JPH10334102A - Key word extraction device and medium where control program is recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To acquire a key word without requiring information depending upon a data base. SOLUTION: Information on characters as a predetermined key word is stored in a character kind information part 3. A primary key word segmenting process part 2 extracts a character string as a primary key word from the data base according to the information of the character kind information part 3 and stores it in a primary key word storage part 4. When an arbitrary primary key word stored in the primary key word storage part 4 is a composite word of primary key words, an unnecessary word removing process part 5 removes it as an unnecessary word. Consequently, the composite word is not registered in a key word information storage part 6 and the key words which constitutes the composite word are registered.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a keyword extracting device for extracting a keyword from a database.

[0002]

2. Description of the Related Art Hitherto, a configuration for extracting a keyword from text data has been considered. For example, a configuration of this type of keyword extraction is disclosed in Japanese Patent Application Laid-Open No. H8-30627. . Such a conventional keyword extracting device discloses a method of extracting words from text data based on character types. For example, in the literature of the above publication, characters are cut out from a text sentence by paying attention to character types, and the cut-out words are further divided into finer character types to increase the types of keywords. Also,
It has a dictionary of words called basic words, and has a function of deleting the same character string as the basic words.

[0003]

However, in the above-mentioned conventional keyword extracting apparatus, keywords are cut out by division according to character types, but in that case, a very large number of words are extracted. In addition, basic words are set in order to reduce keywords, and if the extracted words are basic words, they are deleted. However, in this method, it is necessary to set basic word information in consideration of the contents of the database. In addition, words formed by combining a plurality of basic words or words formed by combining a prefix or suffix with a basic word cannot be deleted.

[0004] In view of the above, it has been desired to realize a keyword extracting apparatus which can acquire a keyword without requiring information depending on a database.

[0005]

The present invention employs the following structure to solve the above-mentioned problems. <Structure of Claim 1> A character type information section having character type information indicating information of a character serving as a predetermined keyword, and a primary keyword cutout for extracting a character string serving as a primary keyword from a database based on the character type information An output processing unit; a primary keyword storage unit for storing the primary keywords extracted by the primary keyword extraction processing unit; and a primary keyword storage unit based on the plurality of primary keywords extracted by the primary keyword extraction processing unit. An unnecessary word removal processing unit that removes unnecessary primary keywords from the primary keywords and outputs the same as keyword information.

<Explanation of Claim 1> The character type information of the character type information part is information indicating information of characters that are keywords and characters that are not. For example, the non-keyword characters are all and some symbols of Hiragana, but other settings may be used. In the primary keyword extraction processing unit,
Characters to be used as keywords of character type information are extracted from the database as primary keywords and stored in the primary keyword storage. The unnecessary word removal processing unit removes unnecessary primary keywords using only the primary keyword information stored in the primary keyword storage unit and outputs the result as a final keyword. The removal process is, for example, a process of removing a word as a primary keyword as an unnecessary word when the word is a composite word of a plurality of words stored as the primary keyword. It may be a method.

[0007] With such a configuration, since the information depending on the database is not used when extracting the keyword, it is possible to apply the invention to texts in all fields.

<Structure of Claim 2> A keyword extracting device according to Claim 1, further comprising a primary keyword extraction processing unit for creating a primary keyword except for a character string in which the keyword is one character. It is.

<Explanation of Claim 2> In the case of a single character such as a kanji character, for example, a hiragana is followed by an adjective or a verb in most cases. Is not often used as a keyword. In addition, a character type other than a kanji rarely appears as a single character string. From such a viewpoint, in the extraction processing of the primary keyword, the primary keyword is created except for the character string in which the keyword character is one character. As a result, words that are not very useful for the search can be removed.

<Structure of Claim 3> According to Claim 1 or 2, of the primary keywords stored in the primary keyword storage unit, it is equal to a keyword obtained by combining a plurality of primary keywords created by the primary keyword extraction processing unit. A keyword extraction device comprising an unnecessary word removal processing unit that removes a primary keyword as an unnecessary word.

<Explanation of Claim 3> According to the invention of claim 3, when a word as a certain primary keyword is a compound word of a plurality of words stored as a primary keyword, it is removed as an unnecessary word. It was made. Thus, it is possible to eliminate words formed by connecting basic words with only the information of the primary keywords, and there is an effect that unnecessary keywords are not output.

<Structure of Claim 4> A process of extracting a keyword of a predetermined character from a database, and a process of extracting a keyword, and after the keyword is extracted, an unnecessary keyword is obtained based on a plurality of extracted keywords. A computer-readable storage medium storing a keyword extraction control program in which a process of obtaining a basic keyword is performed by a computer.

<Explanation of Claim 4> The invention of claim 4 relates to a medium in which a control program for realizing the keyword extracting device of claim 1 on a computer is recorded.

<Structure of claim 5> According to claim 3, a prefix or a prefix for storing information of a predetermined prefix and suffix.
A suffix information section; and an unnecessary word removal processing section that performs an unnecessary word removal process on a character string excluding the prefix or the suffix among the primary keywords stored in the primary keyword storage section. Is a keyword extraction device.

<Explanation of Claim 5> The invention of claim 5 includes, in addition to the invention of claim 3, information of a prefix / suffix,
When a word obtained by removing a prefix or a suffix from a primary keyword is a compound word, this is removed from the primary keyword. As a result, even if the prefix or the suffix is not included in the primary keyword, it is possible to remove a compound word including the prefix and the suffix.

