JPH02156378A - Word retrieve processing system - Google Patents

Abstract

PURPOSE:To decrease the burden of a user by dividing an inputted character string when the maximum number of stored characters is smaller than the number of the characters in the inputted character string to be a retrieve object, and indicating the retrieve of the maximally corresponding character string as one word. CONSTITUTION:When the character code string of an input character code storing part 1 matches with the word in a dictionary in a character code comparing part 3, the number of the maximally corresponding characters and an address in the dictionary are stored into a maximal length correspondence storing part 4. Based on the number of the characters stored into the storage part 4, a comparison objective character code control part 5 separates the character code string by using a partitioning code, and a comparative collating processing with a character code storage part 2 is executed again. That is, when the word to be the retrieve object of the input character string is not stored into the dictionary in an original form, the word to be the retrieve object, the number of the maximally corresponding characters and the address in the dictionary at such a time point are stored. Based on the information, by applying a control such as division to the word to be the retrieve object, the optimal word information stored into the dictionary can be obtained by binary search for one time.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a word search processing system, and in particular to a word search processing system in which a search target word is controlled and a search result is obtained by only one binary search. It concerns a processing system.

[Conventional technology]

Due to the development of database technology and the increasing complexity of user tasks, there is a need for an information retrieval system that allows not only dedicated operators but also users who are not familiar with computers to search for information stored in databases with simple operations and at high speed. There is. One system that meets this demand is a method of querying a computer in natural language.

Conventionally, the longest-match method and the three-way search method have been the mainstream methods for querying in natural language and searching for words.

The longest match method is to cut out the input character code string to the longest character code number of the word stored in the word storage device, and if there is no matching word as a result of search processing, the input character code string is In this method, the search process is performed again on the input character code string with the last character code deleted.

In addition, the three-way search method first matches the input character code string with the word in the center of the word array, and if there is a mismatch, it is determined which half of the array should be searched, and if there is a mismatch again, then half of the word is searched again. This is a method of searching for a matching word by repeating the operation of matching with the central word in the array.

A word search processing method using the longest match method and the three-way search method is described, for example, in Japanese Patent Laid-Open No. 109088/1988. The method disclosed in the above publication eliminates the need for the repeated process of deleting the last character code and searching again when a matching word cannot be found as a result of searching for the longest character code in the word storage device. . That is, for each word in the word storage device, specify and store the word that matches the first position of the word and has the largest number of character codes within a range that is less than the number of character codes of the word; This word is used as the output of the search results.

[Problem to be solved by the invention]

In this way, conventional word search methods involve repeating the search from the beginning many times by deleting or changing the search target word one character at a time until a matching word is found. , which was affecting the overall system performance. Also, in the method described in the above publication, the user adds specific information to the longest word match, and the target word information is obtained through a single three-way search. The problem was the burden on the user of adding information. This specific information describes the address of the largest word when the beginning of each word in the word storage device matches another word. For example, ``Kuroiro'' stored at a certain address matches ``Kuro'' at another address A by two characters from the beginning, so address A is stored as specific additional information at that address. . However, it is extremely troublesome and time consuming for the user to add such specific information in advance. Also, update the dictionary contents,
When the user adds, deletes, etc., the specific information also needs to be changed, which wastes a lot of time.

The purpose of the present invention is to solve these conventional problems, and to quickly obtain target word information through a single three-way search without adding specific information and storing the word in the character code storage device. The purpose of the present invention is to provide a word search processing system that allows users to search for words.

[Means to solve the problem]

In order to achieve the above object, the word search processing system of the present invention includes means for storing input character strings to be searched, and a character storage dictionary for storing character string words and next search addresses associated with the character string words. , a word search processing system comprising a character comparison means for comparing whether or not the input character string matches a word in the character storage dictionary; When it is determined that this is the case, a maximum match storage means for storing the maximum number of matched characters is activated from the maximum match storage means, and when the maximum number of characters stored is smaller than the number of characters in the input character string to be searched, The present invention is characterized in that it has a comparison target character control means that divides the input character string to be searched and instructs the search to be performed again using the maximum matched character string as one word.

[For production]

In the present invention, (a) when an input character code string and a word in the dictionary match, the longest matching number of characters and the address of the word in the dictionary are stored; and (b) character codes to be compared. The control unit has two new processing functions: controlling the specification of words to be searched. In other words, the search target word in the input string is
If it is not stored in the dictionary in its raw form, the system.

