JPH07121547A - Information retrieval device - Google Patents

Abstract

(57) [Summary] [Purpose] By converting the search conditions into regular expressions,
Make it possible to search accurately even if it contains a typographical error or a line feed code. [Structure] A regular expression conversion unit 12 for converting a regular expression is provided so that even if a typographical error is included in each keyword input by the search condition input unit 11, a regular expression conversion unit 12 is provided so that different notations can be searched. Further, the different notation extension unit 13 performs the extension, and the similar notation extension unit 14 performs the extension so that the similar notation can be searched, and the search condition is converted to improve the search accuracy.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval apparatus used for retrieval processing from an electronic document database.

[0002]

2. Description of the Related Art In recent years, with the spread of word processors and the spread of character recognition devices, the number of electronic documents created by them has increased. Therefore, interest in a document database for accumulating a large amount of document information and retrieving the document information as needed is increasing. In a conventional document database, when searching for a document, a keyword search using a keyword attached to each document is general. However, problems such as keyword addition work not keeping up with the number of stored documents, keywords becoming obsolete over time, and being unable to respond to searches with keywords that exceed the expectations of the database administrator, resulting in frequent omission of searches. was there. From such a background, a document database called a full-text database has recently been attracting attention.
In the full-text database, the search condition provided by the user is compared with all the information of the stored documents, and the document satisfying the search condition is output. However,
When the user searches the full-text database, there is a problem that the character type registered in the full-text database and the type of the character to be searched are different from each other. Therefore, in order to reduce the omission of document search, hiragana /
It was possible to prevent omission of search by converting into katakana, kanji and romaji and using different notation expansion using different notation dictionary.

A conventional information retrieval device will be described below. FIG. 6 shows the configuration of a conventional information retrieval device. In FIG. 6, 1 is a search condition input unit, 2 is a character type conversion unit, 3 is a different notation expansion unit, 4 is a search unit, and 5 is a search result display unit.

The operation of the information retrieving apparatus configured as above will be described below. First, the search condition input unit 1 inputs search conditions. Subsequently, the character type conversion unit 2 converts each keyword of the input search condition into a character type such as hiragana, katakana, kanji, and roman characters. For example, if "search" is entered as the search condition, it is converted into "kensaku", "kensaku", and "kensaku". Next, the different notation expansion unit 3 performs different notation expansion for each word converted by the character type conversion unit 2. For example, "kensaku" is "K
It is expanded to "ENSAKU". Next, the search unit 4 performs a search based on the search condition expanded by the different notation expansion unit 3, and the search result display unit 5 displays the search result.

[0005]

However, in the above-described conventional information retrieval, when a document input by an optical character recognition device (OCR) has a recognition error or a document created by a word processor or the like contains a typographical error. However, there is a problem that the search cannot be performed when the word is divided, and the search cannot be performed even when the word is divided by a line feed code, a page break code, a space code, or a tab code in the document.

The present invention solves the above-mentioned problems of the prior art, and an object of the present invention is to provide an information retrieving apparatus capable of retrieving correctly even a document erroneously recognized by OCR or a document containing a typographical error or a special code. To do.

[0007]

In order to achieve the above object, the present invention provides a search condition input section for inputting a keyword for search, and a regular expression conversion for converting the input keyword into a regular expression. A search unit for performing a search based on a search condition converted into a regular expression, and a search result display unit for displaying a search result.

Further, in order to achieve the above object, the present invention provides a different notation extension part for extending a search condition converted into a regular expression into different notations and / or a similar notation extension part for extending notations having similar notations. Be prepared.

In order to achieve the above object, the present invention is characterized in that a search condition converted into a regular expression is expanded to a different notation by a different notation expansion unit and then expanded to a similar notation at a similar notation expansion unit. It is what

Further, in order to achieve the above object, according to the present invention, the regular expression conversion unit is configured to include a line feed code, a page break code,
It has a skipping conversion unit that converts blank codes and tab codes into regular expressions that are ignored.

Further, in order to achieve the above object, the present invention is such that the regular expression conversion unit is provided with an adjacency matching conversion unit for converting an arbitrary character into a regular expression representing a character number limitation between keywords.

Further, in order to achieve the above object, the present invention is such that the regular expression conversion unit is provided with a variable length character string conversion unit for converting into a regular expression in which characters except for punctuation and punctuation are matched between keywords. Is.