<Structure of Claim 6> In claim 3, in accordance with Claim 3, information of a predetermined basic character string and a modified character string composed of different character types for modifying the basic character string is stored. A partial character string information section, and an unnecessary word removal processing section for outputting a final keyword by removing a keyword existing in the primary keyword from the primary keywords stored in the primary keyword storage section. It is a keyword extraction device characterized by comprising:

<Explanation of Claim 6> The character type of the basic word character string and the modifier word character string is, for example, as follows.
And “Katakana”, and the modifier character string is “Kanji”, but the character type of the character string that is a modifier for the basic character string is not limited to this.

With such a configuration, in order to remove a certain keyword, it is not necessary that all of the constituent partial character strings exist in the primary keyword, and therefore, it is possible to further output a keyword useful for retrieval. There is.

<Structure of claim 7> A keyword extracting apparatus according to claim 6, further comprising an unnecessary word removal processing unit that does not remove a primary keyword whose leading character string is a modifier character string as an unnecessary word. It is.

<Explanation of Claim 7> The invention of claim 7 is based on the fact that, when a modifier character string is present at the beginning, it is often a useful keyword for the entire character string. In the case of, it is not removed.

<Structure of Claim 8> In Claim 6 or 7, the character type of the basic word character string is English letters, numbers and katakana, and the character type of the modifier word character string is Kanji. This is a keyword extraction device.

<Explanation of Claim 8> The invention of claim 8 specifies the character type of the basic character string and the modifier character string. Thus, in order to remove a certain keyword, it is not necessary that all of the constituent character strings are present in the primary keyword, and such a combination of character types can be removed.

<Structure of Claim 9> In Claim 3, an unnecessary word information section for storing information of a predetermined unnecessary word,
When extracting a primary keyword, a primary keyword extraction processing unit that excludes an unnecessary word from the primary keyword if it is an unnecessary word stored in the unnecessary word information unit, and a compound word that stores information of a predetermined compound word A keyword extraction device comprising: an information unit; and an unnecessary word removal processing unit that does not perform unnecessary word removal processing on a character string equal to a composite word among primary keywords stored in a primary keyword storage unit.

<Description of Claim 9> According to the invention of claim 9, the information of the unnecessary word which is not extracted as the primary keyword and the word which is determined to be a compound word in the unnecessary word removal processing is provided.
It is provided with compound word information which is not removed as a compound word. Thus, unnecessary primary keywords can be removed at the stage of the primary keyword extraction process. In addition, we find a word that is a compound word but should be registered as a keyword,
Can be saved.

<Structure of Claim 10> In Claim 9,
A non-target character type information section that stores information on non-target character types that are not a predetermined compound word, and a portion of a character string composed of one type of non-target character type among the primary keywords stored in the primary keyword storage section Is a keyword extraction device provided with an unnecessary word removal processing unit that does not process as a synthesis part.

<Explanation of Claim 10> According to the invention of claim 10, even if a word determined to be a compound word in the unnecessary word removing process is a compound word of one kind, the compound word Is provided with character type information that is not removed.
This character type is, for example, katakana or English characters. Thereby, even if the "key" and the "word" exist in the primary keyword, a word such as "keyword" can be output as a keyword.

<Structure of Claim 11> Claim 9 or 10
In the case of performing unnecessary word removal processing among the primary keywords stored in the primary keyword storage unit, together with the primary keywords stored in the primary keyword storage unit,
A keyword extraction device comprising an unnecessary word removal processing unit that uses unnecessary words stored in unnecessary word information.

<Explanation of Claim 11> In the invention of claim 11, unnecessary word information is used in the unnecessary word removal processing in the same manner as a primary keyword. As a result, it is possible to find and remove a compound word based on a word that does not exist in the database.

<Structure of Claim 12> In Claim 3,
A statistical information part indicating the number of appearances of each primary keyword in the database, and among the primary keywords stored in the primary keyword storage part, a word determined as a composite word is predetermined as a primary keyword of the statistical information part. A keyword extraction device comprising: an unnecessary word removal processing unit that does not remove as an unnecessary word if the number of appearances is higher than a threshold value.

<Description of Claim 12> According to the invention of claim 12, in the unnecessary word removal processing, if the number of appearances of a word determined as a compound word in the database is equal to or more than a certain value, the word is output as a keyword. It was done. Thus, it is not necessary to have all the valuable composite words as keywords as information, and such composite words can be automatically determined and stored.

<Structure of Claim 13> In Claim 3,
Among a plurality of data stored in the database, the number of data including a specific primary keyword is determined as a compound word among a statistical information part corresponding to the primary keyword and a primary keyword stored in the primary keyword storage. An unnecessary word removal processing unit that does not remove the word as an unnecessary word as a primary keyword of the statistical information unit when the number of data is higher than a predetermined threshold value. It is a keyword extraction device.

<Explanation of Claim 13> According to the invention of claim 13, in the unnecessary word removal processing, if the number of data in a database having a word determined as a compound word is a certain value or more, it is output as a keyword. It was made. That is, in the present invention, the number of data in which a word appears is used, and it does not matter how many times the word appears in one data. With such a configuration, even in a database in which specific words frequently appear in specific data, keyword extraction without bias can be performed.

