JPH07319895A - Device and method for retrieving document - Google Patents

Abstract

PURPOSE:To retrieve a target document at high speed even if a retrieval key is the synthesis word of a word and a character or a sentence and even if the quantity of the document to be retrieved is large. CONSTITUTION:When the keyword of the document is inputted from an input part 220, a keyword extraction part 224 dissolves the keyword into the lines of plural keyword ID. A keyword matching part 226 searches document ID including all obtained keyword ID by searching an index in an index storage buffer. Then, an adjacent information matching part 227 refers to the adjacent information index of an adjacent information storage buffer 273, retrieves document ID having ID having the same line as that of keyword ID in the keyword inputted among adjacent information of searched document ID and it is set to be a retrieval answer. Thus, retrieval speed can considerably be improved since all the documents are not necessary to be retrieved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval apparatus for retrieving a document containing a retrieval keyword, and more particularly to a document retrieval method when two or more retrieval keywords are input.

[0002]

2. Description of the Related Art Some conventional database search devices employ a full-text search method that can search for a target document by using an arbitrary character string in a document to be searched. Such a search device must create an index in preprocessing in order to search a large amount of documents at high speed. This index expresses a document including all the words and characters extracted from all the search target documents. When the user uses a word or character in the index created in the preprocessing as a search key, the target document can be searched at high speed by referring to this index. However, the user may input the search key with a word or character composite word or a sentence. In such a case, the database search device divides the input search key into words or characters, and Documents containing words and characters are searched by referring to the index. Further, there is an apparatus that determines whether or not the same word or character as the search key appears in the original document by character string matching, and finally searches only the document including the same word or character.

As described above, according to the conventional method, when the search key input by the user is a word or character composite word or a sentence, it is compared by character string matching whether or not the search key appears in the original document. By contrast, they were searching for the desired document. In this case, the larger the document to be searched is, the more the device licks the document, which takes time, and the search speed is significantly reduced.

[0004]

In the conventional database search apparatus employing the full-text search method, when the search key input by the user is a word or character composite word or a sentence, the search key is included in the original document. The target document was searched by comparing and contrasting whether or not it appeared in the text by using character string matching or the like. In this case, the larger the document to be searched is, the more the device licks the document, which takes time, and the search speed is significantly reduced.

Therefore, the present invention eliminates the above-mentioned drawbacks, and the target document can be searched at high speed even if the search key is a compound word or sentence of words and characters and the number of documents to be searched is large. An object is to provide a document search device and a document search method.

[0006]

The present invention has, for each search key, index information in which the identification number of the document to be searched including this search key is associated, and two or more search keywords are input. And a document retrieval device having a function of decomposing the search keyword into word units, replacing each word unit with an identification number, and then retrieving a document including all the identification numbers constituting the search keyword from the index information. In the above, in the search target document, the entire content of the document is decomposed into word units, and the word units are replaced with identification numbers to create adjacent information arranged in word order, and the index information is used to search the adjacent information. Search unit that searches the adjacent information corresponding to the input document for the same sequence as the sequence of identification numbers obtained by decomposing the input search keyword. If, having the configuration and an output means for outputting a document with the sequence of the relevant, which is searched by the search means as a retrieval answer document.

[0007]

In the document retrieval device of the present invention, the adjacency information creating means decomposes the entire contents of the document for each document to be retrieved into word units, and replaces the word units with identification numbers to arrange the adjacency information in word order. To create. The search means searches the adjacent information corresponding to the document searched from the index information for the same sequence as the sequence of the identification numbers obtained by decomposing the input search keyword. The output means outputs the document having the corresponding sequence searched by the search means as the search response document. This eliminates the need for the search means to swipe the contents of the key search target document, thereby increasing the search speed and improving the search efficiency.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic block diagram showing an embodiment of a document search apparatus of the present invention to which the document search method of the present invention is applied and a document creation apparatus of the present invention equipped with this document search apparatus. Reference numeral 1 denotes a control device composed of a CPU and a memory for performing document search and document creation control, a keyboard and a mouse for inputting respective instructions and information for a user to input a search key or perform a search operation. An input device 2, a search key input by the input device 2 and a search operation instruction,
A display device 3 such as an LCD that displays the search result and the content of the document searched by the search, an external storage device 4 that stores the search target document (including the saved document obtained by creating the document), adjacent information, and the like. It is composed of a division dictionary 5 (configured on a memory) used for dividing a document into words or characters when creating information.