By memorizing the longest number of characters matching the search target word and the address in the dictionary at that time, and adding control such as splitting the search target word based on that information,
It is possible to obtain optimal word information stored in the dictionary with a single three-way search. [Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a word search processing system showing one embodiment of the present invention.

In Fig. 1, 1 is an input character code storage section for setting input character codes (natural language "kana characters" for inquiries), and 2 is a three-way storage section for storing words consisting of character code strings and information accompanying each word. Next search address for search (
a character code storage section in which left and right pointers) are stored, 3
4 is a character code comparison unit that determines whether or not the input character code string matches the word stored in the character code storage unit 2 by comparing the character codes, and 4 is the result of the determination by the character code comparison unit 3. , detecting the number of consecutive matching characters from the first character position of the word in the character code storage unit 2 currently being compared with the input character code string, and comparing this with the maximum number of matching characters stored up to now. When a signal is sent from the longest match storage unit 4, the longest match storage unit 5 stores the larger one, separates the characters using a delimiter code based on the number of characters stored in the storage unit 4, and stores the characters again. A comparison target character code control unit is used to start comparison processing with the code storage unit 2, and 6 is an output area unit that stores the matched character code string in the character code storage unit 2 when it is output. .

The comparison determination by the character code comparison unit 3 is performed by comparing the magnitude of each character code for each character from the beginning, that is, by comparing the characters in binary format. In other words, binary numbers that increase in the order of ``Aiweo'' are assigned as character codes, so A is the smallest, O is the largest in the A line, and then the code becomes larger in the C line, Su line, Yugo, etc. .

FIG. 4 is a diagram showing the stored contents of the character code storage section in FIG. 1.

As shown in FIG. 4, in the character code storage section 2.

There are columns for address, word code reading, word character, left pointer, and right pointer. In addition, in the conventional character code storage unit 2, in addition to the left and right pointer fields, an additional field is provided to store addresses of character codes whose readings partially overlap with the word readings in that field as specific additional information. was.

As is clear from the word table in Figure 4, the words are arranged in the order of 'Aiweo' starting from the smallest address, so as a result of first comparing with the word in the middle,
For input character codes that are smaller than the binary value assigned to that word, the pointer moves to a comparison with the array position before that word, that is, the word that starts with the first character with the lower binary value, and this By repeating this, the probability of a match is increased.

The character code comparison unit 3 determines whether the code of the current character position in the input character code string is the above-mentioned one in the word currently being compared in the character code storage unit 2, in character-by-character determination from the beginning. When it is determined that the address is larger than the code of the character position, one of the two next search addresses predetermined associated with the word in the character code storage section 2 is selected. For example, if the first character of the input character code is larger than the binary value assigned to "ch" as a result of first comparing it with "cha j" at address 8 in the center, then the word at address 12 pointed by the right pointer Let's move on to a comparison. Conversely, if it is determined that the character position code is smaller than the character position code, the other one of the next search addresses is selected and the next word in the character code storage unit 2 to be compared is specified. As a result of the comparison with "cha" in FIG. 4, if the first character of the input character code is larger than the binary value assigned to "chi", the process moves on to comparison with the word at address 4 indicated by the left pointer.

On the other hand, if the sizes of the two codes at the character position are equal, the comparison input character position is advanced by one character and the above process is repeated.

As a result, when the input character code and the consecutive character codes from the beginning to the end of the character code string match, the character code string in the character code storage section 2 is outputted to the output area section 6.

However, if there is no continuous match from the beginning to the end, but only a partial match, it depends on the content stored in the longest match storage section 4. The longest-number storage unit 4 stores the number of consecutive characters that match the input character code string from the first character position of the word in the character code storage unit 2 that is currently being compared as a result of the comparison in the one-character code comparison unit 3. , and compares this with the previous maximum number of matching characters already stored in the longest match record section 4, and if the current number of matching characters is larger, this is stored as the maximum number of matching characters,
At the same time, the address of the current word in the character code storage section 2 is also stored. For example, as a result of comparing the input character code "Kuroirono" with r KuroJ at address 5 in Figure 4, the first 2
Since the characters matched, we stored the number of characters 2 and the matching address 5 in the longest-number storage unit 4, and compared it with the address 6, which was already stored in the longest-number storage unit 4. Since four characters are matched, which is more than the current number of matching characters, the contents of the longest-number storage unit 4 are rewritten to the current matching characters 4 and address 6.

