JP3415214B2 - Document search device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document search device for quickly and accurately searching a document containing text data.

[0002]

2. Description of the Related Art Conventionally, when a document including text data is registered as a database, in order to easily specify a desired document at the time of search processing, a characteristic word for the document is added as a keyword together with the document data And register. When searching for a required document from a large number of documents registered in the database, one or more words are given as a search keyword, and a document to which a keyword corresponding to this search keyword is given is extracted as a candidate.

According to this method, although the search speed is relatively high, it takes a lot of work to add a keyword to a document to be registered. Furthermore, it is difficult to assign an appropriate keyword, and the document targeted by the searcher may not be obtained during the search.

On the other hand, in recent years, along with the increase in the processing speed of computers and the increase in capacity of storage devices, full-text search methods are being put to practical use. In this method, all the character strings (text data) contained in the document are searched when searching, so it is not necessary to assign a keyword to each document in advance, and the word specified by the searcher Documents including (keyword) can be extracted without omission.

However, according to this full-text search method, since the entire contents of the text data must be read at the time of searching, the processing time becomes longer as the number of registered document data increases. Therefore, when the number of registered documents is large, even if the processing power of the current computer is used, a huge amount of processing time is required to search for the specified character string, which makes it impractical. It was happening.

Furthermore, according to a general full-text search method, it is difficult to handle the concept of a word, and all documents including the input keyword as a partial character string are searched. For example, when searching for documents that contain the word "microphone,"
There was a problem that documents containing words not related to "Mike" were also searched. That is, the character string “microcomputer” includes a partial character string “microphone”, and matching is performed on this partial character string.

[0007]

As described above, the conventional document search apparatus can always search for a document including a character string designated by the searcher and a keyword is added by the full-text search method. It is excellent in that it does not require any work, but on the other hand, there is a problem that as the amount of data increases, processing time increases, and when the amount of data increases, the search capability decreases significantly. .

Further, in the full-text search method, in order to make a distinction between a word, which is one unit of meaning, and a simple character string at the time of searching, a character string having a target word as a partial character string therein is provided. Such as searching for documents containing
There was a problem that the word intended by the searcher was not searched correctly.

The present invention has been made in consideration of the above circumstances, and provides a document search apparatus capable of quickly and accurately searching a large number of documents for an appropriate document including a designated word. The purpose is to provide.

[0010]

The present invention SUMMARY OF THE INVENTION from among the plurality of documents containing text data consisting of character code strings, the document search apparatus for searching the candidate documents that contain keywords of strings given arbitrarily, For each character code included in the document, a character appearance position table that stores information indicating the position of each character in the document and the position within the word that contains the same character Table management file storage means for storing a table management file for managing, keyword input means for inputting a character string as a keyword, and table management corresponding to each character of the keyword input by the keyword input means. files, and table management file selection <br/> detecting means for selecting from said table management file storage means, before Based on the character appearance position table of each table management files selected by the table management file selection means, the character is determined whether include keywords entered by the keyword input unit in the text data of one document It comprises a column search means and a search result display means for displaying the content of the candidate document determined to contain the keyword by the character string search means.

[0011]

According to this structure, by using the appearance position information stored in the character appearance position table and the information indicating the position within the word in the character string search means, the character specified as the keyword It is possible to quickly determine whether or not a column exists in each document data, and further,
By using the word existence determination means, it is possible to distinguish between a word that is one unit of meaning and a simple character string when searching, and use a character string that internally has the word of the keyword as a partial character string. Prevents the containing document from being searched as a candidate document.

[0012]

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 block diagram showing a schematic configuration of a document search device according to this embodiment. As shown in Figure 1,
The document search device includes a control device 10, an input device 11, a display device 12, a memory device 13, and an external storage device 14.

The control device 10 is composed of a CPU, a memory, etc., and controls the entire document retrieval device.
It has the functional configuration shown in FIG. The control device 10 uses, for example, UNIX as the OS (operating system) as the main control unit 21, and operates according to a program stored in the memory device 13.

The input device 11 is a command for the device,
This is a device for inputting keywords and the like, and is composed of a keyboard, a mouse and the like. The display device 12 is a device that displays a prompt message, a search result, and the like that this device gives to the user, and is configured by, for example, a VRAM, a CRT, and a controller.

