JPH09190450A - Information processor and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processor whereby even a keyword corresponding to personal interest while picking-up a general keyword in a composition. SOLUTION: In a document processor, a document holding part 1 stores an inputted document and a sentence segmenting part 2 segments a sentence from the document. A word segmenting part 3 segments a word from the segmented sentence. A user keyword holding part 4 stores a user keyword at every person. A user keyword pick-up part 5 picks-up the segmented word as a keyword candidate when it is the user keyword. An unrequired word list holding part 6 stores an unrequired word being the word which does not become the keyword generally such as '(matter)' and '(thing)', etc. A keyword candidate pick-up part 7 recognizes that the word segmented by the word segmenting part 3 is not the unrequired word and picks-up as the keyword candidate. A keyword candidate holding part 8 holds the keyword candidates picked-up by the user keyword pickup part 5 and the keyword candidate pickup part 7. A keyword selecting part 9 converges the proper number of keywords from the keyword candidates and its result is provided to a user by an output part 10.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus for extracting a keyword representing a sentence from a document.

[0002]

2. Description of the Related Art Conventionally, a method of extracting a keyword from a sentence in a workstation or a personal computer has been studied. Although the keyword is useful as an index role for the purpose of retrieval, it is an intellectual task that requires understanding the sentence, so it is difficult for humans to accurately assign the keyword.

In the method of automatically extracting a keyword by a computer, first, a noun is cut out from a sentence and the frequency of each is examined. Next, words that frequently appear in general (words that cannot be keywords) are excluded from the cut out nouns. Finally, from the remaining nouns, it is decided whether or not to adopt it as a keyword in consideration of the appearance position in the sentence.

[0004]

However, the above-mentioned conventional examples have the following problems, and their improvement is desired. That is, since the above-mentioned keyword extraction method uses only the information written in the text, it is intended to extract the keyword considered by the writer. In general, sentences are written assuming unspecified readers, so it is thought that this extraction method can only extract general keywords used by people without background knowledge. Each person has different background knowledge, and different people have different interests. Therefore, there should be times when a word is a keyword for some people, but not for others.

For example, when a person who is investigating market trends reads a newspaper, he cannot afford to miss the name of the company he is paying attention to. Articles in which the company name appears many times are used as general keywords, but articles that appear only once in one corner are unlikely to be keywords. In the latter case, there is a high risk that even humans will miss it, so it is only in this case that the computer should properly extract it as a keyword.

Therefore, an object of the present invention is to provide an information processing apparatus capable of extracting general keywords in a sentence and also extracting keywords according to individual interests.

[0007]

In order to achieve the above object, an information processing apparatus according to claim 1 of the present invention is a sentence cutting-out means for cutting out a sentence from an input document, and a sentence cutting-out means for cutting out the sentence. A word cutting-out means for cutting out a word, a user keyword holding means for holding a word registered by a user as a user keyword, and the held user keyword and the cut-out word are compared, and match the user keyword. User keyword extracting means for extracting words as keyword candidates, unnecessary word list holding means for holding an unnecessary word list in which words not suitable for keywords are registered, words registered in the unnecessary word list and the cut-out words And if they do not match, the keyword candidate extraction for extracting the cut-out word as the keyword candidate A keyword candidate holding means for holding the stage, the extracted keyword candidates, characterized in that a keyword selection means for selecting the keyword from among the keyword candidates the holding.

An information processing apparatus according to claim 2 is the information processing apparatus according to claim 1.
In the information processing apparatus according to the above, the keyword candidate is provided with a score assigning unit for assigning a score according to the importance, and the keyword selecting unit selects the keyword according to the assigned score. To do.

An information processing apparatus according to claim 3 is the information processing device according to claim 2.
In the information processing apparatus according to the above, a type is provided for the user keyword, and the keyword candidate that matches the user keyword of a specific type is set as the most important keyword without being scored by the score assigning means. The keyword selecting means selects the most important keyword as the keyword.