On the other hand, the number of matching characters this time is the longest - number storage unit 4
If the number of characters is smaller than the number of characters already stored in , the signal is transferred to the comparison target character code control section 5. Also,
When the number of matching characters in the current comparison is equal to the number of matching characters already stored in the longest-number storage unit 4, the processing is divided into the following two cases. In other words, (a) there is a next search address attached to the word in the character code storage unit 2 (that is, the left and right pointers are pointing to the next address), and (b) the next search address is If it does not exist (if the left and right pointers do not point to the next address)
It is.

If there is a next search address associated with the word in the character code storage unit 2 that is being compared, one of the two predetermined next search addresses is selected. At this time, the information stored in the longest-number storage section 4 is not changed.

If the next search address does not exist, a signal is transferred to the comparison target character code control unit 5 in the same way as in the case where the number of characters is smaller than the maximum number stored in the maximum minus number storage unit 2.

When the signal is transferred from the longest-number storage unit 4, the comparison target character code control unit 5 converts the input character code string in the input character code storage unit 1 into the maximum number of matching characters stored in the longest-number storage unit 4. Based on the information, the characters are separated using a delimiter code (for example, a space code), and the
The character code string to be determined is recognized as the character code string up to the delimiter code, and the character code is compared in size with the word indicated by the address in the character code storage unit 2 stored in the longest-number storage unit 4. Start the dictionary matching process again.

The method of dictionary matching processing at this time only determines whether or not the word to be compared and the word in the character code storage section 2 match.

If there is a match, the word is output to the output area section 6. If they do not match, the last character of the character string to be compared is separated and moved to the beginning of the character string following the delimiter code.

At this time, the character string with the last character cut off is set as a new comparison target character string, and the character code storage unit has the address in the character code storage unit 2 stored in the longest-number storage unit 4 as pointer information. Dictionary matching is performed by comparing the size with the words in 2.

At this time, it is assumed that the data in the longest-number storage section 4 is initialized.

For example, as a result of comparing the input character code string "Kuropoi" with "Kuro" in the character code storage section 2, the first two characters match, so after storing the number of characters 2 and the address in the longest-number storage section 4, the next When compared with “ri”, the first 1
Although the characters match, since the number is smaller than the number of characters already stored (2), a signal is sent to the comparison target character code control unit 5.

In the character code control unit 5, the number of stored characters “2” is stored.
Separate them into ``Kuro J and r Tsupoi'' and start dictionary matching again with ``Kuro J''.
If you compare it with , it doesn't match, so cut out the last character of Kuro J and move it to the beginning of "Tsuboi" that follows the 1-delimiter code, and make it "Roppoi J." Dictionary matching processing is started using the deleted "ri" as a comparison target character string.

The output area unit 6 outputs words that have been found to match the words in the input character code storage unit 2 continuously from the beginning to the end code or delimiter code by the one-character code comparison unit 3.

FIG. 2 is a diagram of a processing binary tree using the character code storage section of FIG. 4, and FIG. 3 is an operation flowchart showing one embodiment of the present invention.

A specific example will be explained according to FIGS. 2 to 4.

Now, consider "Kurotsupoi J" as an input character code string. First, "Kurotsupoi J" is set in the input character code storage section 1 as a character to be compared (Step 31 in FIG. 3).

As shown in FIG. 2, in this embodiment, it is assumed that address 8 is stored in the initial address register in the character code storage section 2. As a result, the character code comparison unit
is first compared with (step 32). Since the character code is assigned as a binary number that increases in the order of 'Aiweo', the first character code 'ri' in the input character code string is the first character 'r' of the word in the character code storage unit 2 to be compared. It is determined that it is smaller (step 33). Therefore, the left pointer 4 of "cha" in FIG. 4 specifies that the next word in the character code storage section 2 to be compared is "key mouth" (step 34). The longest at this time -
The number storage unit 4 stores '0' as an initial value at this stage. Therefore, since the maximum number of characters does not decrease (step 35), the process returns to the beginning and compares again with the words in the dictionary (step 32). Namely.

In the comparison of the first character of "Kuropoi J" and "Kiikuchi", it is determined that the input character code string is larger than the word in the character code storage unit 2 to which it is referred (step 33), and the right The pointer 6 specifies that the word in the character code storage unit 2 to be compared next is "kuroiro" (
Step 34).