FIG. 2 is a block diagram showing a detailed example of the control device shown in FIG. The control device 1 includes an initialization unit 201,
A document capturing unit 202, a word cutting unit 203, a word frequency counting unit 204, a keyword sorting unit 205,
Adjacent information conversion unit 206, optimum data conversion unit 207, adjacency information storage unit 208, keyword list storage unit 209, input unit 220, index reading unit 221, keyword list reading unit 222, adjacency information reading unit 22.
3, keyword extraction unit 224, keyword adjacent information conversion unit 225, keyword matching unit 226, adjacent information matching unit 227, search response unit 228, output unit 229, and other processing units, and document number storage buffer 250 and document storage buffer 251. , Divided word storage buffer 252, frequency storage unit 253, frequency sort storage buffer 254, optimum data storage buffer 255, index storage buffer 271, keyword list storage buffer 272, adjacency information storage buffer 273, search key character string buffer 27
4, a keyword storage buffer 275, an adjacent information conversion buffer 276, a document ID storage buffer 277, a search response buffer 278, and the like.

Next, the processing contents of each functional block constituting the processing unit of the control unit 1 will be described. Initialization unit 201
Initializes each buffer unit. Document capture unit 2
02 stores the document number and the original document data from the external storage device 4 in the document number storage buffer 250 and the document storage buffer 251, respectively. Word cutout unit 203
Divides the document stored in the document storage buffer 251 into keywords such as words and characters using the division dictionary 5. The word frequency counting unit 204 counts the number of appearances of the keyword stored in the divided word storage buffer 252. The keyword sorting unit 205 calculates the appearance frequency based on the keywords stored in the frequency storage unit 253 and the number of appearances of the keywords, and sorts the appearance frequencies in descending order of appearance frequency. The adjacent information conversion unit 206 converts the data stored in the divided word storage buffer 252 into a keyword ID. The optimum data conversion unit 207 represents the keyword ID with optimum bits. The adjacent information storage unit 208 stores the adjacent information stored in the optimum data storage buffer 255 and the document number stored in the document number storage buffer 250 in the external storage device 4. The keyword list storage unit 209 uses the frequency sort storage buffer 2
The keyword list stored in 54 is stored in the external storage device 4
To store. The input unit 220 manages a search operation by the user using the input device 2. The index reading unit 221 reads an index indicating in which document the keyword stored in the external storage device 4 appears in the index storage buffer 271. The keyword list reading unit 222 selects “0, 1, ...” in order of the keywords having a high frequency of appearance in the search target document with respect to the keywords extracted from the search target document.
The keyword list with the keyword ID "2, ..." Is read from the external storage device 4 into the keyword list storage buffer 256. Adjacent information reading unit 2
23 reads the adjacent information index stored in the external storage device 4 into the adjacent information storage unit 253. The keyword extracting unit 224 converts the search key stored in the search key character string buffer 274 by the user into the units indexed from the keywords extracted from the documents to be searched, and the keyword units from the search key. It divides and stores the divided keywords in the keyword storage buffer 275. The keyword adjacency information conversion unit 225 uses the keyword list storage buffer 27 to convert the keywords stored in the keyword storage buffer 275.
According to the keyword list assigned with the keyword ID for each keyword stored in No. 2, the keyword is converted into the keyword ID in the arrangement of the search keys stored in the search key character string buffer 274. The keyword matching unit 226 represents a document ID that includes a keyword stored in the keyword storage buffer 275 that constitutes a search key input by the user and a keyword stored in the index storage buffer 271. Matching with the index is performed, and the document ID including all the keywords stored in the keyword storage buffer 275 is extracted. Adjacent information matching unit 2
The reference numeral 27 matches whether the arrangement of the keyword IDs stored in the adjacent information conversion buffer 276 is the same as the arrangement of the keyword IDs in the document ID unit index stored in the adjacent information storage buffer 273. , All matching document IDs are extracted. The extracted document ID is stored in the search response buffer 278. The search response section 228 obtains the result stored in the search response buffer 278. The output unit 229 displays the search results on the display device 3
Is displayed.

Next, the flow of operation of this embodiment will be described with reference to the flow charts of FIGS. 3, 4 and 5. First, the operation for creating the keyword list will be described with reference to the flowchart of FIG. First, the initialization unit 201 of the control device 1 is activated in step 301 to clear each buffer. Next, the document acquisition unit 202 is activated in step 302, reads the data of the r original document stored in the external storage device 4, and stores it in the document storage buffer 251. Here, the word cutout unit 203 makes the step 303
Starts, and the document stored in the document storage buffer 251 is divided into keywords such as words and characters by the division dictionary 5. An example of the division dictionary is composed of keywords that can be divided as shown in FIG.
The document divided into keyword units by the word cutout unit 203 is stored in the divided word storage buffer 252 in the keyword unit format as shown in FIG. The document capturing unit 202 determines in step 304 whether or not there is a document to be read, and if there is, returns to step 302, and if not, proceeds to step 305.