The memory device 13 is composed of a RAM and a ROM, and stores programs for controlling the CPU of the control device 10 and various data. Further, a work area for performing various kinds of processing is secured.

The external storage device 14 is composed of, for example, a magnetic disk device, and stores a character appearance position table and various information for managing these. In addition, text data or chart data necessary for displaying the search results is also stored. Data is managed as files and can be accessed randomly in the file system.

Next, the control device 10 and the memory device 13
Details of the functional configuration of (based on the program stored in) will be described. As shown in FIG. 2, the functional configuration includes a processing unit 20 and a work area unit 30.

The processing unit 20 is provided with a main control unit 21, an initialization unit 22, a keyword input unit 23, a character appearance position table selection unit 24, a character string reach unit 25, a document selection unit 26, and a search result display unit 27. ing. Work area 30
Is a keyword character number storage buffer 31, a keyword storage buffer 32, a table management file storage buffer 3
3, a candidate document number storage buffer 34, a candidate document number storage buffer 35, and a display document number storage buffer 36 are provided.

The main control section 21 is a section for controlling the processing of the entire apparatus, and gives each functional section an instruction for displaying a prompt message to the user and for branching the processing.

When executing the database search process, the initialization unit 22 initializes each hardware device required for the search process and initializes each buffer of the work area unit 30. When executing the database search process, the keyword input unit 23 inputs, from the input device 11, a character string serving as a search keyword for searching a desired document from a large number of documents in the database stored in the external storage device 14. The keyword input unit 23 stores the input keyword in the keyword storage buffer 32, and stores the keyword character number in the keyword character number storage buffer 31.

The character appearance position table selecting unit 24 receives a table management file corresponding to each character in the keyword according to the keyword input from the keyword input unit 23 and stored in the keyword storage buffer 32, and the work area unit 30. It is selected from the table management file storage buffer 34 and stored in the table management file storage buffer 33.

The character string search unit 25 refers to the contents of the table management file stored in the table management file storage buffer 33 and refers to the keyword storage buffer 32.
A candidate document to be searched including the word of the keyword stored in is searched from the database stored in the external storage device 14, stored in the corresponding candidate document number storage buffer 35, and the number of candidate documents is stored in the number of candidate documents. The data is stored in the buffer 34.

The document selection unit 26 displays a list of candidate documents indicated by the candidate document numbers stored in the candidate document number storage buffer 35 which are searched by the character string search unit 25, and when there are a plurality of candidate documents, the document selection unit 26 searches. A document corresponding to the target is selected, and the display document number indicating the selected document is stored in the display document number storage buffer 36.

The search result display unit 27 is a character string search unit 2.
Document selection unit 26 among the candidate documents searched by
The content of the document identified by the display document number stored in the display document number storage buffer 36 specified by
Display on screen 2.

The keyword character number storage buffer 31 is for storing the number of keyword characters input from the input device 11 via the keyword input unit 23. The keyword storage buffer 32 is for storing input keywords (character strings), and is stored in an array (character buffers 0, 1, ...) As shown in FIG.

Table management file storage buffer 33
Is for storing a table management file for managing the character appearance position table, and has an array of structures as shown in FIG. That is, the table management file storage structures 0, 1, 2, ... Are arranged.

Each table management file storage structure of the table management file has a structure as shown in FIG. 4 (b). In one table management file storage structure, a file size storage area for storing the file size of this file, a table number storage area for storing the number of character appearance position tables stored in this file, Further, a character appearance position table storage area for storing each body of the character appearance position table (0 to N-1) and a pointer for storing each pointer to each character appearance position table (0 to N-1) Storage area,
And the document number (0-N-) corresponding to each pointer storage area.
A document number storage area for storing 1) is provided.

Further, the table management file storage structure is provided with an area for storing variables and the like necessary for various processes at the time of character string search. In this area, a reference table number storage area for storing which internal character appearance position table is referred to during processing, and which appearance position data in the character appearance position table are referenced An appearance position storage area is provided for storing whether or not it is being performed.

In each character appearance position table storage area, as shown in FIG. 5B, an appearance total number storage area for storing the total number of appearances and appearance position data (0, 1, ...) Are stored. An appearance position data storage area is provided.