[0013]

Therefore, according to the present invention, when there is a typographical error in the target document, there are three types: one in the regular expression conversion unit, one in which the keyword is incorrect, one in which the keyword is missing, and one in which there is more. A highly accurate search can be performed by creating and searching for the regular expression.

Further, according to the present invention, by expanding the search condition converted into the regular expression by the different notation expansion unit, it is possible to search even if there is a typographical error or omission in a document created by a word processor or the like.

The present invention also extends the OCR by expanding the search condition converted into a regular expression with the similar notation expanding means.
You can also search for characters that were misrecognized by.

The present invention also makes it possible to accurately generate a document containing a linefeed code by creating a regular expression in the skipping conversion unit by inserting a repetition of characters such as a linefeed code to be skipped between each character of a keyword. You can search.

In addition, according to the present invention, when the limitation of the number of characters between keywords is designated, the adjacency matching conversion unit creates a regular expression in which an arbitrary character is connected by the logical sum up to the designated limited number of characters, thereby inputting search conditions. It can be done easily.

In the present invention, the variable-length character string conversion unit creates a regular expression that matches with characters excluding punctuation and punctuation, so that even in the case of arbitrary matching between keywords, the search can be performed without straddling clauses. can do.

[0019]

【Example】

(First Embodiment) A first embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of an information search apparatus according to the first embodiment of the present invention. Figure 1
, 11 is a search condition input unit, 12 is a regular expression conversion unit, 13 is a different notation expansion unit, 14 is a similar display expansion unit, 15
Is a search unit, and 16 is a search result display unit. The different notation extension unit 13 and the similar notation extension unit 14 are connected in parallel between the regular expression conversion unit 12 and the search unit 15.

The operation of the information retrieval apparatus configured as described above will be described. First, the search condition input unit 11
Regular expression converter 1 for each keyword entered in
Convert to regular expression in 2. There are three conversion methods: one character is incorrect, one character is missing, and one character is more, but whether to convert or not can be controlled independently. For example, when the keyword is "abcd", the conversion when one character is incorrect is "bcd | a.cd | a
b. d | abc ”(. represents an arbitrary character, and | represents a logical sum). If one character is missing, the conversion is "bcd
| Acd | abd | abc ". The conversion when there is one more character is “a.bcd | ab.cd | abc.d”. By performing three types of conversion in this way, it is possible to search even if one character is missing.

Next, the different notation expansion unit 13 expands these characters. There are two expansion methods. One is the case where characters are all expanded with the same character type, and the other is not. For example, when the keyword is "abcd", the former is expanded to "ABCD | abcd",
Matches with either "ABCD" or "abcd". In the latter case, "(a | A) (b | B) (c | C) (d | D)"
And so on, and it matches either "ABCD", "ABCd", "ABcd", ... "Abcd".

Next, the similar notation expansion unit 14 converts the character string into a character string that is likely to be erroneously recognized by OCR. For example, "Kou" and "E", "ichi" and "-(minus)" that have similar notations are registered in advance, and either of these letters can be registered. When is displayed, the condition is expanded.

Next, the search unit 15 performs a search based on the search conditions thus expanded, and the search result display unit 16
To display the search results. The search is performed by converting the condition-expanded character string into a state transition table and performing character string matching using the finite state automaton algorithm. Since the character string matching using such a finite state automaton has been introduced in detail in the literature (Tunesuke Takahashi, "Text Search Processor" The Institute of Electronics, Information and Communication Engineers), its explanation is omitted here.

As described above, according to the present embodiment, since the different expression extension unit 13 and the similar notation conversion unit 14 are provided in parallel between the regular expression conversion unit 12 and the search unit 15, they are created by OCR. Even if there is a recognition error in the created document, or if there is a typographical error in the document created by the word processor, it is possible to accurately search. In the present embodiment, the different notation extension 13 and the similar notation extension unit 14 have a simpler configuration.
By omitting and, the input search condition can be simply converted into a regular expression for searching.

(Second Embodiment) Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 2 shows the configuration of the information search device according to the first embodiment of the present invention. In this embodiment, a read-skipping conversion unit 17 is added to the regular expression conversion unit 12 in the first embodiment shown in FIG. 1, and other configurations are the same as those in the first embodiment. The same reference numerals are given to the elements, and duplicated description will be omitted.