According to a fourth aspect of the information processing method, a sentence is cut out from an input document, a word is cut out from the cut out sentence, the word registered by the user is held as a user keyword, and the held user is held. The keyword is compared with the cut-out word, a word that matches the user keyword is extracted as a keyword candidate, an unnecessary word list in which words that are not suitable for the keyword are registered is held, and the unnecessary word list is registered. When a word is compared with the cut-out word and if they do not match, the cut-out word is extracted as the keyword candidate, the extracted keyword candidate is held, and the keyword is selected from the held keyword candidates. Is selected.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an information processing apparatus according to the present invention will be described. The information processing device in this embodiment is applied to a document processing device. FIG. 1 is a block diagram showing the functional configuration of the document processing apparatus according to the embodiment. In the figure, 1 is a document holding unit that stores a document entered under the user, 2 is a sentence cutout unit that cuts out a sentence from an input document, and 3 is a word cutout unit that cuts out a word from the cut out sentence. is there.

Reference numeral 4 denotes a user keyword holding unit for storing a word which should be a keyword for each individual as a user keyword, and 5 determines whether or not the word cut out by the word cutout unit 3 is a user keyword of the user keyword holding unit 4. When it is a user keyword, the user keyword extraction unit extracts the keyword as a keyword candidate.

Reference numeral 6 denotes an unnecessary word list holding unit for storing a list of unnecessary words, such as "koto" and "things", which are frequently found in any document and are not generally used as keywords. Reference numeral 7 denotes a keyword candidate extraction unit that confirms that the word cut out by the word cutout unit 3 is not in the unnecessary word list holding unit 6 and extracts it as a keyword candidate.

Reference numeral 8 is a keyword candidate holding unit for holding the keyword candidates extracted by the user keyword extracting unit 5 and the keyword candidate extracting unit 7. Reference numeral 9 denotes a keyword selection unit that narrows down an appropriate number of keywords from these keyword candidates, and the result is provided to the user at the output unit 10.

FIG. 2 is a block diagram showing the hardware configuration of the document processing apparatus. In the figure, reference numeral 11 is a control memory that stores a control procedure, and includes a ROM and a RAM. A central processing unit 12 performs processing according to the control procedure stored in the control memory 11, and has a document holding unit 1, a user keyword holding unit 4, and a keyword candidate holding unit 8.

Reference numeral 14 is a keyboard. Reference numeral 15 is a disk, which has an unnecessary word list holding unit 6. A display 16 is composed of a CRT or a liquid crystal display (LCD). The display 16 is used to display the keywords. Reference numeral 17 is a bus for connecting each component.

FIG. 3 is a flowchart showing the keyword extraction processing procedure of the document processing apparatus. FIG. 5 is an explanatory diagram showing a document to be processed and its processing result. First,
Sentences are cut out sentence by sentence from the document taken out from the document holding unit 1 (step S1), and words are cut out from the cut-out sentence (step S2). In this example sentence, "20 days",
Words such as "ABC" and "President" are clipped. When the word is in the inflected form such as a verb, the word is returned to the original form (step S3).

It is determined whether or not each word is registered in the user keyword holding unit 4 (step S4). FIG. 4 is an explanatory diagram showing information stored in the user keyword holding unit 4. For user keywords, "most important" is always extracted, and "important" is emphasized more than other words.
There are two types. Here, "Canon" is set as the most important, and "Color copy" and "Camera" are set as important.

Subsequently, the type is determined (step S
5) If it is "most important", create the most important keyword (step S6). If it is "important", the user keyword is scored (step S7). This score is simply the score added in step S9, which will be described later.
It is doubled. The scored words are registered in the keyword candidate list of the keyword candidate holding unit 8 (step S10). In the case of FIG. 5, since the user keyword “Canon” exists in the second sentence, it is decided as a keyword candidate.

It is determined whether or not the word determined in step S4 as not a user keyword is an unnecessary word (step S8). The unnecessary word list of the unnecessary word list holding unit 6 is a simple list including only words or words and part-of-speech information.