In the present embodiment, hereinafter, the i-th structure in the table management file storage buffer 33 is represented by TMF [i], and the file size stored in the file size storage area of this structure is TMF [i]. ] .fsize, the number of character appearance position tables included in one structure stored in the table number storage area is TMF [i] .nCLT, the reference table number of the reference table number storage area is TMF [i] .iCLT, The appearance position storage area is represented by TMF [i] .iLOC, the jth character appearance position table is TMF [i] .CLT [j], and the document number corresponding to the jth character appearance position table is TMF [i]. It will be represented by .ID [j].

Further, the total number of appearances stored in the total number of appearances storage area provided in the jth character appearance position table storage area is TMF [i] .CLT [j] .N, and the kth character appearance position table is kth. The appearance position of TMF [i] .CLT [j] .LOC [k] is represented by TMF [i] .CLT [j] .PIW [k].

The candidate document number storage buffer 34 is for storing the number of candidate documents obtained by the search by the character string search unit 25. The candidate document number storage buffer 35 is for sequentially storing ID numbers of candidate documents obtained by the search by the character string search unit 25, and has an array as shown in FIG. The display document number storage buffer 36 determines that the user wants to display the candidate documents among the document candidates.
It is for storing the ID number of the document selected by.

Next, the structure of each file stored in the external storage device 14 will be described. As shown in FIG. 11A, the external storage device 14 stores a plurality of document data to be searched as document files. Each document data has an integer document number (0,
1, ...) Are added, and the names of the respective document files are also corresponding thereto. One document file is
As shown in FIG. 11B, it includes text data forming a document and data of a chart or an image included in the document.

In the external storage device 10, in addition to the document data shown in FIG. 11A, a file (table management file) for managing the table including the character appearance position table is shown in FIG. 11C. It is stored as shown. First, the contents of the character appearance position table included in the table management file will be described.

The character appearance position table has a structure as shown in FIG. 7 (a), and in the text data of an arbitrary document stored in the external storage device 10, all the characters in which an arbitrary character appears. The position is described along with the total number of occurrences. In addition, the information on the position of the character in the word to which the character belongs is described in association with each appearance position.

In FIG. 7A, the total number N of occurrences of an arbitrary character in text data in an arbitrary document is stored at the head of the character appearance position table, and subsequently, in each document. The appearance positions (0 to N-1) in are stored as N integer strings. The position information of the character in the word to which the character belongs, that is, in-word position information (0 to N-1) is stored in association with each appearance position. This information is obtained by subjecting the text data of the document to morphological analysis in advance and cutting the words.

Here, a specific example of the data stored in the character appearance position table will be described. The in-word position information is expressed by numbers, and its definition is set as shown in FIG. That is, the character is "1" at the beginning of the word, the character is "2" at the end of the word,
The character is defined as "3" if it is the beginning and end of a word, and "0" if it is neither the beginning nor end of a word.

For example, when morphological analysis is performed on the text data "Taro goes to junior high school" and word cutting is performed, "Taro / wa / junior high school / ni / go" (/ is word cutting It shows the position).

Therefore, the in-word position information of each character in the text data shown in FIG. 9 is represented as shown in FIG. That is, for example, for the character "thick", since it is at the beginning of the word "Taro", the in-word position information is "1".
For other characters, depending on the position in the word,
As shown in 0, in-word position information is set for each.

FIG. 7B shows a specific example in which the character appearance position is described in the text data having an arbitrary character, for example, the character "sentence". The character "sentence" appears 7 times in the text data, and its appearance position is 56,
123, ..., 6011, and the in-word position information determined as described above corresponding to each appearance position is described.

Next, the entire table management file will be described. The table management file is a file for managing a character appearance position table in which character appearance position tables for a plurality of documents are arranged inside the file. That is, the same character code (for example, JI
This is for gathering the character appearance position tables for a plurality of documents for the character of S code). The external storage device 14 stores a plurality of table management files each corresponding to a character code of an arbitrary character, and is given a file name corresponding to the character code.

FIG. 11C shows the general structure of the table management file. From the head of the table management file, the file size and the number of documents (N) including the target character are stored. Subsequently, the document number of the document including the target character and the position (table position) from the beginning of the file of the character appearance position table corresponding to this number are stored in association with each other. Subsequent to them, the body of the character appearance position table (0 to N-1) corresponding to each document having the structure shown in FIG. 7 is stored.

