the above summary text by reducing or cutting out the original document information, During a search, we have invented a method that displays a list of auxiliary information for document information that corresponds to a specified keyword, and specifies desired document information from among the auxiliary information. (Special Public Service 56-5
3788.57-8499, patent application 1982-20166
6) L, rudder: All of these are for registering as images, and for code data that is the output result of a word processor, etc.

The drawback was that it could not be applied.

[Purpose of the invention]

An object of the present invention is to simplify the registration process when document information is registered as a code by automatically extracting and registering search keywords and summary text as search auxiliary information from text data. It is.

[Summary of the invention]

Generally, when searching for document information, the types of words that can be used as search keywords are limited. Particularly in the case of academic papers, patent information, etc., keywords are strictly defined and often have special codes attached to them.

Furthermore, in the case of academic papers and patent information, the above-mentioned keywords are always included in sentences that explain the subject (so-called tobitsku sentences).

In the present invention, focusing on the above properties, (1) automatically extracting words that can be keywords from the main text; ■ A day consisting of keyword input, summary sentence creation, and input by extracting sentences that include the keywords extracted in ■ above, re-editing them into occurrences, and making them into summary sentences.

This automates the evening registration process.

[Embodiments of the invention]

Hereinafter, the present invention will be explained in detail using Examples.

FIG. 1 shows a processing overview of the document information registration method according to the present invention. 10 in the figure indicates automatic keyword extraction and automatic summary creation processing according to the present invention. In this embodiment, a list of words that can be used as keywords, that is, a keyword table 101 is provided, and words extracted from a text file 102 are provided. It is checked whether the word matches the word stored in the keyword table 101. If they match, that word is registered in the keyword file 103 as a keyword for the document. For example, when symbols, codes, etc. associated with keywords are defined, such as classification codes in patent information, these codes are also stored in the keyword table 101 and registered in the keyword file 103 along with the keywords. Furthermore, a sentence containing the keyword is extracted from the main text file 102,
This is stored in the summary sentence file 104. The words extracted from the text file 102 are stored in the keyword table 101.
If the word does not match the word stored in the document information, it is assumed that the word cannot be a keyword of the document information,
Shifts processing to the next word extraction.

The above process is repeated for all words in the text file 102, thereby completing extraction of all keywords and creation of a summary sentence.

FIG. 2 shows an example of the automatic extraction process described above. In the figure, 102 is a main text file, and the "image" present in the first line of six sentences matches the content of the keyword table. Therefore, “image” is registered in the keyword table 103.

Along with this, the sentence containing this "image" is "In the present invention...
・I made it so. ” is stored in the summary sentence file 104.

FIG. 3 shows the flow of the processing described above. 201 in the figure is a process for extracting words from the text file 102.

In English, words can be identified as strings of characters separated by blanks, periods, commas, etc. On the other hand, in Japanese, in addition to the English rules, there are also particles, conjunctions, and suffixes.

Extract as a character string separated by conjugated endings, etc.

Furthermore, since all words that can be used as keywords are nouns, by determining one part of speech and extracting only one noun, it is possible to reduce the number of times the determination process 202 that follows is performed.

Process 202 is a process of checking whether the same word as the above-mentioned extracted word exists in the keyword table 101. Generally, the number of entries in the keyword table 101 increases, so it is necessary to speed up the checking process 202 by using a method such as a binary check.

As described above, in the process 203, if a symbol, number, etc. (hereinafter referred to as a key code) is defined along with the keyword, the key code is stored in the keyword table 101.
This is the process of selecting from.

This key code is stored in the keyword table 101 in association with the keyword.

204 in FIG. 2 is a process for storing the keywords and key codes selected in the above processes 202 and 203 in the keyword file 103. The keyword file 103 has the same code entered as the identifier of the document information in the text file 102.
Following this, keywords and key codes are stored. Therefore, in process 204, the empty area following the identification code corresponding to the text data currently being processed is
．． The keywords and key codes selected in step 203 are stored.

Process 205 is a process of extracting a sentence containing the keyword as a summary sentence. The sentence ends before a period, semicolon, or comma followed by a conjunction.

The period from the end point of such a sentence to the end point of the next sentence is extracted as one sentence. Furthermore, conjunctions are removed to create a summary sentence.

Processing 206 stores the sentences extracted in the above processing into the summary sentence file 104. Summary text file 104
Similar to the keyword file 103, an identification code corresponding to one text data is entered, and the sentence extracted in process 206 is stored in the empty area following this.

Process 207 determines the end of the above processing,
When processing is completed for all words in the text, automatic extraction of keywords and automatic creation of a summary sentence are completed.

In the above embodiment, all keyword candidates in the text are extracted. In this case, the word used for convenience of explanation of the main text happens to be a keyword candidate in the keyword table 10.
1, it may also be extracted as a keyword for the document. In order to eliminate this, methods can be considered, such as adopting only keyword candidates that appear more than a specified number of times as true keywords, or adopting only keywords that appear as subjects or objects in syntactic analysis. It will be done.

FIG. 4 shows a display screen when searching for documents registered using the document information registration method according to the present invention.

Three pieces of document information are selected in response to the keyword "r image". In order to specify a desired one from among these, a summary sentence automatically created by the document information registration method according to the present invention is displayed as a guidance screen. By looking at this summary text, the searcher can identify the document information that is truly needed.

〔Effect of the invention〕

According to the present invention, when registering document information such as academic papers and patent information, only the text data is input, and then,
It automatically extracts keywords and also automatically creates a summary sentence, which was traditionally considered an intellectual activity for humans. Therefore, it not only simplifies the task of inputting document information, but also makes it easier to find the desired information by using automatically created summaries as auxiliary information when searching for information that cannot be specified using keywords alone. Can be identified. Furthermore, conventional document information registration/retrieval systems have often been hindered from becoming popular due to the difficulty of inputting them, but the present invention can eliminate this obstacle.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the overall configuration of the document information registration method according to the present invention, FIG. 2 is a diagram showing an example of automatic keyword extraction and automatic summary creation processing, and FIG. 3 is the process. 41!1 is a diagram showing an example of a guide screen using a summary sentence automatically created according to the present invention as auxiliary information for searching. 10...Keyword/summary sentence automatic creation section, 101...
-Keyword table, 102...Main text file. 103... Keyword file, 104 Summary sentence file, 201... Word extraction processing section, 202 River keyword check section, 203... Key code selection section, 20
4...Keyword registration section, 205...Sentence extraction processing section, 2'06...Summary sentence registration section, 2o7...End determination section. ￥ 1 mouth 2nd mouth ￥ 13 mouth

Claims (1)

[Claims] 1. A document characterized in that, in a file system that stores document information, search keywords are automatically extracted from the main text, and the results are used to automatically generate a summary sentence as search auxiliary information. Information registration method. 2. The document information registration method according to item 1, which displays a list of the summaries and selects desired information from among them.