<Structure of Claim 14> Claim 12 or 1
3, the keyword extraction device is characterized in that the case where the value is higher than the threshold value is a case where the value of a word determined as a compound word is higher than the value of a basic word constituting the word.

<Explanation of Claim 14> The invention of claim 14 relates to, for example, a case where the number of appearances of the composite word “mobile phone” is higher than the number of occurrences of the basic words “mobile phone” and “telephone”. It is defined as higher than the threshold. Thereby, a word useful as a keyword can be output.

[0035]

Embodiments of the present invention will be described below in detail with reference to the drawings. << Specific Example 1 >> Specific example 1 is based on data from a database.
First, a primary keyword is created, and then a compound word is deleted from the obtained primary keyword.

<Structure> FIG. 1 is a diagram showing a specific example 1 of a keyword extracting apparatus according to the present invention. The apparatus shown in the figure is constituted by, for example, a microcomputer or the like, and comprises a database 1, a primary keyword extraction processing unit 2, a character type information unit 3, a primary keyword storage unit 4, an unnecessary word removal processing unit 5, and a keyword information storage unit 6. .

The database 1 is an aggregate of data.
Each data is stored in a hard disk or the like, and each data has a text portion.

The primary keyword extraction processing unit 2 reads out each text part from the database 1 and
Has a function of extracting a character string (primary keyword) that is likely to be a keyword using the character type information. The character type information section 3 stores information on the type of character cut out from the text.

FIG. 2 is an explanatory diagram of the contents of the character type information section 3. In the character type information section 3, characters that can be keywords are
It indicates information with characters that are not. Here, characters that can be keywords are called keyword characters, and characters that are not are called non-keyword characters. All characters always belong to only one of them, and the character type information section 3 describes all non-keyword characters as shown in the figure. That is, non-keyword characters are defined as all and some symbols (, .., etc.) in Hiragana.

The primary keyword storage unit 4 is configured on a hard disk or a memory, and includes a primary keyword extraction processing unit 2.
This is a storage unit for storing the primary keywords extracted by. The unnecessary word removal processing unit 5 deletes a word that is considered unnecessary from the primary keywords stored in the primary keyword storage unit 4 after receiving the signal after the primary keyword extraction processing unit 2 has completed the primary keyword extraction processing. The processing unit generates the final keyword. The keyword information storage unit 6 is a storage unit that is configured on a hard disk or a memory and stores the keyword output from the unnecessary word removal processing unit 5.

<Operation> First, the target database 1
Is configured as follows. FIG. 3 is an explanatory diagram of the contents of the database 1. Here, the data number is
The text part is a number assigned to a group of data, and the text part represents the data of the text part of each data.

The primary keyword extracting unit 2 reads the database 1 and performs a primary keyword extracting process on the text part of each data. This extraction processing is performed by extracting a character string composed of keyword characters in the text based on the character type information of the character type information unit 3. That is, a character string composed of characters other than the non-keyword characters indicated in the character type information section 3 is extracted. At this time, even if it is a keyword character, a character string of one character is deleted as a target. This is because in the case of one kanji character, hiragana is followed by an adjective or verb in most cases, and such words are rarely used as search keywords. In addition, a character type other than a kanji rarely appears as a single character string. By this processing, words that are not very useful for the search can be removed. For example, in the example shown in FIG. 3, “key”, “table”, and “difference” of data number 3 are composed of keyword characters, but are not output to the primary keyword storage unit 4 because they are one character.

FIG. 4 is an explanatory diagram showing primary keywords extracted by the primary keyword extraction processing unit 2. As shown in the figure, the non-keyword characters shown in the character type information section 3 and character strings other than one character string are extracted as primary keywords for each data number.

The unnecessary word removing section 5 removes unnecessary words from the primary keywords obtained in this way. In this process, a primary keyword equal to a character string formed by combining a plurality of primary keywords is removed. Here, a character string formed by combining a plurality of primary keywords is called a composite word. That is, a compound word is a "word formed by connecting a plurality of meaningful and independent minimum unit words (= basic words)".

FIG. 5 is a flowchart of the unnecessary word removing process. In the figure, check (w [1, n]) is a process of determining whether the primary keyword w [1, n] of the target n characters is a removed character string (a compound word to be removed). Note that w [x, y] is obtained from the x character of the target character string to y.
Represents a substring up to the character. For example, “development cost” stored in the primary keyword storage unit 4 will be described as an example of the target character string. Therefore, in this case, ch
eck (w [1,5]).

In such unnecessary word removal processing, first, i
= 1 (step S1). That is, the process starts from the first character. Then, since the target character string is two or more characters, the value of i is incremented by 1 and the second character is targeted (step S2), and it is determined in the next step S3 whether i = n. That is, it is determined whether the target character string is only two characters. In this step S3, if the target character string is only two characters, it becomes false and the word is left as it is. This is because, as described above, since one character is not a target as a keyword, there is no compound word of two characters.

On the other hand, if i = n is not satisfied in step S3, that is, if the target character string is three or more characters like "development cost", the process proceeds to step S4.
It is determined whether w [1, i] exists in the primary keyword. In this case, since “development” is to be determined and the primary keyword “development” exists in data number 3, the process proceeds to step S5, and it is determined whether i = n. This step S5
Since i is not equal to n, the process proceeds to step S6.
If w [1, i] does not exist in the primary keyword in step S4, the process returns to step S2.

