JPH0231273A - Word dictionary retrieving device - Google Patents

Abstract

PURPOSE:To retrieve all words coincident with partial character strings of sentences out of a word dictionary by extracting successively the unit character strings of the prescribed length out of the sentences and collating to the partial word dictionaries divided into plural parts. CONSTITUTION:A unit character string extracting means 3 extracts a unit character string of the same length as the maximum expression length of a work dictionary 1 out of the sentence received from a sentence input means 2. The dictionary 1 is divided into plural partial word dictionaries for each word having the same head character. Then the unit character string extracted out of the sentence is stored into a character string retrieving means 4 which stores plural character strings and can detect all partial character strings coincident with plural stored character strings out of those character strings to be inputted. Then each of those words stored in the partial word dictionaries is inputted. Thus it is possible to extract all partial character strings coincident with each word of the word dictionaries out of the sentences.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a word dictionary search device that is used in a text processing device such as a word processor and searches a word dictionary for all words that match a partial character string of a sentence. .

(Prior Art) In recent years, text processing devices such as word processors have become rapidly popular. One of the most frequently performed processes in such text processing devices is character string search processing.For example, searching for a certain character string to find a desired position in a sentence, or searching for a certain character in a sentence. String search processing is performed when replacing all strings with other strings (this is referred to as the first type of string search processing). Furthermore, it also inspects sentences (detecting mistakes and inappropriate parts in sentences).

As the first step in performing more advanced text processing such as automatic extraction of keywords, conversion of text into audio, and automatic translation into other languages, the text is compared with a word dictionary, and partial characters of the text are extracted from the word dictionary. Single character string search processing is also performed when searching for all words that match a string (this is referred to as the second type of character string search processing). Comparing sentences with a word dictionary is
This is a character string search process in which the character strings of multiple words registered in the word dictionary are compared with the character strings of the text. The first type of character string search and the second type of character string search processing are the same in that they search for a searched character string from one sentence, but the 1st type! The number of character strings to be searched in the 1/i character string search is usually one, or at most several, and the number of character strings to be searched in the 21st S1 character string search process is as follows.

The major difference is that the number is usually huge, ranging from tens of thousands to tens of words (=number of words stored in a word dictionary).

Therefore, conventionally, a registration mode and a search mode are provided exclusively for character string search processing, and in the registration mode, multiple input character strings are stored internally, and in the search mode,
A character string search means is realized that detects all partial character strings that match a plurality of character strings stored in registration mode from an input character string (which allows wildcard characters to be included). Such a string search method is described in the following document (
As shown in 1), it is implemented as an LSI. In addition,
A wildcard character is a special character predetermined to match any character.

r String Search LSIJ (Yamada, Hirata, Mizuben, Takahashi, Institute of Electronics, Information and Communication Engineers Technical Research Report CAS87-25198
(May 29, 2007) Literature (1) The above-mentioned character string search means can be easily applied to @1 type of character string search processing. That is, by first registering a character string to be searched in the character string search means in the registration mode and then inputting a sentence into the character string search means in the search mode, it is possible to search for a desired character string from a single sentence.

(Practice to be Solved by the Invention) However, there is a problem in applying the above-described character string detection means to the second method of character string search processing. This is because, as described above, the number of character strings to be searched is usually a huge number ranging from several blades to several tens of blades. Therefore, if you try to realize the second type of character string search process in the same manner as the first character string search process (register 6 units of word dictionary in registration mode and input sentences in search mode) method), the string search method can register a large number of strings, or
Alternatively, it is necessary to prepare a large number of single character string search means. For example, the character string search means realized as an LSI in the above-mentioned document (11).

The number of character strings that can be registered is 64, and 1,000 character strings are required to register a word dictionary of 64,000 words.

Therefore, a method may be considered in which sentences are registered in the character string search means instead of a word dictionary. That is, first, register a sentence in the character string search means in registration mode.

This method involves entering a word dictionary in search mode. The length of a sentence, which is the basic unit of writing, is generally several dozen characters, so... Compared to the method of registering a word dictionary, the method of 1.: requires fewer character string search means. less. However, it is necessary to register in the character string search means all possible patterns formed by changing the starting character position and character string length for a sentence. For example, "Learn Japanese"
For the 8-character sentence, it is necessary to register the following variations (36 variations). Since there are 36 ways to search for a sentence of about 8 characters, there are hundreds to thousands of ways to search for a sentence of several tens of characters, which again requires a large number of character string search means.