If it is determined that the word is an unnecessary word, the process returns to step S2 to process the next word. If it is not an unnecessary word,
Scores are assigned as keyword candidates (step S9) and registered in the keyword candidate list (step S10).

Various methods may be used for scoring. For example, 3 points are added at the beginning and the last sentence of a document, and 2 points are added at the beginning of a paragraph. Also, if it is a noun, it is 3 points when it is "ha ・ ga case" and 2 points when it is "wa case" according to the case particle
The points and others are allotted to be 1 point. Since "ABC" in the first sentence of FIG. 5 is the first sentence of the document and "ha case", it has 6 points.

The "color copy" has 4 points because it is the first sentence and other cases, and has 8 points in total because it is a user keyword and the score is doubled.

As shown above, all keyword candidates with scores are stored in the keyword candidate holding unit 8. Since the number of keyword candidates is large, it is necessary to reduce the number when presenting it to the user. The number of keywords is given in advance by the user. In this case, since the sentence is short, it is set to two.

The keyword candidates are arranged in the order of scores (step S11), and two keywords with the highest scores are adopted as keywords. Here, "color copy" of 8 points and "ABC" (6 points + 3 points = 9 points) that appear twice are created as keywords (step S12).

The most important keyword created in step S6 and the keyword created in step S12 are connected (step S13). In this example, the most important keyword is “Canon”, the important keyword is “color copy”, and the normal keyword is “ABC”.

At this stage, the keywords are arranged in score order, so that all the keywords are rearranged so that they are easy to understand (step S14). In the case of FIG. 5, the documents are rearranged in the order of appearance. All the rearranged keywords are presented to the user (step S15), and the process ends.

In the above embodiment, only the keyword is displayed when outputting the keyword, but the entire body may be displayed and the keyword portion may be highlighted. As an emphasizing method, there are methods such as changing the font, increasing the font size, underlining, and changing the character color. On the contrary, the characters other than the keyword may be made small or lightly colored to make them inconspicuous.

Further, in the above-mentioned embodiment, when the keywords are rearranged, the documents are arranged in the order of appearance, but other arrangements such as score may be arranged.

Further, in the above embodiment, the types of user keywords are classified into two types, "most important" and "important", but they may be classified more finely. In the case of fine classification, a score multiplying factor may be designated for each type. Furthermore, a fixed point may be added instead of the magnification.

Further, in the above embodiment, the unit of the user keyword is a word, but a partial match such as a compound noun or an affix may be considered. For example, by registering the classifier "yen", the expression of the price such as "59,800 yen" in the document can be surely detected.

Further, in the above-mentioned embodiment, the keyword to be detected is the user keyword, but conversely, the word to be ignored may be registered so that it is not a general keyword.

Further, in the above-mentioned embodiment, the keyword extraction is performed for the Japanese document, but a document in a language other than English or Japanese may be targeted.

Furthermore, the present invention may be applied to a system composed of a plurality of devices or may be applied to an apparatus composed of one device. Needless to say, the present invention can be applied to a case where the present invention is achieved by supplying a program to a system or an apparatus. In this case, by reading out a storage medium storing a program represented by software for achieving the present invention into a system or an apparatus, the system or the apparatus can enjoy the effects of the present invention.

[0035]

According to the information processing apparatus according to the first aspect of the present invention, a sentence is cut out from the document input by the sentence cutout means, and a word is cut out from the cutout sentence by the word cutout means. Holds the word registered by the user by the keyword holding means as a user keyword,
The user keyword extracting means compares the held user keyword with the cut-out word, extracts a word that matches the user keyword as a keyword candidate, and the unnecessary word list holding means registers a word not suitable for the keyword. Holding the unnecessary word list, comparing the words registered in the unnecessary word list by the keyword candidate extraction means and the cut out words, if they do not match, extract the cut out words as the keyword candidates, The extracted keyword candidates are held by the keyword candidate holding means, and the keyword is selected from the held keyword candidates by the keyword selecting means. Therefore, it is possible to extract general keywords in the sentence and meet the individual interest. You can also extract keywords. Therefore,
It becomes possible to extract keywords that take personal interests into consideration, and appropriately determine whether to read the text by looking at the keywords. Further, even when searching for documents by keywords, all necessary documents can be searched for.