Next, the operation of this embodiment will be described with reference to the flowcharts shown in FIGS. First, the initialization unit 22 is activated by the program stored in the memory device 13. The initialization unit 22 performs initialization of the control device 10, the input device 11, the display device 12, and the external storage device 14, initialization of a work area (work area unit 30) in the memory device 13, display of an initial screen, and the like. Perform (step A1).

Next, the keyword input section 23 is activated. The keyword input unit 23 causes the input device 11 to input a keyword (character string) for searching for a desired document from the document file stored in the external storage device 14 (step A2).

In the present embodiment, it is assumed that each character of the words forming the keyword is represented by a 2-byte JIS code. The keyword input unit 23 sequentially inputs the character codes of the respective characters in the order in which they are input into the keyword storage buffer 32.
And the number of input characters is stored in the keyword character number storage buffer 31.

Next, the character appearance position table selecting section 24 is activated (step A3). The character appearance position table selection unit 24 calls a table management file corresponding to the character code stored in the keyword storage buffer 32 from the external storage device 14, and sequentially calls each structure in the table management file storage buffer 33. To store.
The character codes stored in the keyword storage buffer 32 are sequentially represented as C0, C1, ..., Cn-1. Further, n indicates a numerical value (keyword character number) stored in the keyword character number storage buffer 31.

In the present embodiment, as the file name of the table management file stored in the external storage device 14, a file identifier ".tmf" is added to the corresponding character code expressed in ASCII format. For example, the file name of the table management file corresponding to the character "love" whose JIS code is hexadecimal "3026" is "3026.tmf".

Therefore, when calling one table management file, first the character code of the target character is 16
Generates a character string in which the file identifier ".tmf" is added to the character string represented by a decimal number. Character appearance position table selection unit 2
4 reads the corresponding file from the external storage device 14 via the main control unit 21 with the generated character string as the file name. The character appearance position table selection unit 24 stores the file read by the main control unit 21 in FIG.
Also, the data is stored in the table management file storage buffer 33 with the configuration shown in FIG.

When the contents of each table management file corresponding to the target character are stored in the table management file storage buffer 33, the character string search unit 25 is activated. The character string search unit 25 searches for a document including a keyword while referring to the data of the table management file structure stored in the table management file storage buffer 33, and stores the document numbers of the corresponding documents in order of candidate document numbers. The process of storing in the buffer 35 is executed.

Regarding the processing of the character string search unit 25, FIG.
This will be described with reference to the flowchart shown in FIG. FIG.
Shows the outline of the processing for searching the candidate documents.
First, the character string search unit 25 initializes each variable used for searching a document, i.e., the value of i for specifying the table management file storage structure and the value of j for specifying the character appearance position table (step B1). ). Note that i and j are integer type variables. In addition, the character string search unit 25
Substitutes 0 into the value of the candidate document number storage buffer 34 (nCand), and then the table management file storage buffer 33
Reference table number storage area (0≤i <n included in each structure (corresponding to each character in the keyword) stored in
0 is assigned to TMF [i] .iCLT of.

Next, the character string search unit 25 makes a common document in all the table management files (all structures) from the 0th to the (n-1) th stored in the table management file storage buffer 33. It is determined whether or not the character appearance position table generated from is included (step B2). That is, it is determined whether or not there is a document including all the characters that form the character string of the keyword (character existence determination process). The character presence determination process in step B2 of the character string search unit 25 will be described with reference to the flowchart shown in FIG.

Here, the reference table number TMF [0] .iCLT indicates which number character appearance table in the 0th table management file is referred to. The value of this reference table number TMF [0] .iCLT is 0 in the initial state and is updated in the subsequent processing. This reference table number TMF [0] .iCLT
The document number of the document corresponding to the character appearance position table indicated by can be represented by TMF [0] .ID [TMF [0] .iCLT].

The character string search unit 25 first sets i = 1 (step C1), and the value of j in the table number TMF [i] .nCLT from the reference table number TMF [i] .iCLT (i-th table management). Shows each character appearance position table in the file)
, The document number TMF [i] .ID [j] is the reference table number TMF in the 0th table management file (structure 0).
Document number TMF [0] .ID [TMF [0] .iCLT] corresponding to [i] .iCLT
It is checked whether there is a value of j that is equal to (step C2).

The reference table number TMF [i] .iCLT is a variable whose initial value is 0 and is updated in the processing of step C3, and the number of tables TMF [i] .nCLT is a fixed value and one table management data. The total number of character appearance position tables included in the table is shown.