With respect to the information retrieval device configured as described above, the operation of the skip-read conversion unit 17 will be mainly described below. The skip-read conversion unit 17 inserts the skipped characters between the characters of the keyword. For example, when searching from "abcd", "a (\ n | \ f | \ s | \ t) *
b (\ n | \ f | \ s | \ t) * c (\ n | \ f | \ s
| \ T) * d ”(\ n is a line feed, \ f is a page break, \
s is a blank, \ t is a tab, and * is a continuation of 0 or more). Such a character string is skipped and inserted as a character string.

As described above, according to the present embodiment, by providing the regular expression converting unit 12 with the read skip converting unit 17 for inserting the read skip character string, in addition to the operation of the first embodiment, Even if a line feed code, a page break code, a space code, or a tab code is included in the middle, it is possible to ignore the search and prevent the search omission.

(Embodiment 3) Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 3 shows the configuration of an information retrieval apparatus according to the third embodiment of the present invention. In this embodiment, an adjacency matching conversion unit 18 is added to the regular expression conversion unit 12 in the first embodiment shown in FIG. 1, and since the other configurations are the same as those in the first embodiment, the same elements are used. Are denoted by the same reference numerals and redundant description will be omitted.

The operation of the adjacency matching conversion unit 18 will be mainly described below for the information retrieval apparatus configured as described above.
The adjacency matching conversion unit 32 creates a regular expression based on the specified maximum number of characters between keywords.
A regular expression is created by arranging arbitrary characters (.) In order from 1 up to the maximum number of characters and connecting them with a vertical bar (|) of logical sums. For example, if the keyword is "information" and "device" and a maximum number of characters is 4 between them, "information (. | ... | ... | ...) device"
To a regular expression. When converted in this way, a character string such as "information retrieval device" or "information input / output device" can be retrieved.

As described above, according to the present embodiment, by providing the adjacency matching conversion unit 18 in the regular expression conversion unit 12, in addition to the operation of the first embodiment, setting between keywords is performed. Since it can be converted into a regular expression using the specified maximum number of characters, it is possible to input search conditions more easily. The regular expression conversion unit 12 in this embodiment may be provided with the read skip conversion unit 17 in the second embodiment.

(Fourth Embodiment) Next, a fourth embodiment of the present invention will be described with reference to the drawings. FIG. 4 shows the structure of an information retrieval apparatus according to the fourth embodiment of the present invention. In this embodiment, a variable length character string conversion unit 19 is added to the regular expression conversion unit 12 in the first embodiment shown in FIG. 1, and since the other configurations are the same as those in the first embodiment, The same elements will be denoted by the same reference symbols and redundant description will be omitted.

The operation of the variable-length character string conversion section 19 of the information search apparatus configured as described above will be mainly described below. The variable-length character string conversion unit 42 creates a regular expression that represents the continuity of the character set between the keywords, excluding the punctuation marks (.) And the punctuation marks (,) from arbitrary characters. For example, if you want to search for “international” or “meeting”, you can search for “international [^ ,.] * meeting” ([] matches one of the characters in the brackets. ^ Is the character in []. Matches all characters except.)).

As described above, according to the present embodiment, by providing the variable length character string conversion unit 19 in the regular expression conversion unit 12, in addition to the operation of the first embodiment, collation of arbitrary characters is performed. By not matching the punctuation marks and the punctuation marks when performing the search, it is possible to omit the matching over the clauses and perform the search more accurately. The regular expression conversion unit 12 in the present embodiment can be provided with the read skip conversion unit 17 in the second embodiment and / or the adjacent matching conversion unit 18 in the third embodiment.

(Fifth Embodiment) Next, a fifth embodiment of the present invention will be described with reference to the drawings. FIG. 5 shows the configuration of the information retrieval apparatus in the fifth embodiment of the present invention. In this embodiment, the different notation extension unit 13 and the similar notation extension unit 14 in the first embodiment are connected in series. Since the other configurations are the same as those in the first embodiment, the same elements are the same. The same reference numerals are given and duplicate explanations are omitted.

The operations of the different notation extension unit 13 and the similar notation extension unit 14 of the information retrieval apparatus configured as described above will be mainly described below. First, the different notation extension 13 is used to extend the input keyword. For example, the word “processing” is expanded to “down”, “temporary structure”, “kawaguchi”, and the like. Next, each of these characters is expanded to a similar notation by the similar notation expansion unit 14. For example, "processing" is "Ka", "falling" is "Shimotaka", "temporary" is "plate", and "Kawaguchi" is "Kuro". To be extended.