Learn JapaneseLearn JapaneseLearn JapaneseLearn JapaneseLearn JapaneseJapaneseJapaneseLearn VocabularyLearn VocabularyLearn VocabularyBookIF#Learn LanguageLearn VocabularyAlso . Although there is a method to implement the second type of string search processing without using the string search means described here, it is better to use the string search means for the second type of string search processing as well. This is desirable as a text processing device in that the first type of string search processing and the second type of string search processing can share a single string search means (as a text processing device, an efficient configuration can be achieved). ).

An object of the present invention is to provide a word dictionary search device using the character string search means as described above.

It is an object of the present invention to provide a word dictionary search device that can be configured with a smaller number of character string search means than conventional ones.

(Means for Solving the Problems) The present invention is a word dictionary search device that searches a word dictionary for all words that match a partial character string of a sentence, and which divides the word dictionary into a plurality of partial word dictionaries. , the partial word ′W
The calligraphy has a form in which words are notated in a solid manner with de + J mita in between, and a unit character string extraction means for extracting a unit character string of the same length as the maximum notation length in the word dictionary from all character positions of the sentence. , has a registration mode and a search mode, and in the registration mode stores a plurality of input character strings internally, and in the search mode, the input character strings including wildcard characters are selected from the input character strings stored in the registration mode. a character string search means for detecting all partial character strings that match a plurality of character strings; and a registration mode set for the character string search means to retrieve the plurality of unit character strings extracted by the unit character string extraction means. Selecting a partial word dictionary to be searched from among the plurality of partial word dictionaries based on the unit character string registration means input to the character string search means and the unit character string extracted by the unit character string extraction means. and a wild card adding means that adds wild card characters to the end of a given character string until it has the same length K as the unit character string.

A search mode is set for the character string search means, and the wild card adding means inputs the notation of each word in the partial word dictionary selected by the dictionary selection means to the character string search means with a wild card character added thereto. The invention is characterized by comprising a word dictionary input means.

(effect) The present invention will be described in detail.

The unit character string extraction means in the present invention extracts n types of unit character strings from a sentence with a length of n characters. The length U of the unit character string at this time is the same as the maximum notation length in the word dictionary (there is no point in matching with a partial character string of a sentence that is longer than the maximum notation in the word dictionary). In the conventional method of registering a sentence such as the one shown in the section of the problem to be solved by the invention in a character string search means, n unit character strings with different starting positions extracted from a sentence of length n are Furthermore, since the number of partial character strings is expanded to different lengths, the total number of one partial character string (the number of character strings registered in the character string search means) is linear.

When n≦U n(n+1) When n)u (nu-u)u+↓u(u+1)=”
u (2n-u+1) For example, if u=8, the value of the above equation is 36. 100 when n=16, 228 when n=32
It is. In the present invention, the number of unit character strings registered in the character string search means is 8 when n=8. 16 when n=16. n=3
2, the number is 32, so compared to the conventional method, the number of string search means is much smaller (therefore, the number of string search means is also smaller).

1. In the present invention, the unit character string is divided into parts of different lengths
Instead of adding text to a character string, a wildcard addition means adds a wildcard character to a word in a word dictionary, and the character string search means uses a function to match the wildcard characters.

We are trying to detect multiple words of different lengths from a Hijii Bunko string. FIG. 2 shows an example of the contents of the word dictionary 1. In FIG. 2, each word is filled with delimiters r/J*+-,z. For example, "Learn all Japanese"
From the beginning of the sentence, "Japanese", "Japan", "Japan" and (1
In the case of extracting five word groups, the conventional method would involve comparing the following eight partial character strings with the word dictionary 41 as shown in Figure 2 of Maki.

Learn JapaneseLearn JapaneseJapan rf! However, in the present invention, a wildcard character up to a length U is added to each expression in the word dictionary, and then matching is performed. FIG. 3 is a diagram showing a form in which a wild card character is added to each notation in the word dictionary 1 of FIG. 2 (in the case of n=8, the wild card character is represented by "$"). Therefore, in the present invention, it is sufficient to match one unit character string "Learn Japanese" with the character string shown in FIG.

