JPH09237281A - Text data retrieving device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to register also an unregistered word mixedly including HIRAGANA (cursive form of Japanese syllabary), KANJI (Chinese character), etc., in a dictionary by providing the text data retrieving device with a means for judging whether a key word is a word registered in a dictionary file or not and a means for registering a key word in the dictionary file as a word when the key word is not registered. SOLUTION: A morpheme analyzing means 3 executes morpheme analysis for text data inputted by a dictionary file 4. A retrieving means 8 retrieves a word coincident with a key word received by a key word receiving means 7 within words divided by the means 3 and outputs a retrieved result. A judging means judges whether an inputted key word is a word not registered in the file 4 or not, and in the case of a word not registered in the file 4, a registering means 9 registers the key word in the file 4. In addition, the means 9 registers a key word received by the means 7 more than the prescribed number of times as a word.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text data search device and a text data search method for searching text data using an input keyword.

[0002]

2. Description of the Related Art Recently, an index for full-text search is created in advance on a word-by-word basis for text data written in a language such as Japanese, which is not separated (words are separated from each other by spaces). Incidentally, there has been proposed a text data search device capable of searching text data using this index when a keyword is input. This text data search device performs a morphological analysis process that first divides text data input using a dictionary into word units, and creates an index that registers words such as nouns and verbs based on the results of the morphological analysis process. I'll do it. Then, when the keyword is input, it is determined whether the created index includes the keyword. In this determination, if the index includes the keyword, the text data of the corresponding portion (the portion including the keyword) is output, and if the index does not include the keyword, the corresponding text data does not exist. A message to that effect is displayed.

However, when the morphological analysis process is performed on the text data including the word that is not registered in the dictionary, the word division fails. For example, if the word "ra-chi" is not registered in the dictionary, morphological analysis processing is executed on the text data "rare-arrested.""Suspect""De"
It will be divided into "arrest", "sa", "re" and "ta",
The word "Latch" is not registered in the index. Therefore, when a word that is not registered in the dictionary, such as “Latch”, is input as a keyword, there is a problem that the text data cannot be searched correctly. Therefore, when a word that is not registered in the dictionary (hereinafter referred to as an unregistered word) is detected from text data or the like, it is important to register the unregistered word in the dictionary.

[0004]

As a method of registering the unregistered word in the dictionary, when a character string in which the same character type such as kanji or katakana continues is not registered in the dictionary, this character string is not registered. There is a method of detecting it as a word and registering it in a dictionary. However, with this method, a word having a mixture of character types such as “hiragana” and “kanji” such as “ra-ki” as in the above example cannot be detected as an unregistered word. Therefore, there is a problem that it is not possible to register an unregistered word in which such character types are mixed in the dictionary.

An object of the present invention is to provide a text data search device and a text data search method capable of registering an unregistered word having a mixture of character types such as "Hiragana" and "Kanji" in a dictionary. .

[0006]

The present invention described in claim 1 is a morpheme analyzing means for performing a morpheme analysis of text data input using a dictionary file, and a keyword accepting means for accepting an input of a keyword. A text data search device comprising: a search unit that searches the morphologically analyzed text data for a word that matches the keyword accepted by the keyword accepting unit and outputs a search result, wherein the keyword accepting unit accepts the word. Determining means for determining whether or not the keyword is a word registered in the dictionary file, and registering the keyword as a word in the dictionary file when the accepted keyword is a word not registered in the dictionary file And a registration means for performing the registration.

In this structure, the morpheme analysis means performs morpheme analysis of the text data input using the dictionary file. The search means searches the words divided by the morpheme analysis means for a word that matches the keyword accepted by the keyword acceptance means, and outputs the search result. Also,
The determination means determines whether or not the input keyword is a word that is not registered in the dictionary. If the result of this determination is a word that is not registered in the dictionary, the registration means registers this keyword in the dictionary.

Further, the registration means is means for registering a keyword, which has been received a predetermined number of times or more by the keyword receiving means, as a word in a dictionary file.

With this configuration, when the keyword receiving unit receives a keyword that is not registered in the dictionary a predetermined number of times or more, the keyword is registered as a word in the dictionary file.