In the case of the first embodiment, since the different notation expansion and the similar notation expansion are performed in parallel, the different notation expansion for "processing" is the same, but the similar expansion is "processing".
Will be performed only for the
The vocabulary becomes richer and the search omission becomes smaller accordingly.

As described above, according to the present embodiment, since the search condition is expanded by the different display and then the similar expansion is performed, it is possible to further reduce the omission of the search. In addition,
Also in the present embodiment, the regular expression conversion unit 12 may be provided with the read skip conversion unit 17 and / or the adjacent matching conversion unit 18 and / or the variable length sentence string conversion unit 19.

In each of the above embodiments, the regular expression conversion unit 12 does not particularly limit the number of characters of the keyword when converting, but if the number of characters is 2 or less, the matching rate becomes poor, so the number of characters is limited to 3 or more. Can be provided. Further, the different notation extension unit 13 and the similar notation extension unit 1
4 can be operated independently so that either one or both can be selected.

[0039]

As described above, according to the present invention, when there is a typographical error in the target document, one character is erroneous in the regular expression conversion unit, one character is deficient in the keyword, and one character is missing. It is possible to perform a highly accurate search by creating and searching three types of regular expressions when there are many characters.

Further, according to the present invention, by expanding the search condition converted into a regular expression by the different notation expansion unit, it is possible to search even if there is a typographical error or a missing character in a document created by a word processor or the like.

The present invention also expands the search condition converted into the regular expression by the similar notation expanding means to obtain the OCR.
You can also search for characters that were misrecognized by.

The present invention also makes it possible to accurately generate a document containing a line feed code by creating a regular expression in the skipping conversion unit by inserting a repeating character such as a line feed code to be skipped between each character of a keyword. You can search.

Further, according to the present invention, when the limitation of the number of characters between keywords is designated, the adjacency collation conversion unit creates a regular expression in which an arbitrary character is connected by the logical sum up to the designated limited number of characters, thereby inputting search conditions. It can be done easily.

According to the present invention, the variable-length character string conversion unit creates a regular expression that matches with characters excluding punctuation and punctuation, so that even in the case of arbitrary matching between keywords, searching can be performed without straddling clauses. can do.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing the configuration of an information search device according to a first embodiment of the present invention.

FIG. 2 is a schematic block diagram showing the configuration of an information search device according to a second embodiment of the present invention.

FIG. 3 is a schematic block diagram showing the configuration of an information search device according to a third embodiment of the present invention.

FIG. 4 is a schematic block diagram showing a configuration of an information search device according to a fourth embodiment of the present invention.

FIG. 5 is a schematic block diagram showing the configuration of an information search device according to a fifth embodiment of the present invention.

FIG. 6 is a schematic block diagram showing the configuration of a conventional information search device.

[Explanation of symbols]

 11 search condition input unit 12 regular expression conversion unit 13 different notation expansion unit 14 similar notation expansion unit 15 search unit 16 search result display unit 17 read skip conversion unit 18 adjacency matching conversion unit 19 variable length character string conversion unit

Claims (6)

[Claims] 1. A search condition input unit for inputting a keyword for a search, a regular expression conversion unit for converting the input keyword into a regular expression, and a search condition converted into the regular expression. An information search device comprising a search unit for performing a search and a search result display unit for displaying a search result. 2. The information search device according to claim 1, further comprising: a different notation extension unit that extends the search condition converted into the regular expression to a different notation, and / or a similar notation extension unit that extends the notation having a similar notation. 3. The information search device according to claim 1, wherein the search condition converted into the regular expression is expanded to a different notation by the different notation expansion unit and then expanded to a similar notation at the similar notation expansion unit. 4. The information according to claim 1, wherein the regular expression conversion unit includes a read skip conversion unit that converts a line feed code, a page break code, a space code, and a tab code into a regular expression. Search device. 5. The information search device according to claim 1, wherein the regular expression conversion unit includes an adjacent matching conversion unit that converts an arbitrary character into a regular expression that represents a character number limitation between keywords. 6. The information according to claim 1, wherein the regular expression conversion unit includes a variable length character string conversion unit for converting a regular expression in which characters that match punctuation marks and punctuation marks are matched between keywords. Search device.