Furthermore, when inputting a word dictionary into the character string search means in the search mode, the large capacity of the word dictionary poses a problem. Therefore, in the present invention, a dictionary selection means is provided so that instead of always inputting the entire word dictionary, only the part selected by the dictionary selection means is inputted. As a result, the time required for searching can be significantly reduced compared to inputting the entire word dictionary.

At this time, the portion of the word dictionary to be searched is selected based on the unit character string registered in the character string search means. For example, if a word dictionary is divided into partial word dictionaries by the first character of each word, only the partial word dictionary corresponding to the first character of each unit string registered in the string search method should be selected as the search target. Bye. String search methods include "Learn Japanese Literature,""Learn Japanese Literature," and "Learn Literature."
If "Learn about", "Learn", "Learn", "Study", "Suru", and "Ru" are registered, instead of inputting all the words in the dictionary to the string search method, Input only partial word dictionaries such as

A partial word dictionary containing words whose first letter is ``日'' ``A partial word dictionary containing words whose first letter is ``hon'' ``A partial word dictionary containing words whose first letter is ``bun''``Gaku'' as the first letter A partial word dictionary that contains words whose first character is ``wo'' A partial word dictionary that contains words whose first character is ``shu'' A partial word dictionary that contains words whose first character is ``su'' DESCRIPTION OF THE PREFERRED EMBODIMENTS Partial Word Dictionary (Example) Partial Word Dictionary Containing Words Starting with "ru" (Example) The present invention will be described in more detail with reference to the drawings.

FIG. 1 is a block diagram showing the structure of the male side of the word dictionary search device of the present invention.

In FIG. 1, a word dictionary 1 is a dictionary in which words are registered. Magnetic disk device, magnetic tape device, IC memo 1
This can be realized using an optical disk device or the like. The word dictionary 1 is composed of a plurality of partial word dictionaries.

As a method of dividing the word dictionary l into partial word dictionaries, for example, there is a method of dividing the word dictionary by the first character of each word. FIG. 6 shows the word dictionary 1 divided into N partial word dictionaries 1-1.1-2. for each first character of a word. This is a diagram showing a configuration divided into 1-N (in FIG. 6, L, M, and N are integers, and 1<L<M<N). Furthermore, the word dictionary 1 according to the present invention has a form in which the notation of each word is filled in with delimiters in between. FIG. 2 is a diagram showing an example of the contents of the word dictionary 1 for the 1-part ratio word dictionary 1-M.

In FIG. 2, the delimiter is represented by "/".

The text input means 2 is a means for inputting text.

It may take a form in which sentences are input while being created on the spot, or it may take a form in which sentences created in advance on another device are read all at once. The text input means 2 can be realized by a keyboard, a character recognition device, a voice recognition device, a kana-kanji conversion device, a magnetic disk reading device, a magnetic tape reading device, etc. The text inputted from the text input means 2 is sent to the unit character string extraction means 3.

The unit character string extraction means 3 is a means for extracting a unit character string having the same length as the maximum notation length U in the word dictionary 1 from all character positions of a sentence inputted from the sentence input means 2. For example, u = In the case of 8 (below).

In the operation example, the explanation will be made assuming u = 8), "
FIG. 1 is a block diagram showing the configuration of an embodiment of the word dictionary search device of the present invention. ”, the following unit character strings are extracted. The unit character string extraction means 3 sends the extracted unit character strings to the unit character string registration means 5 and the dictionary selection means 9.

Figure 1 shows the present invention. Figure 1 shows the unit of the present invention. The figure is the word of the invention is the vocabulary of the invention Unit salary dictionary of the present invention This is a block diagram. is a block diagram It is a lock diagram.

This is a diagram.

This is a diagram.

It is a diagram.

It is.

be.

Ru.

The character string search means 4 has a registration mode and a search mode. In the registration mode, it internally stores a plurality of input character strings, and in the search mode, it stores input character strings (including wildcard characters). This means detects all substrings that match a plurality of strings stored in registration mode. The character string search means 4 is a known means, and can be realized, for example, as described in the above-mentioned document (1). String search means 4 based on document (1)
When this is realized, switching between registration mode and search mode is performed by receiving and receiving signals from the outside. In the registration mode, first, the registration address (the location stored inside the character string search means 4) is set from the outside, and then, if the character string to be registered is input from the outside, 1
The character string search means 4 internally stores one character string. When this is repeated, multiple character strings are searched by the character string search means 4.
will be stored internally. In the search mode, the character string search means 4 searches one character string each time a character string that matches a character string stored in the registration mode is input.