The searching means further comprises an index creating means for creating an index in which the position information indicating the position of the word in the input text data is registered together with the word based on the result of the morphological analysis. A unit for outputting the search result including the position information is also included.

In this configuration, the index creating means creates an index in which a word and position information indicating the position of the word in the text data are registered. Then, the search means outputs the search result including the position information.

Further, the inventions described in claims 4 to 6 are configurations in which the text data search device described in claims 1 to 3 is regarded as a method invention.

[0013]

1 is a block diagram showing the function of a text data search apparatus according to an embodiment of the present invention. The text data search device 1 includes a text data storage unit 2, a morphological analysis unit 3, a dictionary file 4, an index creation unit 5, an index storage unit 6, a keyword input unit 7, a text data search unit 8, and unregistered. The word registration unit 9 is provided. The text data storage unit 2 stores a plurality of text data. The morpheme analysis unit 3 takes in the text data stored in the text data storage unit 2 and executes a morpheme analysis to divide the text data into words. The dictionary file 4 stores a character string and an attribute such as a part of speech corresponding to each word. The index creating unit 5 creates an index based on the result of morphological analysis. The index storage unit 6 stores the created index. The keyword input unit 7 receives input of a keyword. The text data search unit 8 searches whether or not the accepted keyword is in the index stored in the index storage unit 6. The unregistered word registration unit 9 executes registration of unregistered words in the dictionary file 4. The word part of speech is stored in the dictionary file 4 in order to test the connectivity of the divided words with respect to the result of the morphological analysis. For example, by predetermining a connection relation between a part of speech and a part of speech such that an adjective is not connected after a noun, the result of the morphological analysis of the character string "don't think" is "think (noun) + no ( Not as two verses of "adjective)"
It can be obtained as one phrase of "thought (lower verb) + no (auxiliary verb)".

The operation of the text data search device 1 of this embodiment will be described in detail below. Dictionary file 4 now
2, the word shown in FIG. 2 is registered, and the word “Latch” is an unregistered word that is not registered in the dictionary file 4. As shown in FIG. 2, the dictionary file 4 stores a character string and its part-of-speech corresponding to each word. FIG. 3 is a flowchart showing a process in which the text data search device creates an index of text data.
The morphological analysis unit 3 reads the designated text data stored in the text data storage unit 2 (n1).
The morphological analysis unit 3 executes a morphological analysis process of dividing the text data read in n1 into words using the dictionary file 4 (n2). For example, the text data A of "arrested on suspicion of being killed" is divided into nine words as shown in FIG. 4 by the morphological analysis process of n2.
In this morphological analysis process, the part-of-speech of a word (in this example, "ra" and "ki") not registered in the dictionary file is not registered.

The index creating section 6 detects a word whose part of speech is noun or unregistered in the words divided by the morphological analysis (n3), and creates an index (n4) (see FIG. 5). In the index, text information indicating the text data containing the word is registered in association with the word. Then, the index created by the index creating unit 5 is stored in the index storage unit 6 and the process is completed (n5).

As described above, the text data retrieving apparatus 1 of this embodiment divides text data into words by morphological analysis, and registers the words whose part of speech is not registered in the noun or the dictionary file 4 in the index. To do. Further, a word and text information indicating text data including the word are stored in the index in association with each other.

Next, the process of searching the text data with the input keyword will be described. FIG. 6 is a flowchart showing the search process. Text data retrieval device 1
When a document search keyword is input from the keyword input unit 7 (n11), the text data search unit 8 determines whether the input keyword is registered in the index stored in the index storage unit 6. (N12). When the input keyword is registered in the index storage unit 6, the text data search unit 8 reads out the text information stored in association with the keyword from the index (n13). Then, the text data search unit 8 outputs the read text information (n14), and the processing is completed.