In step S6, check (w [i +
1, n]). In this example, it is determined whether the “cost” of the third to fifth characters exists in the primary keyword. In this case, similarly to the above-described process, it is determined whether a three-character character string exists in the primary keyword. As a result, since "cost" exists in data number 2,
The process proceeds from step S4 to step S5. Since i = n in step S5, the result is true, and the word “development cost” is deleted from the primary keyword.

By applying such processing to all the primary keywords, the compound word is removed. For example, as described above, the primary keyword “development cost” of data number 1 is a composite word of “development” of data number 3 and “cost” of data number 2. Also, data number 4
“Sales trend” is a composite word of “sales” of data number 2 and “trend” of data number 1. Therefore, these two words are removed as keywords. FIG. 6 is an explanatory diagram of the finally obtained keyword.

<Effect> As described above, in the first embodiment, it is possible to automatically assign a keyword from data including a text portion. In addition, there is an effect that words formed by connecting basic words can be removed. Further, in extracting such a keyword, since the information depending on the database is not used, the present specific example can be applied to texts in all fields.

<< Specific Example 2 >> In the specific example 2, in addition to the configuration of the specific example 1, prefix / suffix information is provided, and unnecessary word removal processing is performed based on the prefix / suffix information. It is.

<Structure> FIG. 7 is a diagram showing the structure of a second embodiment of the keyword extracting apparatus according to the present invention. The apparatus shown in the figure includes a database 1, a primary keyword extraction processing unit 2, a character type information unit 3,
It comprises a primary keyword storage unit 4, an unnecessary word removal processing unit 5a, a keyword information storage unit 6, and a prefix / suffix information unit 7. Here, database 1 to keyword information storage unit 6
Among them, the configuration excluding the unnecessary word removal processing unit 5a is the specific example 1.
The description is omitted here.

The prefix / suffix information section 7 is an information storage section for a predetermined prefix and suffix, and is configured as follows, for example. FIG. 8 is an explanatory diagram of the contents of the prefix / suffix information section 7.

Further, the unnecessary word removal processing section 5a has a specific example 1.
In addition to the functions described above, among the primary keywords stored in the primary keyword storage unit 4, for the character string excluding at least one of the prefix or suffix stored in the prefix / suffix information unit 7 And has a function of performing unnecessary word removal processing. That is, the primary keyword has a function of removing a primary keyword whose partial character string excluding the prefix or suffix already exists as the primary keyword in the primary keyword storage unit 4.

<Operation> The description of the operation common to the first embodiment will be omitted, and only the characteristic operation of the second embodiment will be described.

The unnecessary word removal processing by the unnecessary word removal processing section 5a is performed in the following procedure. 1. If the first character string of the target primary keyword matches the prefix, the prefix is deleted. 2. If the character string at the end of the target primary keyword matches the suffix, the suffix part is deleted. 3. It is determined whether the character string remaining after removing the prefix and suffix exists as a primary keyword. If it exists, delete the target primary keyword.

By performing the above processing for all the primary keywords, it is possible to remove unnecessary words.
For example, to explain using the example of the specific example 1, first, assuming that the target database 1 is given as shown in FIG. 3, the primary keyword generated by the extraction processing of the primary keyword is shown in FIG. Become like Here, "each manufacturer" of the data number 2 has a prefix "each" defined in the prefix / suffix information part 7, and this prefix "each".
The character string "maker" excluding is present in the primary keyword. Therefore, since each of the conditions matches the removal condition, “each maker” removes the primary keyword. Also, “manufacturer-specific” of data number 3 also has the suffix “other”, and a character string “maker” excluding the suffix “other” exists in the primary keyword. Therefore, “by manufacturer” is also eliminated.
Similarly, this process is performed for all primary keywords to obtain a final keyword.

FIG. 9 shows the keywords extracted in this way. In order to perform the unnecessary word removal processing based on the prefix / suffix, for example, as compared with the keyword of the specific example 1 shown in FIG. 6, “each manufacturer” of data number 2 and “by manufacturer” of data number 3 Has been removed.

<Effects> As described above, according to the specific example 2, the following effects are obtained in addition to the specific example 1. That is, in the first specific example, in order to remove a certain keyword, all of the constituent partial character strings had to be present in the primary keyword. On the other hand, in the specific example 2, when a prefix or a suffix is attached to a basic word and cut out as a primary keyword, the effect that the prefix or the suffix can be removed even if it does not exist in the primary keyword is obtained. is there.

<< Specific Example 3 >> In the specific example 3, in addition to the structure of the specific example 1, the character type of the predetermined basic word character string, the basic character string is modified, and the character type is different. The apparatus is provided with partial character string information for storing information on the composed decoration character string, and performs an unnecessary word removal process based on the partial character string information. The basic word character string indicates a character string that is valid as a keyword.

<Structure> FIG. 10 is a diagram showing the structure of a third embodiment of the keyword extracting apparatus according to the present invention. The apparatus shown in the figure includes a database 1, a primary keyword extraction processing unit 2, a character type information unit 3, a primary keyword storage unit 4, an unnecessary word removal processing unit 5b,
A keyword information storage unit 6 and a partial character string information unit 8 are provided. Here, database 1 to keyword information storage unit 6
Among them, the configuration excluding the unnecessary word removal processing unit 5b is the specific example 1.
The description is omitted here.