At this time, the character string search unit 25 sets the minimum value of j to TMF [i] .iCLT and the maximum value of j in order to speed up the processing.
Perform ballina research in the range of TMF [i] .nCLT-1.
That is, in step C2, the character string search unit 25 is incremented by incrementing the value of i in step C6.
Is the document number TMF [i] .I in each file for each of the first to n-1th table management files.
Is D [j] the same as the document number TMF [0] .ID [TMF [0] .iCLT] corresponding to the reference table number of the 0th table management file (whether there is a corresponding j value? ) Is determined.

After the processing of step C2, the character string search unit 1
5 updates TMF [i] .iCLT (step C3).
In step C2, the character string search unit 25 can obtain the value of j that satisfies the condition (when there is the character appearance table obtained from the common document), the reference table number TMF [i] .iCLT. The value of j is stored as the value of. Here, if the value of j is TMF [i] .nCLT-1, then j is used as the value of the reference table number TMF [i] .iCLT.

On the other hand, if the value of j satisfying the condition does not exist, the document number TMF [i]. Which satisfies the condition that the value of the document number TMF [0] .ID [TMF [0] .iCLT] is exceeded. The smallest ID [j] is used as the value of the reference table number TMF [i] .iCLT. If the document number TMF [i] .ID [j] that satisfies this condition does not exist, the value of TMF [i] .nCLT-1 is used as the value of TMF [i] .iCLT.

At step C4, the process branches depending on the result of the process at step C2. The character string search unit 15
If the search is unsuccessful in step C2, a determination failure flag is set and the process of step B2 is ended (step C5).

On the other hand, when the search is successful, the character string search unit 15 increments the value of i (step C6). Then, the process of step C2 is executed as described above.

By executing the process of step C2 while sequentially incrementing the value of i, if the search is successful for all the table management files, that is, if the document including all the characters forming the character string of the keyword is found. If it exists, a flag indicating success in determination is set and the processing in step B2 ends (step C8).

When the character string search unit 15 completes the character presence determination process of step B2, if the result of the process indicates that the determination failure flag is set, the process proceeds to post-process of step B7 (step B3). If the determination success flag is set, character connection determination and word presence determination processing are performed (step B4).

In the character presence determination process of step B2, it is only determined whether or not there is a document including all the characters constituting the character string of the keyword, the connection relation of each character is checked, and the character string to be searched is searched. Is not determined as a word, and these processes are performed in step B4.

Details of the processing in step B4 are shown in FIG.
This will be described with reference to the flowchart shown in FIG. here,
Reference table number TMF [i] .iCL determined in step B2
The value of T, that is, 0 corresponding to the first character of the keyword
With respect to the character appearance position table for the document number TMF [0] .ID [TMF [0] .iCLT] corresponding to the reference table number for the second table management file, the character appearance position data therein is referenced.

Then, for 0≤i <n-1, TMF [i + 1] .CLT [TMF [i + 1] .iLOC] = TMF [i] .CLT [TMF [i] .iL
It is checked whether there is a TMF [i] .iLOC pair (0 ≦ i <n) satisfying OC] +1.

Thus, in the document, there is a portion in which all the characters forming the character string of the keyword exist adjacent to each other, that is, the specified keyword is included as the partial character string in the corresponding document. It is determined whether or not there is.

First, the character string search section 15 is a variable for storing the current reference number in each character appearance position data (these are mainly used in step D4 to improve the efficiency of the binary search). Are initialized (step D1).

Next, the character string search unit 15 initializes a variable i indicating which number of the table management file is referred to (i = 0) (step D2), and further, the first character of the keyword character string. Location of TMF [0] .CLT [TMF [0] .iLOC] .LO
C [TMF [0] .iLOC] is set to the variable P (step D3).

In this case, all the appearance positions stored in the appearance position data in the 0th table management data are the positions of the first character of the keyword character string. The variable TMF [0] .iLOC indicates the number of the reference.

Next, the character string search unit 15 determines whether the appearance position data in each appearance position table has a value P + i, that is, a value k after i characters of the first character position of the keyword. Find out by searching. In this binary search, the lower limit of TMF [i] .iLOC in the character appearance position table is
The upper limit is limited to only the TMF [i] .CLT [TMF [i] .iCLT.Nth appearance position data (step D4).