For example, it is assumed that "suspect" is input as a keyword while the index storage unit 6 stores the index shown in FIG. Since the word “suspect” is registered in the index, the text data search unit 8 outputs the text information “text data A” stored in association with the word “suspect” as the search result. From this search result, it is possible to easily read out the text data containing the keyword “suspect”. Further, when the text data is displayed on the display unit as shown in FIG. 7A, if only the keyword portion in the text data A is underlined and is displayed separately from other portions, the display The keyword can be easily found from the text data A displayed. Alternatively, the keyword portion may be highlighted so as to be distinguished from other portions (see FIG. 7B).

If the keyword entered in the index is not registered, the text data retrieval unit 8 inputs this keyword into the unregistered word management unit 9. The unregistered word management unit 9 determines whether this keyword is a word registered in the dictionary file 4 (n15). n
If it is determined in 15 that the input word is a word registered in the dictionary file 4, a message "text data containing the relevant keyword does not exist" is output as a search result (n16), and the process is executed. Complete.

For example, when the words shown in FIG. 2 are registered in the dictionary file 4 and the index shown in FIG. 5 is stored in the index storage unit 6, it is assumed that “captivity” is input as the keyword. Although the word “confinement” is not registered in the index, the word “confinement” is registered in the dictionary file 4, so there is text data that includes “corresponding keyword (“ confinement ”) as a search result. I don't. Message is output.

When it is determined that the keyword input in n15 is an unregistered word not registered in the dictionary file 4, the unregistered word management unit 9 registers this keyword in the dictionary file 4 (n17). Then, as a search result, a message "The keyword is not registered in the dictionary and the keyword cannot be searched. This keyword was registered in the dictionary earlier" is displayed and the processing is completed (n18).

For example, when the words shown in FIG. 2 are registered in the dictionary file 4 and the index shown in FIG. 5 is stored in the index storage unit 6, it is assumed that “Latch” is input as the keyword. . Since the word "Latch" is not registered in the index and dictionary file 4, "Keyword (" Latch ") is not registered in the dictionary as a search result and cannot be searched with this keyword. I registered this keyword in the dictionary earlier. Message is displayed. By this processing, the word "Latch" is newly registered in the dictionary file 4 as shown in FIG. Further, in this embodiment, the part of speech of the word registered in the dictionary file 4 at n17 is a noun.

When the indexing process is executed again on the text data A ("arrested on suspicion of being killed"), it is divided into eight words by morphological analysis (FIG. 9).
(See (A)), and the index shown in FIG. 9 (B) is created. Therefore, it becomes possible to search the text data by inputting "Latch" as the keyword.

Further, when an index creating process is executed for the text data B of the character string "The general public is confined." In the morphological analysis, it is divided into seven words as shown in FIG. 10 (A). (Dictionary file 4 is shown in Figure 8
It is assumed that the words shown in are registered. ), The index is updated and changes as shown in FIG. 9 (B) to FIG. 10 (B). At this time, if the text data is searched by inputting “match” as the keyword, two of the text data A and the text data B can be detected as the text data including the keyword as the search result.

As described above, in the text data search device 1 of this embodiment, if the keyword input during the search is a word that is not registered in the dictionary file 4, the keyword is registered as a word in the dictionary file 4. To do. Therefore, from the next time, it will be possible to search by this registered word. Moreover, since the information indicating the text data including the keyword is output as the text information as the search result, the text data including the keyword can be easily found.

In the above-described embodiment, if the input keyword is a word that is not registered in the dictionary file 4, it is registered in the dictionary file 4. However, registration of a word in the dictionary file 4 will be described below. You may make it perform the process shown. In this example, the unregistered word management unit 9 includes an auxiliary dictionary 9a. As shown in FIG. 11, the auxiliary dictionary 9a stores a keyword and the number of times the keyword is input in association with each other. Text data retrieval device 1
When detecting that a keyword that is not registered in the dictionary file 4 is input from the keyword input unit 7, determines whether this keyword is a keyword registered in the auxiliary dictionary 9a. If this keyword is registered in the auxiliary dictionary 9a, the number of times it is stored correspondingly is increased by one. Also, this keyword is the auxiliary dictionary 9a.
If not registered in, use this keyword in the auxiliary dictionary 9
While registering in a, the number of times corresponding to this registered keyword is set to 1. That is, when a word that is not registered in the dictionary file 4 is input as a keyword, that keyword is registered in the auxiliary dictionary 9a. Also,
The auxiliary dictionary 9a stores the number of input times for each registered keyword. Then, the text data search device 1 is configured such that the auxiliary dictionary 9 a
If there is a keyword that has been input a predetermined number of times or more among the keywords registered in, the keyword is registered in the dictionary file 4. It should be noted that the keywords registered in the auxiliary dictionary 9a are not stored in the auxiliary dictionary 9a when registered in the dictionary.
deleted from a.