In this case, the first character "rij" and the second character "1 kuchi" match. In the comparison of the third character, the input character code string is "tsu" and the word in the character code storage unit 2 to be referenced is "i", so the input character code string is determined to be thicker. Ru. Therefore, the right pointer of TI Croiro'
7' specifies that the word in the character code storage section 2 to be compared next is "r Shiro" (step 34). In addition, the maximum length matched this time is stored in the longest-number storage unit 4.
2' and the corresponding word "
Address 6 of Kuriro 1 is stored (step 35)
. Next, input character code string "Kuropoi" and address 7
and "Shiro" (step 32). In this case, in the comparison of the first character, it is determined that the input character code string is smaller than the word in the character code storage unit 2 that is referred to, and the character code storage unit 2 that the left pointer of "SHIRO" points to (step 34). However, since the maximum number of matched characters '2' is stored in the longest-number storage unit 4, which is larger than the maximum number of characters matched this time '0', the comparison target character code is stored in the longest-number storage unit 4. The activation signal is transferred to the control unit 5 (step 35
). This activation signal makes it clear that there is no word matching "Kuropoi" in the character code storage section 2. Also, based on the maximum number of matched characters '2' stored in the longest-number storage unit 4 at that time, the comparison target character code control unit 5 converts the first two characters 'Kurotsuboi' in the input character code storage unit 1 to 'Kurotsuboi'. " and the remaining character string "tsupoi" (step 36). Then, based on the information stored in the longest-number storage section 4, it is determined whether or not a word matching the character code string "Kuro" exists in the character code storage section 2 (step 31). This determination process is restarted from the address 6 of "Kuroiro" stored in the longest-number storage unit 4. First, the character code comparison section 3 compares "Kuro ■" with "Kuroiro" at address 6 in the character code storage section 2 (step 32). In this case, the first character and second character match, but when comparing the third character, the input character code string is a delimiter code (in this example, it is a blank code), and the reference Since the character in the character code storage unit 2 is "i", it is determined that the input character code string is smaller. That is, here, the space increases in the order of ``A'', but the space is smaller than the first letter ``A''. Therefore, the left pointer 5 of "Kuroiro" specifies that the next word in the character code storage section 2 to be compared is "r Kuro" (step 34). Next, the input character code string "kuro" is compared with "r kuro" at address 5 (step 32).

At this time, since the input character code string and the word in the character code storage section 2 match (step 33), the information of the word "r" in the 5-character code storage section 2 is output to the output area section 6 (
Step 37).

The word "r" in the character code storage unit 2 at this address 5.
is the longest consecutive word that matches from the beginning of the input character code string.

In this way, in order for a computer to understand the meaning of a word contained in a makeshift sentence, it needs to be checked against a dictionary.
The speed of dictionary matching processing is an important factor in overall system performance. In this embodiment, the longest-number storage unit 4
By newly providing the comparison target character code control section 5, it is possible to detect words in the dictionary that are equivalent to or coincide with the search target word in a single search, and the performance of word search can be improved. . Also, in this example.

In addition to improving the performance of a system that performs word search processing, it is possible to omit the amount of work required to construct a dictionary and change the words in the dictionary, which occur with improved performance. In other words, in the conventional character code storage unit, it was necessary to add specific information as table information to store words, but in the present invention, this can be omitted, so it is possible to reduce the user's workload. . In particular, it is possible to omit the user's workload associated with changing the specific information when updating, adding, or deleting words in the character code storage unit.

〔Effect of the invention〕

As explained above, according to the present invention, a character code string in the character code storage unit that matches the input character code string for the longest time is searched from the beginning of the input character code string by performing only one three-way search when searching for a word. It is possible to specify quickly.

It also has the advantage of not increasing the user's load when creating or updating a dictionary.

[Brief explanation of the drawing]

Fig. 1 is a functional block diagram of a word search processing system showing an embodiment of the present invention, Fig. 2 is a diagram showing a binary tree showing the processing operation of Fig. 1, and Fig. 3 is an embodiment of the present invention. FIG. 4 is a flowchart showing the word search process, and FIG. 4 is a diagram showing the stored contents of the character code storage section in FIG. Input character code storage section, 2: Character code storage section, 3:
Character code comparison section, 4: longest-number storage section, 5: comparison target character code control section, 6: output area section. Figure □: The flow of Deday!

Claims (1)

[Claims] 1. means for storing an input character string to be searched; a character memory dictionary for storing a character string word and a next search address associated with the character string word; and a means for storing an input character string to be searched; In a word search processing system comprising a character comparison means for comparing whether or not they match, when the character comparison means determines that there are consecutive matches from the beginning of a word, a maximum number of characters that are matched is stored. A match storage means and a maximum match storage means are activated to divide the input string to be searched when the maximum number of stored characters is smaller than the number of characters of the input string to be searched, and to divide the input string to be searched for, and to divide the input character string to be searched to obtain the maximum matched characters. A search word processing system comprising: comparison target character control means for instructing a search again using a string as one word.