When the keyword division of all the documents is completed, the word frequency counting unit 204 is activated in step 305 to count the number of appearances of the keyword stored in the divided word storage buffer 252 to determine the keyword and its keyword. The number of appearances is stored in the frequency storage unit 253 as shown in FIG. Next, the keyword sorting unit 205 is activated, and in Step 306, the number of appearances of the keyword stored in the frequency storage unit 253 and the number of keywords are used to calculate how many such keywords appear in the document. , And the numbers are assigned to these rearranged keywords in order from 1 to 0, 1, 2, 3, ... As shown in FIG. )), Step 3
At 08, the data is stored in the frequency sort storage buffer 254 as shown in FIG.

For example, if the appearance frequency of the keyword "no" is highest, the keyword ID "0" is given to the keyword "no". On the contrary, a keyword having a low appearance frequency is assigned a large keyword ID such as "3,000". The keyword ID and the keyword group stored in the frequency sort storage buffer 254 are called a keyword list. When the keyword ID storage is completed, the keyword list storage unit 209 is activated in step 309, stores the keyword list stored in the frequency sort storage buffer 254 in the external storage device 4, and ends the processing.

Next, a specific example of creating the adjacent information will be described with reference to the flowchart of FIG. First, the initialization unit 201
Is activated in step 401 to clear each buffer. Next, the document capturing unit 202 performs step 4
In step 02, the original document and the original document number data stored in the external storage device 4 are read and stored in the document number storage buffer 250 and the document storage buffer 251, respectively. In step 403, the keyword list reading unit 222 is activated to load the keyword list stored in the external storage device 4 into the keyword list storage buffer 272.
Store to. In step 404, the word cutout unit 2
03 starts, and the document stored in the document storage buffer 251 is divided by the division dictionary 5 into keywords such as words and characters. The document divided into the keyword units in this way is stored in the divided word storage buffer 252 in the keyword unit format. Next, the adjacent information conversion unit 20
6 is started in step 405, the document divided in the keyword unit stored in the divided word storage buffer 252 is compared with the keyword list stored in the keyword list storage buffer 222, and in this keyword list Is converted to a keyword ID assigned to the same keyword. At this time, the optimum data conversion unit 20
7 starts in step 405, and after changing to a bit expression most suitable for expressing the keyword ID, the changed keyword ID is stored in the optimum data storage buffer 255.

The optimum bit is represented by 1 in 1 byte.
By using one bit as a flag, the keyword ID is represented by 1 byte or 2 bytes. In the example shown in FIG. 13, when the first bit of the byte is “1”, it is indicated that the next byte is continuous. For example, the keyword ID "10 ₍₁₀₎ " is "0a
₍₁₅₎ ", and the keyword ID" 400 ₍₁₀₎ "is expressed as" 8310 ₍₁₆₎ ". When all the keywords stored in the divided word storage buffer 252 are converted into keyword IDs, the adjacent information storage unit 208 is activated in step 406 to store the adjacent information and the document number stored in the optimum data storage buffer 255. Buffer 250
The document number (hereinafter referred to as “document ID”) of the original document used to create the main adjacency information stored in (4) is stored in the external storage device 4. What is stored in the external storage device 4 by the adjacent information storage unit 208 as shown in FIG. 11 is called an adjacent information index.

Next, a specific example of the search operation of the apparatus shown in FIG. 1 will be described with reference to the flowchart of FIG. First, in step 501, the initialization unit 201 initializes each buffer. In step 502, the index reading unit 221
Starts and stores an index indicating a keyword included in the search target document from the external storage device 4 in the index storage buffer 271. The index storage buffer 271 stores a keyword index indicating a keyword and a document number (document ID) including the keyword. The keyword index is composed of, for example, "keyword character string / document ID" as shown in FIG. Further, in the document ID, the numbers of a plurality of documents including the keyword character string are arranged. In the example of FIG. 14, the keyword “document” has a document ID “1, 4, 7, 9, 17, ...”
It is shown to be included in.