Next, the character string search section 15 updates the value of TMF [i] .iLOC used in the next step D4 (step D5). If the search is successful in step D5, the value of k is substituted as the value of TMF [i] .iLOC. If the search is unsuccessful, TM in step D4 that satisfies the condition that the value of TMF [i] .iLOC exceeds the value of P + i
Use the smallest of F [i] .CLT [j] .LOC [k]. When there is no one satisfying the condition, the value of TMF [i] .CLT [j] .N-1 which is the number of elements-1 in the corresponding appearance position table is used.

In step D6, the process branches depending on the result of step D4. If the binary search is successful, the process moves to step D7, and if it fails, the process goes to step D1.

If the binary search is successful, the character string search unit 25 increments the value of i (step D7). The character string search unit 25 uses the incremented i
The value of n and the value of n, that is, the keyword is compared with the number of characters in the character string. As a result, if i ≧ n, the condition is satisfied for all characters. Therefore, the existence of the character string corresponding to the keyword is confirmed. It will be. In this case, the character string search unit 25 further performs a word existence determination process, and performs a word determination process of whether or not the searched character string is regarded as one word as a result of the morphological analysis. Perform (step D9).

Details of the word determination process in step D9 will be described below with reference to the flowchart shown in FIG. In FIG. 16, first, the intra-word position information of the leading character of the character string searched by the processing up to step D7 is referred to, and it is checked whether or not this is the beginning of the word (step E1).

Here, the in-word position information of the first character of the character string searched by the above-mentioned processing is set to TMF [0] .CLT [TMF
It is represented by [0] .iCLT] .PIW [TMF [0] .iLOC], and it is determined whether or not this is the intra-word position information “1” indicating the beginning of the word.

As a result, if the determination is successful, step E3
If it fails, the process branches to step E7. If the determination is successful, the character string search unit 25 refers to the in-word position information from the first character to the n-1th character of the character string searched by the processing up to step D7, and all of these are found. , It is determined whether or not it is the in-word position information "0" indicating that it is neither the beginning nor the end of the word (step E3).

The position information in each word is TMF [k] .CLT [TMF [k] .iCLT] .PIW [TMF [k] .iLOC] (k = 1,
2, ... n-1). If the determination is successful here, step E5.
If it fails, the process branches to step E7. If the determination is successful, the character string search unit 25 refers to the in-word position information of the character at the end of the character string searched in the processing up to step D7, and checks whether this is the end of the word (step E5).

Here, the in-word position information of the first character of the searched character string is TMF [n-1] .CLT [TMF [n-1] .iCLT] .PIW [TMF [n-1] .iLOC. ], And it is determined whether this is the intra-word position information “2” indicating the end of the word. If the determination is successful here, the process branches to step E8 if it fails, and the process branches to step E7.

If the determination is successful, the character string search unit 25
Sets a flag for successful determination and returns (step E
8). If the determination fails, the character string search unit 25
Sets the flag of judgment failure and returns (step E
7).

The above is the flow of processing in step D9. Here, a case has been described in which a word having three or more characters is input as a keyword, but in the case of a word having a two-character keyword, only the first and last characters are checked in the processing described above.

When the keyword is a word consisting of one character, only the value of TMF [0] .CLT [TMF [0] .iCLT] .PIW [TMF [0] .iLOC] is referred to, and this is the beginning of the word. It is only determined whether or not it is the intra-word position information "3" indicating that it is the end position.

After the word existence determination process, the character string search unit 2
If the determination failure flag is set according to the processing result in step D9, step 5 returns to step D4 and repeats the above-described processing. If the determination success flag is set, the determination success flag is set again and the process returns (step D11).

On the other hand, if the search in step D4 fails (step D6), the character string search unit 25 indicates a variable TMF [which indicates which number in the character appearance position table in the 0th table management data is referred to. 0] .iLOC value is incremented (step D12).

First, the character string search section 25 executes step D1.
The value of TMF [0] .iLOC incremented by 2 is compared with the number of elements TMF [0] .CLT [TMF [0] .iCLT] .N of the corresponding appearance position table of the 0th table management data. If the former value is greater than or equal to the latter value, a judgment failure flag is set and the process returns (step D13).

Otherwise, the character string search unit 25
Variable TMF for binary search corresponding to 1 ≦ i <n
Increment the value of [i] .iLOC. If even one of these values is equal to or larger than the number of elements in the corresponding appearance position table, a judgment failure flag is set and the process returns (step D14).