Outputs the registered address of the matched string. Also.

At this time, the matching position (-matching address) in the input character string is obtained by counting how far the character string has been input into the single character string search means 4 when the registered address is output.

At the same time as the registered address is output, the single character string search means 4 outputs it.

The unit character string registration means 5 is means for setting a registration mode for the single character string search means 4 and inputting a plurality of unit character strings sent from the unit character string extraction means 3 to the character string search means 4. be. FIG. 4 is a flowchart showing the processing procedure of the unit character string registration means 5 when the character string search means 4 is implemented as in document (1). The unit character string registration means 5 first sends a signal to set the registration mode to the character string search means 4 (the signal causes the character string search means 4 to enter the registration mode). Then, the unit character string registration means 5
The unit character strings are retrieved one by one from the unit character string extraction means 3 and the retrieved unit character strings are registered in the character string search means 4, and this process is repeated. The registration of a unit character string in the character string search means 4 is a process of setting a registration address and then inputting the unit character string to be registered into the character string search means 4. When there are no more unit character strings sent from the unit character string extraction means 3, the unit character string registration means 5 sends a signal to the character string search means 4 to cancel the registration mode, and the process j! (However, the signal for canceling the registration mode may be omitted and replaced by the search mode setting signal issued by the word dictionary input means 6).

The dictionary selection means 9 is means for selecting a partial word dictionary to be searched from among a plurality of partial word dictionaries based on the unit character string extracted by the unit character string extraction means 3. The method for selecting a partial word dictionary depends on the method of dividing the word dictionary l into partial word dictionaries, but in the case where the method of dividing the word dictionary l into partial word dictionaries for each first character of a word is adopted, as shown in FIG.

The partial word dictionary corresponding to the first character of each unit character string extracted by the unit character string extraction means 3 may be selected.

That is, a dictionary selection table 90 as shown in FIG. 7 that associates the first character with the dictionary number of the partial word dictionary is prepared, and the part to be searched is retrieved by using the first character of the unit character string as a key. All you need to do is get the dictionary number of the word dictionary. The dictionary selection means 9 sends the dictionary number of the partial word dictionary to be searched to the word dictionary input means 6.

The wild card addition means lO is means for adding wild card characters to the end of a given character string until the length U is the same as the unit character string (the same length as the maximum notation length in the word dictionary l). It is.

The system operates in such a manner that it receives the notation (variable length) of a word from the word dictionary input means 6 and returns the result of adding a wild character to the word dictionary input means 6. For example, the word dictionary input means 6 reads "Japan". If the notation ``Japanese'' is removed, Uke returns ``Japanese $$$$$$$J'' to the word dictionary input means 6;
Return J.

A method of forming a character string in which wildcard characters are added up to a length U is, for example, to prepare a buffer in which U wildcard characters are stored in advance, and to input the received notation from the word dictionary input means 6. , only the character string length valve can be set in the buffer from the beginning and overridden and written), and the resulting X character string in Batza can be extracted.

The word dictionary input means 6 sets a search mode for the character string search means 4, and the wild card addition means 10 sets the search mode to the character string search means 4.
This means adds a wild card character to the notation of each word in the partial word dictionary selected by the dictionary selection means 9 and inputs it to the character string search means 4. The processing procedure is almost the same as that of the unit character string registration means 5. First, a signal is sent to the single character string search means 4 to set the search mode, and then the word dictionary! F1
are input into the character string search means 4, and the partial word dictionaries of the dictionary numbers instructed by the dictionary selection means 9 are sequentially input), and when the input is completed, a signal is sent to the character string search means 4 to cancel the search mode. However, when inputting the partial word dictionary to the character string search means 4.