Next, the keyword list reading section 222
Starts in step 503, reads the keyword list from the external storage device 4, and stores it in the keyword list storage buffer 272. The keyword list storage buffer 272 stores keywords corresponding to keyword IDs. For example, as shown in FIG. 15, the keyword list storage buffer 272 stores data including “keyword ID / keyword”. This FIG.
In the example, the keyword ID “0” is “no”. Next, the adjacent information reading unit 223 is activated in step 504, reads the adjacent information index from the external storage device 4, and stores it in the adjacent information storage buffer 273. Here, this "keyword sequence" is the sentence itself of the document number. This sentence is divided into units of keywords as shown in FIG. 12, and each divided keyword is replaced by a number (keyword ID) assigned to the keyword in the keyword list, whereby the keyword sequence is expressed by the keyword ID. It

Thereafter, in step 505, it is determined whether or not the user has input information for selecting [end] from the input device 2, and when the end is selected, the present device is ended. However, if the user does not select the end described above and inputs the search key from the input device 2 such as a keyboard, the following operation is performed. The input search key is, for example, "create document". When such a search key is input, the search key is stored in the search key character string buffer 274 by the input unit 220. FIG. 16 shows this search key character string buffer 27.
4 is a storage example of No. 4. When the search key is input, the keyword extraction unit 224 is activated in step 507, and the search key character string is divided into units of keywords registered in the keyword index. Here, the search key character string is divided into, for example, "document / no / create". The divided keywords are stored in the keyword storage buffer 275 in FIG.
It is stored as shown in. At this time, the keyword adjacency information conversion unit 225 is activated in step 509, and according to the keyword list stored in the keyword list storage buffer 256, the keyword stored in the keyword storage buffer 256, for example, “document / of / create”. Is converted into a keyword ID and stored in the adjacent information conversion buffer 276 as shown in FIG. For example, "23,001.
It is converted to "0.23,000 ₍₁₀₎ ". After that, the matching process is performed.

In the matching process, first, the keyword matching unit 226 is activated in step 510 to match the keywords between the search keyword stored in the keyword storage buffer 275 and the index stored in the index storage buffer 271. Then, the document ID of the document including the keyword is extracted. In this example, the keywords "document", "no", and "creation" are "1, 4" as the document ID of the document appearing in the document.
Is extracted. These document IDs extracted by the keyword matching unit 226 are stored in the document ID storage buffer 27.
7 is stored as shown in FIG. In step 511, it is determined whether or not the document ID is extracted. If not extracted, the process returns to step 505, and if extracted, the process proceeds to step 512.

Next, the adjacency information matching unit 227 is activated in step 512, searches the adjacency information index of the document ID stored in the document ID storage buffer 277, and searches the keyword stored in the adjacency information conversion buffer 276. A document ID having adjacent information in the same sequence as the sequence of IDs is searched, only the relevant document ID is extracted, and the extracted document ID is stored in the search response buffer 278 as shown in FIG. Search for the corresponding document ID Response buffer 2
When it is stored in 78, the search response section 228 stores it in step 514.
Then, the document ID stored in the search response buffer 278 is acquired as a search response. In the example of FIG. 14,
The search response has the document ID “4”. Then step 5
The output unit 229 is activated at 15 and the search result is displayed on the display device 3. After that, in step 516, it is determined whether or not to continue narrowing down the search result, and if so, the process returns to step 505. When performing a new search without narrowing down, the user selects [Clear] with the input device 2 and returns to step 501.

According to the present embodiment, when the user inputs a word-character composite word or a sentence as a document search keyword, the keyword is decomposed into a plurality of keyword IDs and all the obtained keyword IDs are included. Figure 1 for the document ID
11 is extracted from the keyword index shown in FIG.
Document I obtained from the adjacent information index as shown in
The arrangement of the keyword IDs of D and the arrangement of the keyword IDs of the input keywords are compared, a document ID having the same arrangement of the keyword IDs as the arrangement of the keyword IDs of the input keywords is searched, and thus searched. Since the document indicated by the document ID is used as the search answer, there is no need to revise the document to be searched by the compound words and sentences that compose the input keyword, and the search speed is significantly increased to improve the search efficiency. Can be made. Therefore, in the document creation device equipped with such a document search device having high search efficiency, it is possible to quickly search the saved document or the like stored in the external storage device 4 or the like, thereby improving the document creation efficiency. be able to.

[0022]

As described above, according to the present invention, even if the search key is a compound word or sentence of words and characters and the number of documents to be searched is large, the target document can be searched at high speed, The search efficiency can be improved, and the document creation efficiency can be improved.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing an embodiment of a document search device of the present invention to which a document search method of the present invention is applied and a document creation device of the present invention equipped with this document search device.

FIG. 2 is a block diagram showing a detailed example of a control device shown in FIG.

FIG. 3 is a flowchart showing a flow of processing for creating a keyword list in the apparatus of FIG.

4 is a flowchart showing a flow of a process of creating adjacent information in the device shown in FIG.

5 is a flowchart showing a flow of document search processing in the apparatus shown in FIG.