The purpose of this specific example is to remove a character string formed by connecting a character string effective as a keyword (= basic word character string) with a modifier character string. Here, the character type of a character string valid as a keyword is called a basic word character type, and the character type of a modifier-like character string is called a modifier character type. For this reason, the partial character string information section 8 describes rules in which English characters, numbers, and katakana are used as basic word character types, and kanji are used as modifier word character types.

The unnecessary word removal processing unit 5b, based on the rules of the partial character string information unit 8, sets the character type of the primary keyword in the primary keyword storage unit 4 to “English character or number or katakana” + “Kanji”. Is a string consisting of
It is divided into “English characters or numbers or katakana” and “Kanji”, and if the “Kanji” part exists as a primary keyword, the primary keyword is deleted. Also, if a modifier character string is present at the beginning, it is often a useful keyword for the entire character string. Therefore, such a character type arrangement is not removed.

<Operation> FIG. 11 is a flowchart of the unnecessary word removing process of this embodiment. FIG. 12 is an explanatory diagram of a primary keyword in this specific example.

The unnecessary word removal processing section 5b first divides the primary keyword into a basic word character string and a modifier word character string based on the information of the partial character string information section 8 (step S1). For example, “computer industry” and “new memory” of data number 1 and “EC leader” of data number 2 shown in FIG. 12 are subject to division because the character type is “Katakana + Kanji”, and “computer” and “ It is divided into "Industry", "New" and "Memory" and "EC" and "Leader".

Next, in step S2, it is determined whether or not the first partial character string is a basic word character string. Here, since the first partial character string of “computer”, “industry”, “EC”, and “leader” is the basic word character string, the process proceeds to the next step S3. Since the character string is a modifier character string, the character string does not satisfy the removal condition, and the process proceeds to step S5 to be “Return”.

In step S3, it is determined whether all the modifier character strings exist as primary keywords. For example, in the case of “computer” and “industry”, since the modifier character string “industry” exists in the primary keyword, the process proceeds to step S4, in which “computer industry” is removed, and “Retur” is removed.
n ”(step S). In addition, "EC""leaders"
Also, since the modifier character string “leader” exists in the primary keyword, “EC leader” is also removed. By performing such processing for all primary keywords, a final keyword is obtained. FIG. 13 is an explanatory diagram of keywords finally obtained.

<Effects> As described above, according to the third embodiment,
The following effects are obtained in addition to the effects of the first embodiment. That is,
In the first specific example, in order to remove a certain keyword, all of the constituent character strings must exist in the primary keyword, but in this specific example, basic words having different character types are combined and cut out. If a kanji character string exists as a primary keyword in such a case, such a word can be eliminated.

Moreover, even in the specific example 2, the word in which the modifier word is combined with the basic word is removed, but in the specific example 2, the modifier word needs to be specifically described. On the other hand, in this specific example, this is not necessary, and efficient unnecessary word removal can be performed.

<< Example 4 >> In Example 4, when extracting a primary keyword, unnecessary word information indicating an unnecessary word to be removed as a primary keyword, and synthetic word information indicating a synthetic word to be left in the unnecessary word removal processing And non-target character type information indicating character type information of a character string that is not removed by the unnecessary word removal processing.

<Structure> FIG. 14 is a diagram showing the structure of a fourth embodiment of the keyword extracting apparatus according to the present invention. The apparatus shown in the figure includes a database 1, a primary keyword extraction processing unit 2, a character type information unit 3, a primary keyword storage unit 4, an unnecessary word removal processing unit 5c,
It comprises a keyword information storage section 6, an unnecessary word information section 9, a composite word information section 10, and a non-target character type information section 11. Here, the configuration of the database 1 to the keyword information storage unit 6 except for the primary keyword extraction processing unit 2a and the unnecessary word removal processing unit 5c is the same as that of the first embodiment, and thus the description thereof is omitted.

The unnecessary word information section 9 is a storage section for information indicating a word which is a character string composed of one or more keyword characters but is unnecessary as a keyword. FIG. 15 is an explanatory diagram of the contents of the unnecessary word information section 9.

When the character string extracted from the database 1 matches a word in the unnecessary word information section 9, the primary keyword extraction processing section 2a stores the word in the primary keyword storage section 4.
Is configured not to output. The unnecessary words shown in the unnecessary word information section 9 are treated as words equivalent to the primary keywords when the unnecessary word removal processing section 5c performs the composite word removal processing described in the first specific example.

The synthesized word information section 10 extracts a character string to be extracted as an independent keyword from a plurality of primary keywords or a character string which may be formed by combining unnecessary words indicated in the unnecessary word information section 9. Is stored.

FIG. 16 is an explanatory diagram of the contents of the composite word information section 10. As shown in the figure, the words described in the composite word information section 10 are words having a unique meaning in which two words are combined, such as “mobile phone”.

In the non-target character type information section 11, even when a plurality of primary keywords are connected in the unnecessary word removal processing in the unnecessary word removal processing section 5c, if all of the primary keywords are of the same character type, the word is not a composite word. The information of the character type not to be stored is stored. Here, the character types are English characters,
Numbers and katakana are specified.

The unnecessary word removal processing unit 5c treats the word indicated in the unnecessary word information unit 9 as a word equivalent to the primary keyword as described above when performing the compound word removal process. The word shown in 10 and the word of the character type specified in the non-target character type information section 11 are configured to be processed as non-target words as composite words.