That is, in this example, even if a word that is not registered in the dictionary file 4 is input as a keyword,
Not registered in dictionary file 4 immediately. This keyword is registered in the dictionary file 4 when the number of input times exceeds a predetermined number. Therefore, the keywords that have been repeatedly used are registered in the dictionary file 4, so that the keywords that are erroneously input are not registered in the dictionary file 4. Therefore, it is possible to prevent unnecessary words (mis-inputted keywords) from being registered in the dictionary file 4.

[0028]

As described above, according to the present invention, if the entered search keyword is a word that is not registered in the dictionary, the word is mixed with a character type such as "Hiragana" and "Kanji". Even the keywords that are included are automatically registered in the dictionary.

Further, since the number of times of input as a keyword exceeds a predetermined number of times, it is registered in the dictionary, it is possible to prevent an erroneously input keyword from being registered in the dictionary unnecessarily. be able to.

Further, the position information indicating the position where the input keyword is present, which is output as the search result, makes it possible to easily read the text data containing the keyword.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a text data search device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a dictionary.

FIG. 3 is a flowchart showing a process of creating an index of text data.

FIG. 4 is a diagram showing an example of a morpheme analysis result.

FIG. 5 is a diagram showing the contents of an index created.

FIG. 6 is a flowchart showing a search process.

FIG. 7 is a diagram showing a display example of text data.

FIG. 8 is a diagram showing a configuration of a dictionary in which “Latch” is newly registered.

FIG. 9 is a diagram showing a morphological analysis result and updated indexes.

FIG. 10 is a diagram showing a morphological analysis result and updated indexes.

FIG. 11 is a diagram showing a configuration of an auxiliary dictionary.

[Explanation of symbols]

 1-text data search device 2-text data storage unit 3-morpheme analysis unit 4-dictionary 5-index creation unit 6-index storage unit 7-keyword input unit 8-text data search unit 9-unregistered word registration unit 9a- Auxiliary dictionary

Claims (6)

[Claims] 1. A morpheme analysis means for performing a morpheme analysis of text data input using a dictionary file, a keyword acceptance means for accepting an input of a keyword, and a morpheme analysis for a word matching the keyword accepted by the keyword acceptance means. A text data search device comprising a search means for searching the obtained text data and outputting a search result, wherein the keyword accepted by the keyword accepting means is a word registered in the dictionary file. Text data comprising: a determining unit that determines whether the received keyword is a word that is not registered in the dictionary file, and a registration unit that registers the keyword as a word in the dictionary file. Search device. 2. The registering means is means for registering a keyword, which has been received a predetermined number of times or more by the keyword receiving means, as a word in a dictionary file.
The described text data retrieval device. 3. An index creating means for creating an index that registers position information indicating a position where the word exists in the input text data together with the word based on the result of morphological analysis, and the search means 3. The text data search device according to claim 1, further comprising means for outputting the search result including the position information. 4. A text data search method for performing morphological analysis of input text data using a dictionary file, searching for a word matching an input keyword from the morphologically analyzed text data, and outputting a search result. A method of searching text data, characterized in that when the accepted keyword is a word that is not registered in the dictionary file, the keyword is registered as a word in the dictionary file. 5. The text data search method according to claim 4, wherein words that are received as a keyword less than a predetermined number of times are not registered in the dictionary file. 6. An index in which position information indicating a position where the word exists in the input text data is registered together with a word based on a result of morphological analysis, and the position information is also included as a search result. 6. The text data search method according to claim 4, wherein the text data search method outputs the text data.