FIG. 6 is a diagram showing an example of a division dictionary used in the apparatus shown in FIG.

7 is a diagram showing an example of assigning keyword IDs in order of appearance frequency by the device shown in FIG.

FIG. 8 is a diagram showing a storage example of a divided word storage buffer shown in FIG.

FIG. 9 is a diagram showing a storage example of a frequency storage buffer shown in FIG.

10 is a diagram showing a storage example of the frequency sort storage buffer shown in FIG.

FIG. 11 is a diagram showing a storage example of a frequency sort storage buffer shown in FIG.

FIG. 12 is a diagram showing an example of an adjacent information index created by the device shown in FIG.

FIG. 13 is a keyword I created by the apparatus shown in FIG.
The figure which showed the expression method of D.

FIG. 14 is a diagram showing an example of a keyword index created by the device shown in FIG.

15 is a diagram showing an example of a keyword list created by the device shown in FIG.

16 is a diagram showing a storage example of a search key character string buffer shown in FIG.

17 is a diagram showing a storage example of the keyword storage buffer shown in FIG. ,

18 is a diagram showing a storage example of the adjacent information conversion buffer shown in FIG. ,

FIG. 19 is a diagram showing a storage example of a document ID storage buffer shown in FIG.

20 is a diagram showing a storage example of the search response buffer shown in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Control device 2 ... Input device 3 ... Display device 4 ... External storage device 5 ... Divided dictionary 201 ... Initialization unit 202 ... Document acquisition unit 203 ... Word cutout unit 204 ... Word frequency counting unit 205 ... Keyword sorting unit 206 ... Adjacent Information conversion unit 207 ... Optimal data conversion unit 208 ... Adjacent information storage unit 209 ... Keyword list storage unit 220 ... Input unit 221 ... Index reading unit 222 ... Keyword list reading unit 223 ... Adjacent information reading unit 224 ... Keyword extraction unit 225 ... Keywords Adjacent information conversion unit 226 ... Keyword matching unit 227 ... Adjacent information matching unit 228 ... Search response unit 229 ... Output unit 250 ... Document number storage buffer 251 ... Document storage buffer 252 ... Divided word storage buffer 253 ... Frequency storage unit 254 ... Frequency sort Case Storage buffer 255 ... Optimal data storage buffer 271 ... Index storage buffer 256, 272 ... Keyword list storage buffer 273 ... Adjacent information storage buffer 274 ... Search key character string buffer 275 ... Keyword storage buffer 276 ... Adjacent information conversion buffer 277 ... Document ID storage Buffer 278 ... Search response buffer

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kenichi Nogami 1381 Shinmachi, Ome-shi, Tokyo 1 Toshiba Computer Engineering Co., Ltd. (72) Inventor Yukio Nakamoto 1381 Shinmachi, Ome-shi, Tokyo 1 Toshiba Computer In data engineering Co., Ltd. (72) Toshihiro Ozaki 1381 Shinmachi, Ome-shi, Tokyo Within Toshiba Computer Engineering Co., Ltd. (72) Toshie Yuzawa 1381 Shinmachi, Ome-shi, Tokyo Toshiba Computer Engineering Co., Ltd.

Claims (4)

[Claims] 1. When each search key has index information in which an identification number of a document to be searched including the search key is associated and a search keyword of two or more words is input, the search keyword is used as a word unit. In the document search device having a function of searching the index information for a document including all identification numbers constituting the search keyword after disassembling each word unit into an identification number, Adjacency information creating means for decomposing the entire contents of the document into word units and replacing the word units with identification numbers to create adjacency information arranged in word order; and the adjacency corresponding to the document retrieved from the index information. A search unit that searches the information for the same sequence as the sequence of the identification numbers obtained by decomposing the input search keyword, and a search unit that uses this search unit. A document search device, comprising: an output unit that outputs a document having a corresponding sequence that has been reached as a search response document. 2. When assigning an identification number to each word,
2. The document search device according to claim 1, wherein one bit in one byte expressing the identification number is used as a flag. 3. When assigning an identification number to each word,
3. The document search device according to claim 2, wherein numbers are given in the range of a natural number such as "0.1.2" in order from a word having a high frequency in the searched document. 4. A document search method for searching a document including an input search key from a plurality of searched documents, wherein the entire contents of the document are decomposed into word units for each document to be searched, The adjacency information in which the units are replaced with the identification numbers and arranged in the word order is created in advance, the input search key is decomposed into word units, and each word unit is replaced with the identification number, and the input search is performed. A document search method, characterized in that a key is made into a plurality of identification number sequences, and then a document having an identification number sequence in the same sequence as this identification number sequence is retrieved from the adjacent information.