<Operation> In this specific example, for example, it is assumed that the text to be extracted by the primary keyword extraction processing unit 2a is "a sudden increase in the use of mobile phones has completely changed the way of telephones".

First, the primary keyword extraction processing section 2a divides the target text into keyword characters, as in the first to third examples, and extracts a character string composed of two or more characters from the text. . As a result, (mobile, sudden,
(Increase in popularity, telephone, change). Of these, "rapid" is indicated in the unnecessary word information section 9 (see FIG. 15), and thus is removed as a primary keyword.
Therefore, the primary keyword stored in the primary keyword storage unit 4 is (mobile phone, increase in spread, telephone, change).

The unnecessary word removal processing section 5 first checks whether or not each primary keyword is described in the composite word information section 10. If the word is described in the composite word information section 10, the word is output to the keyword information storage section 6. For example, since “mobile phone” is described in the composite word information section 10, the composite word determination is skipped and output to the keyword information storage section 6. On the other hand, for primary keywords that are not described in the composite word information section 10, the following unnecessary word removal processing is performed.

FIG. 17 is a flowchart of the unnecessary word removing process. In the figure, check (w [1, n]) is a character string in which the primary keyword w [1, n] of the target n characters is a removed character string (synthesized to be removed), as in the specific example 1 shown in FIG. Word).

In the unnecessary word removing process, i = 1 and ct = 0 in step S1, and the value of i is incremented by one in step S2, as in the operation of the first embodiment. Step S
Ct in 1 indicates a character type, and ct = 0 indicates that information on the character type is reset. Next, it is determined in step S3 whether i = n. That is, it is determined whether the target character string is only two characters. In this step S3, if the target character string is only two characters, fa
1se, and leave the word as it is. This is because, as in the first embodiment, there is no compound word of two characters.

On the other hand, if i = n is not satisfied in step S3, for example, if the target character string has three or more characters, such as "increase in spread", the process proceeds to step S4, where w
It is determined whether [1, i] exists in the primary keyword or unnecessary word information section 9. In this case, “spread” is a determination target, and since this word exists in the unnecessary word information section 9, the process proceeds to the next step S5.

In step S5, it is determined whether the target word is a character type described in the non-target character type information section 11. In this case, since the character type is not a non-target character type, the process proceeds to step S8 to determine whether i = n. In step S8, since i = n is not satisfied, the process proceeds to step S9, where check
k (w [i + 1, n]) is performed. Since “increase” is also described in the unnecessary word information section 9, the result is true, and it is determined that the word is a composite word. Therefore, it is not output to the keyword information storage 6.

Steps S5 to S7 are as follows:
It is provided to perform the following processing. That is, when words of a specific character type are excluded from the target in the compound word removal process, the accuracy of keyword extraction may be improved. For example, words composed of alphanumeric characters, katakana, and the like are accidentally compound words, and even compound words are important words in many cases. For example, “in” + “put” = “input”
Or “key” + “word” = “keyword”.
Therefore, by specifying such a character type in the non-target character type information unit 11, a compound word based on these character types can be excluded from the compound words to be removed.

For example, it is assumed that “key”, “word”, and “keyword” exist as primary keywords. Assuming that the word "keyword" is targeted in the unnecessary word removal processing, first, a basic word is found in a partial character string "key" (steps S1 to S4). And
In step S5, since the "keyword" is one kind of non-target character type, the process proceeds to step S6. Step S
In 6, it is determined whether the character type of ct and w [1, i] are the same. However, since ct has been reset in step S1, it is determined that the character type is different, and the process proceeds to step S7.
The information called katakana is stored in ct.

The remaining character string “word” also exists in the primary keyword, but “word” is katakana and is determined to be the same character type as ct.
Returning from step S6 to step S2, since no more character strings remain, finally, from step S3 false
Becomes

Other than this, for example, “NG scene”, “N
Consider the case where the words "G" and "scene" exist as primary keywords. In this case, both “NG” and “scene” are non-target character types, but “NG” is an English character and “scene” is katakana.
Since the character types of ct and w [3,5] do not match, the process proceeds to step S8. Here, since the character string no longer remains, it is finally true, and the “NG scene” is determined to be a composite word, and is not output to the keyword information storage unit 6.

With the above processing, the above-mentioned text "The rapid increase in the use of mobile phones has completely changed the way telephones work."
Will be (mobile phone, phone, all-in-one).

<Effects> As described above, according to the fourth embodiment,
The following effects are obtained in addition to the effects of the first embodiment. That is,
By providing the unnecessary word information section 9, unnecessary words can be removed at the stage of the primary keyword extraction processing. Further, since the compound word is decomposed using the unnecessary word information section 9, it is possible to find and remove a compound word by a word that does not exist in the database 1. Further, the compound word information section 1
0 has the effect that a word that is a compound word but should be registered as a keyword can be found and stored.

<< Example 5 >> In Example 5, a statistical value of a primary keyword is obtained, and unnecessary word removal processing is performed using the statistical value.