Each word in the partial word dictionary is passed to the wild card adding means 10, and the character string returned from the wild card adding means 10 is input to the character string searching means 4. Therefore, the word dictionary input means 6 has one, for example, a second
For the partial word dictionary 1-M as shown in the figure, read it, change it to a character string with wild card characters as shown in Figure 3 using the wild card addition means 1o, and input it to the character string search means 4. do. Note that the word dictionary input means 6 is
It needs to be activated after the unit character string registration means 5 completes the registration process of the unit character string to the character string search means 4.

The search result storage means 7 is a means for storing the search results by the five character string search means 4. It can be realized by a magnetic disk device, a magnetic tape device, an IC memory, etc. Search results are what kind of words appear in which part of a sentence (where the word in the word dictionary matches the sentence)
This is the information. Information about which part of the text is
When a match occurs in the character string search means 4, the two character string search means 4 output (of the matched unit character string)
Corresponds to the registered address. Information about what kind of word it is is 1 when a match occurs in the character string search means 4.
This corresponds to the matching address of the ^t@ dictionary 1 output from the character string search means 4. Therefore, the search result storage means 7
Registered address and word dictionary of unit string when a match occurs 4! Store the ships with matching addresses of l. Since these two addresses are information output from the character string search means 4 when a match occurs, they can be stored as they are.

The control means 8 includes the text input means 2. Unit character string extraction means 3
1 Character string search means 4. Unit character string registration means 5, dictionary selection means 9. A means for controlling the operation of the word dictionary input means 6 and causing the processing as shown in FIG.
U etc. are used.

The control means 8 first activates the text input means 2 to input a sentence (STEPI), and then the control means 8 activates the unit character string extraction means 3 to extract a unit character string from one sentence. (STEP 2). continue.

The control means 8 activates the unit character string registration means 5.

Register the unit character string in the character string search means 4 (STEP 3)
). At this time, the single character string search means 4 is set to the registration mode, and stores the unit character string input from the unit character string registration means 5 internally. Further, the control means 8 controls the command 41! The selection means 9 is activated to select a partial word dictionary to be searched (STEP 4). Subsequently, the control means 8 activates the word dictionary input means 6, adds a wildcard character to the word dictionary 41, and inputs it to the character string search means 4 (STEP 5).
). At this time, the character string search means 4 is set to search mode.

The result of searching the word dictionary 1 for character strings containing wildcard characters against the internally stored unit character strings is output to the search result storage means 7. In this embodiment, the wild card addition means 10 is directly controlled by the word dictionary input means 6 (a configuration in which the control means 8 controls the wild card addition means 10 is also possible).

In the embodiments described above, a method of dividing the word dictionary 1 into partial word dictionaries is shown by dividing each word by the first character, but the present invention does not limit the method of division to this. For the first character of a word, instead of forming one partial word dictionary for each character, a plurality of first letters may be combined into one partial word dictionary. Moreover, as a rougher division method, it is sufficient to simply divide the characters depending on whether they are kanji or kana. As described above, the method of selecting a partial word dictionary to be searched by the dictionary selection means 9 is based on the method of dividing the word dictionary 1 into partial word dictionaries.

In addition, if one character string search means 4 cannot register all the character strings, you can repeat the registration and search for the number of characters that can be registered, or you can use multiple character string search means 4. good. When using a plurality of character string search means 4, in the registration mode, the control means 8 monitors the number of character strings registered in the character string search means 4, and determines the number of character strings that can be registered in one character string search means 4. When this is reached, a command is sent to the unit character string registration means 5 to switch the character string search means 4 to be registered. In the search mode, the word dictionary input means 6 inputs the word dictionary 1 into the plurality of character string search means 4 in parallel, and writes all the search results output from the one character string search means 4 to the search result storage means 7. Make it.

In the embodiment of the first type of character string search processing described in the prior art section, a sentence is input into a character string search means in a search mode. At this time, the comparison is performed from arbitrary character positions in the two sentences. However, when inputting a word dictionary to the character string search means in the search mode as in the present invention, it is not necessary to perform matching from any position of the character string in the word dictionary. It is sufficient to start matching from the first position of the notation of each word. There are two methods to limit such matching start positions: Since the expressions in the word dictionary are separated by de'J mitter, the first method is to control the string search means to perform matching immediately after the delimiter (this is done by This is a matching method generally called one-anchor matching). In the character string search means shown in document (1), K is set so that such an anchor match can be realized. The second method is
In the anchor match kumquat, deIJi is registered at the beginning of the unit character string to be registered. As a result, even if the character string search means does not have an anchor matching function, matching can be performed from the delimiter position in the word dictionary.