According to the information processing apparatus of the second aspect, there is provided a score assigning means for assigning a score according to the degree of importance to the extracted keyword candidates, and the keyword selecting means according to the assigned score. Since the keywords are selected, it is possible to clarify the criteria for selecting the keywords in consideration of individual interests, and it is possible to change the criteria for the selection according to the selected keywords.

According to the information processing apparatus of claim 3, a type is provided for the user keyword, and the keyword assigning means does not give a score to the keyword candidate that matches the user keyword of the specific type. Since the keyword candidate is set as the most important keyword and the keyword selecting means selects the most important keyword as the keyword, it is possible to reliably extract a word of extremely high importance as a keyword and prevent it from being overlooked.

According to the information processing method of the fourth aspect, a sentence is cut out from the input document, a word is cut out from the cut out sentence, the word registered by the user is held as a user keyword, and the word is held. Comparing the extracted user word with the extracted word, and extracting a word that matches the user keyword as a keyword candidate,
Holding an unnecessary word list in which words that are not suitable for keywords are registered, and comparing the words registered in the unnecessary word list with the cut out words, and if they do not match, the cut out words as the keyword candidates Since the extracted keyword candidates are held and the keyword is selected from the held keyword candidates, general keywords in the sentence can be extracted and also keywords according to the individual interest can be extracted.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional configuration of a document processing apparatus according to an embodiment.

FIG. 2 is a block diagram showing a hardware configuration of a document processing apparatus.

FIG. 3 is a flowchart showing a keyword extraction processing procedure of the document processing apparatus.

FIG. 4 is an explanatory diagram showing information stored in a user keyword holding unit 4.

FIG. 5 is an explanatory diagram showing a document to be processed and a processing result thereof.

[Explanation of symbols]

 2 sentence cutout unit 3 word cutout unit 4 user keyword holding unit 5 user keyword extraction unit 6 unnecessary word list holding unit 7 keyword candidate extraction unit 8 keyword candidate holding unit 9 keyword selection unit

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Takaya Ueda 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Yuji Ikeda 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Incorporated (72) Inventor Minoru Fujita 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc.

Claims (4)

[Claims] 1. A sentence segmentation unit for segmenting a sentence from an input document, a word segmentation unit for segmenting a word from the segmented sentence, and a user keyword holding unit for retaining a word registered by a user as a user keyword. And a user keyword extracting means for comparing the held user keyword with the cut-out word and extracting a word matching the user keyword as a keyword candidate, and an unnecessary word list in which a word not suitable for the keyword is registered. Unnecessary word list holding means for holding the word, and a keyword candidate extracting means for comparing the word registered in the unnecessary word list with the cut-out word, and extracting the cut-out word as the keyword candidate when they do not match. A keyword candidate holding means for holding the extracted keyword candidates; The information processing apparatus being characterized in that a keyword selection means for selecting the keyword from among the keyword candidates. 2. The method according to claim 1, further comprising: score giving means for giving a score according to the degree of importance to the extracted keyword candidates, wherein the keyword selecting means selects the keyword according to the given score. The information processing apparatus according to claim 1. 3. A type is set for the user keyword, and the keyword candidate is set as the most important keyword without being scored by the score assigning means to the keyword candidate that matches the user keyword of a specific type. 3. The information processing apparatus according to claim 2, wherein the keyword selection unit selects the most important keyword as the keyword. 4. A sentence is cut out from an input document, a word is cut out from the cut out sentence, a word registered by a user is held as a user keyword, and the held user keyword and the cut out word are stored. And a word matching the user keyword is extracted as a keyword candidate, and an unnecessary word list in which words that are not suitable for the keyword are registered is held, and the word registered in the unnecessary word list and the cut-out word are compared with each other. If the two do not match, the extracted word is extracted as the keyword candidate, the extracted keyword candidate is held, and the keyword is selected from the held keyword candidates. Information processing method.