<Structure> FIG. 18 is a diagram showing the structure of a fifth embodiment of the keyword extracting apparatus according to the present invention. The apparatus shown in the figure includes a database 1, a primary keyword extraction processing unit 2b, a character type information unit 3, a primary keyword storage unit 4, an unnecessary word removal processing unit 5d,
It comprises a keyword information storage section 6 and a statistical information section 12. Here, the configuration of the database 1 to the keyword information storage unit 6 except for the primary keyword extraction processing unit 2b and the unnecessary word removal processing unit 5d is the same as that of the first embodiment, and thus the description thereof is omitted.

The primary keyword extraction processing unit 2b extracts primary keywords from the database 1, stores them in the primary keyword storage unit 4, and outputs them to the statistical information unit 12, as in the above specific examples. Have been. Further, the statistical information section 12 includes a primary keyword cutout processing section 2b.
This is a storage unit for recording statistical information (statistical index) related to the primary keyword in the database 1 based on the information from. The statistical information is the number of appearances of the primary keyword in the database 1.

FIG. 19 is an explanatory diagram of statistical information. In this statistical information, a primary keyword is used as a heading, and an integer variable corresponding to each heading is used as the number of appearances in the database 1.

The unnecessary word removing section 5d uses the statistical information shown in the statistical information section 12 when performing the compound word removing processing, and uses the statistical information stored in the primary keyword storing section 4 for the primary keyword. When a word determined as a compound word has a higher number of appearances than a predetermined threshold, the word is not removed as an unnecessary word.

<Operation> In the primary keyword extraction processing unit 2b, when the primary keyword is extracted from the database 1,
This primary keyword is stored in the primary keyword storage unit 4 and output to the statistical information unit 12. Statistical information section 12
Then, an integer variable whose index is the input word is 1
increase. If there is no integer variable with the input word as an index, a new integer variable with the word as an index is created and 1 is substituted as an initial value. Therefore, when the primary keyword extraction processing is completed, the statistical information section 12 stores the primary keywords and the number of appearances of each primary keyword in the database 1 as a set.

In the statistical information section 12, the number of appearance data may be used instead of the number of appearances. The number of appearing data refers to the number of data in which a word appears when there are a plurality of data in the database 1, and it does not matter how many times the word appears in one data. In the case of a database in which specific words frequently appear in specific data, inconvenience often occurs in the number of appearances in the data. For a database having such properties, it is better to use the number of occurrence data.

Next, the unnecessary word removing section 5d removes unnecessary compound words based on the statistical information of the statistical information section 12.
In this removal process, first, for each primary keyword, FIG.
The unnecessary word determination process of the specific example 1 shown in (1) is performed to check whether the word is a compound word. Next, the statistical information of the primary keyword determined to be a compound word and the basic word constituting the primary keyword as the compound word is extracted from the statistical information unit 12. Then, the statistical information of the compound word and the basic word is compared, and when it is determined that the compound word is useful, the primary keyword is output to the keyword information storage unit 6. Here, as a criterion for determining that a compound word is useful, it is conceivable that the number of appearances of the compound word is larger than the number of occurrences of any of the constituent basic words.

For example, when the statistical information section 12 resulting from the extraction of the primary keyword from the database 1 is in a state as shown in FIG. 19, “mobile phone” is a combination of the basic words “mobile” and “phone”. Although these words are more than the number of appearances of these basic words, they are determined to be useful words and are output as final keywords.

<Effects> As described above, according to the fifth embodiment,
The following effects are obtained in addition to the effects of the first embodiment. That is,
In the specific example 4, all the compound words having a value as a keyword must be held as information, but in this specific example, such a compound word can be automatically determined and stored. I have.

In each of the above specific examples, the final keyword is stored in the keyword information storage unit 6.
The obtained keywords may be integrated by adding them to the database 1 as additional information.

Further, all the operations of each of the above specific examples can be realized by control by a program of a computer serving as a keyword extracting device. Therefore, a method of recording the program on a recording medium such as a floppy disk or a CD-ROM and then installing the program on a computer, or downloading and installing the program from a network, or a method of previously installing the program on a hard disk or the like Thereby, the keyword extracting device of the present invention can be realized.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a specific example 1 of a keyword extraction device of the present invention.

FIG. 2 is an explanatory diagram of the contents of a character type information section in the keyword extraction device of the present invention.

FIG. 3 is an explanatory diagram of the contents of a database in a specific example 1 of the keyword extracting device of the present invention.

FIG. 4 is an explanatory diagram of an extracted primary keyword in a specific example 1 of the keyword extraction device of the present invention.

FIG. 5 is a flowchart of an unnecessary word removing process in the specific example 1 of the keyword extracting device of the present invention.

FIG. 6 is an explanatory diagram of a keyword obtained in a specific example 1 of the keyword extracting device of the present invention.

FIG. 7 is a configuration diagram of a specific example 2 of the keyword extraction device of the present invention.

FIG. 8 is a diagram illustrating the contents of a prefix / suffix information section in a specific example 2 of the keyword extracting device of the present invention.

FIG. 9 is an explanatory diagram of keywords acquired in a specific example 2 of the keyword extracting device of the present invention.

FIG. 10 is a configuration diagram of Example 3 of the keyword extraction device of the present invention.

FIG. 11 is a flowchart of an unnecessary word removing process in a specific example 3 of the keyword extracting device of the present invention.

FIG. 12 is an explanatory diagram of a primary keyword in a specific example 3 of the keyword extraction device of the present invention.