Furthermore, if the word dictionary search device of the present invention is incorporated into a text creation device that performs the 181st string search process using the string search means described above, the word dictionary search device of the present invention may be It can be shared by the filE1-dan character string search process and the second type of character string search process.

(Effects of the Invention) As explained above, according to the present invention, there is provided a word dictionary search device using a character string search means.

Moreover, a word dictionary search device can be obtained that can be configured with a smaller number of character string search means than conventional ones.

Since this is a word dictionary search device using one character string search means, by combining it with a sentence creation device that performs the first type character string search process using one character string search means, two character string search means can be used as the first type. It is possible to realize a text creation device with an efficient configuration that is shared by the character string search process of 1 and the character string search process of the 2nd Wi.

By incorporating the word dictionary search device, the text creation device can use not only character string information but also word information for sentences.
It will be possible to perform more advanced text processing, including text inspection and text-to-speech conversion.

Furthermore, since the partial word dictionary is selected according to the unit character strings extracted from the two sentences, it is possible to limit the range of word dictionaries input to the character string search means in the search mode. Therefore, faster searches can be performed compared to inputting all the words in a dictionary.

In the word dictionary examination device of the present invention, the words registered in each partial word dictionary do not necessarily have to be arranged in the order of their written character codes. Conventionally, with regard to word dictionaries, it has been troublesome to manage the order of words registered in the dictionary, adding a large burden to the work of adding and deleting words. Although it is necessary to divide the word dictionary used in the present invention into partial word dictionaries, it is possible to eliminate the trouble of managing the arrangement of words within a single partial word dictionary.

Furthermore, since the word dictionary can be packed with one notation, its capacity can be kept small compared to a case where the notation area is set to a fixed length.

It is a figure showing an example of contents.

l...Word dictionary, 2...Sentence input means, 3...Unit character string extraction means, 4...
Character string search means, 5...Unit character string registration means,
6... Single candy dictionary input means, 7... Search result storage means, 8... Control means.

9...Dictionary selection means blO...River/Wild card addition means, 9o/River...Dictionary selection table.

Agent Patent Attorney Shinsuke Honsho

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, and FIG. 2 is a diagram showing an example of the contents of a word dictionary ja1. FIG. 3 is a diagram showing the result of adding wildcard characters to the notation of each word in the word dictionary 1, FIG. 4 is a 70-chart showing the processing procedure of the unit character string registration means 5, and FIG. 5 is an example. FIG. 6 is a diagram showing an example of a method of dividing a word dictionary into partial word dictionaries. FIG. 7 is a flowchart showing the processing procedure of the dictionary selection table 90. E3gzssssss e shout $$$$$$$ a death $$$$$$$ a book $$$$$$$ a$-invitation $$$$$$ a-nin $$$$$$ Sunday $$$$$$ a Day of the week $$$$$ Figure 3 Figure 5 (a) Figure 5 (b) Figure

Claims (1)

[Claims] In a word dictionary search device that searches a word dictionary for all words that match a partial character string of a sentence, the word dictionary is divided into a plurality of partial word dictionaries, and the partial word dictionaries are divided into a plurality of partial word dictionaries in which word representations are separated by delimiters. unit character string extracting means for extracting a unit character string in a solid form and having the same length as the maximum notation length in the word dictionary from all character positions of the sentence; and a registration mode and a search mode. stores the input strings internally, and in the search mode searches all substrings that match the strings stored in the registration mode from the input strings containing wildcard characters. a character string search means for detecting; and a unit character string registration means for setting a registration mode for the character string search means and inputting a plurality of unit character strings extracted by the unit character string extraction means to the character string search means. and,
dictionary selection means for selecting a partial word dictionary to be searched from among the plurality of partial word dictionaries based on the unit character string extracted by the unit character string extraction means; a wildcard addition means for adding wildcard characters until the length is the same as the unit string; and a search mode is set for the character string search means, and the wildcard addition means sets a search mode for the character string selected by the dictionary selection means. A word dictionary search device comprising: word dictionary input means for inputting the notation of each word in a partial word dictionary to the character string search means by adding a wild card character.