FIG. 13 is an explanatory diagram of a keyword obtained in a specific example 3 of the keyword extracting device of the present invention.

FIG. 14 is a configuration diagram of a specific example 4 of the keyword extraction device of the present invention.

FIG. 15 is an explanatory diagram of the contents of an unnecessary word information section in a specific example 4 of the keyword extracting device of the present invention.

FIG. 16 is an explanatory diagram of the contents of a compound word information part in a specific example 4 of the keyword extracting device of the present invention.

FIG. 17 is a flowchart of an unnecessary word removing process in a specific example 4 of the keyword extracting device of the present invention.

FIG. 18 is a configuration diagram of a specific example 5 of the keyword extraction device of the present invention.

FIG. 19 is an explanatory diagram of statistical information in a specific example 5 of the keyword extraction device of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Database 2, 2a, 2b Primary keyword extraction processing part 3 Character type information part 4 Primary keyword storage part 5, 5a, 5b, 5c, 5d Unnecessary word removal processing part 6 Keyword information storage part 7 Prefix / suffix information part 8 Partial character string information part 9 Unnecessary word information part 10 Compound word information part 11 Non-target character type information part 12 Statistical information part

Claims (14)

[Claims] 1. A character type information section having character type information indicating information of a character serving as a predetermined keyword, and a primary keyword extraction for extracting a character string serving as a primary keyword from a database based on the character type information. A processing unit, a primary keyword storage unit that stores the primary keywords extracted by the primary keyword extraction processing unit, and a plurality of primary keywords extracted by the primary keyword extraction processing unit, based on the primary keyword storage unit A keyword extraction device comprising: an unnecessary word removal processing unit that removes unnecessary primary keywords from stored primary keywords and outputs the same as keyword information. 2. The keyword extraction device according to claim 1, further comprising a primary keyword extraction processing unit that creates a primary keyword except for a character string in which the keyword character is one character. 3. The unnecessary keyword according to claim 1, wherein, among the primary keywords stored in the primary keyword storage, a primary keyword equal to a keyword obtained by combining a plurality of primary keywords created by the primary keyword extraction processing unit is used. A keyword extraction device comprising an unnecessary word removal processing unit for removing a keyword. 4. A process of extracting a keyword of a predetermined character from a database, and after the keyword extraction process, an unnecessary keyword is obtained based on a plurality of extracted keywords, and a final keyword is determined. A medium on which a keyword extraction control program is recorded, wherein the processing for obtaining is performed by a computer. 5. The prefix or suffix information unit for storing information of a predetermined prefix and suffix, and the prefix or suffix of the primary keywords stored in the primary keyword storage unit. A keyword extraction device comprising an unnecessary word removal processing unit for performing unnecessary word removal processing on a character string excluding a suffix. 6. The partial character string information according to claim 3, wherein a predetermined basic word character string and a modified character string composed of different character types for modifying the basic word character string are stored. And an unnecessary word removal processing unit that outputs the final keyword by removing the keyword existing in the primary keyword from the primary keyword stored in the primary keyword storage unit. Keyword extraction device featuring. 7. The keyword extracting device according to claim 6, further comprising an unnecessary word removal processing unit that does not remove a primary keyword whose first character string is a modifier character string as an unnecessary word. 8. The keyword extracting device according to claim 6, wherein the character types of the basic character string are English characters, numbers, and katakana characters, and the character type of the modifier character string is a Chinese character. 9. The unnecessary word information section for storing information of a predetermined unnecessary word, and a case where a primary keyword is extracted, wherein the unnecessary word is stored in the unnecessary word information section. Is a primary keyword extraction processing unit to be excluded from the primary keyword, a composite word information unit that stores information of a predetermined composite word, and is equal to the composite word among the primary keywords stored in the primary keyword storage unit. A keyword extraction device comprising: an unnecessary word removal processing unit that does not perform unnecessary word removal processing on a character string. 10. The non-target character type information section for storing information on a non-target character type that is not a predetermined compound word and a primary keyword stored in a primary keyword storage section according to claim 9, wherein A keyword extraction device comprising an unnecessary word removal processing unit that does not process a character string portion composed of a non-target character type as a combined portion. 11. The unnecessary keyword along with the primary keyword stored in the primary keyword storage unit when performing the unnecessary word removal processing among the primary keywords stored in the primary keyword storage unit. A keyword extraction device comprising an unnecessary word removal processing unit that uses unnecessary words stored in information. 12. The statistical information section according to claim 3, wherein a statistical information section indicating the number of appearances of each primary keyword in the database, and a primary keyword stored in the primary keyword storage section, wherein the word determined as a compound word is the statistical keyword. A keyword extracting device, comprising: an unnecessary word removal processing unit that does not remove, as a primary keyword of an information part, an unnecessary word when the number of appearances is higher than a predetermined threshold value. 13. The statistical information section according to claim 3, wherein the number of data including a specific primary keyword among a plurality of data stored in the database is stored in a primary keyword storage section corresponding to the primary keyword. In the case where the word determined as a compound word among the primary keywords thus determined has a data count higher than a predetermined threshold value as the primary keyword of the statistical information section, the unnecessary word not removed as an unnecessary word A keyword extraction device comprising a removal processing unit. 14. The method according to claim 12, wherein the case where the value is higher than the threshold value is a case where a value of a word determined as a compound word is higher than a value of a basic word constituting the word. Keyword extraction device.