If none of the above conditions are satisfied, the processing from step D2 is repeated. That is, 0
Among the elements of the character appearance position table in the th table management data, a series of processing is performed for the next one following the previous one.

The above is the flow of processing in step B4 (FIG. 13). In FIG. 13, step B5 receives the result of step B4, and if the determination failure flag is set, the intermediate processing is not performed and the processing proceeds to step B7.
If the determination success flag is set, the character string search unit 25
Registers the relevant candidate document number in the candidate document number storage buffer 35, and at the same time stores the candidate document number storage buffer 34.
The number of candidate documents is stored in (step B6).

That is, in step B6, the ID number of the document currently referred to in step C2, that is, TMF [0] .ID [TMF [at the position indicated by the candidate document number storage buffer 34 in the candidate document number storage buffer 35. [0] .iCLT] is stored, the content of the candidate document number 34 is incremented, and the process proceeds to step B7.

In step B7, the character string search section 25 is a variable indicating which number of appearance position data in the 0th table management data is referred to as post-processing of the series of processing shown in FIG. Increment TMF [0] .iCLT.

At the subsequent step B8, the value of TMF [0] .iCLT incremented at step B7 and the value of the variable TMF [0] .nCLT indicating the total number of appearance position tables included in the 0th table management data. Are compared, and if the former value is equal to or larger than the latter value, the processing in the character string search portion 25 is finished and the process returns (step B9).

If not, the character string search unit 25
Increments the value of TMF [i] .iCLT corresponding to 1 ≦ i <n. If even one of these values is equal to or more than the total number TMF [i] .nCLT of the corresponding appearance position tables, the process returns to the main process (step B9). If none of the above conditions are satisfied, the series of processes described above from step B2 is repeated.

When the character string search processing in step A4 is completed in this way, the document selecting section 26 is driven next. First, the document selection unit 26 displays a list of document titles corresponding to each document number stored in the candidate document storage buffer 35 on the screen of the display device 12.
FIG. 17 shows an example of the status of the screen on which the list of titles is displayed.

After displaying the list of titles, the document selection section 26 inputs the title designation by the user using the input device 11, and stores the document number corresponding to the designated title in the display document number storage buffer 36 ( Step A
5).

Next, the search result display unit 27 is driven (step A6). The search result display unit 27 reads the document file corresponding to the document number stored in the display document number storage buffer 36 from the external storage device 14. The search result display unit 27 expands the text data in the document file into a character font, and expands the chart or image data into a bit image and displays it on the screen of the display device 12. FIG. 18 shows an example of a screen displaying text data, charts, and image data.

After the processing of step A6, the search result display unit 27 displays two icons ("end icon" and "research icon") as shown in FIG. 19 on the screen (step A7).

Here, when another title is selected by the user using the input device 11, as described above,
Display the document corresponding to the selected title. Further, when the “end” icon is designated using the input device 11, all the processes of this device are ended. On the other hand, if the "re-search" icon is specified, the process returns to step A2 in FIG. 12 to execute a series of document search processing based on other keywords.

In the above embodiment, one table management file stored in the external storage device 14 is all loaded into the table management file storage buffer 33. Only the position table may be loaded and stored.

Furthermore, if the data in the table management file is not loaded onto the memory at all and the external storage device 14 is a magnetic disk device, for example, the above-mentioned character string search processing is performed only by seeking the magnetic head. May be.

Further, in the above embodiment, the result of the morphological analysis is used to search only the document including the word input as the keyword, and the document including the character string input as the keyword as the partial character string is searched. Although not searched, the document including the partial character string may be displayed by considering the priority order when displaying.

Further, as the format of the intra-word position information stored in association with each character, a method other than this example can be used as long as the result of the morphological analysis is reflected. In short, the present invention can be implemented with various modifications as necessary without departing from the spirit of the present invention.

[0100]

As described above, according to the present invention, when a document search is performed by a full-text search for a large amount of document databases, it is possible to quickly obtain a target document. At that time, it becomes possible to accurately select the originally desired document in consideration of the document including the character string input as the keyword as the partial character string in the word. It is possible to obtain a great effect in practice, such as eliminating the need to read.

[Brief description of drawings]

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

FIG. 2 is a block diagram showing a functional configuration realized by a control device 10 and a memory device 13 in this embodiment.

FIG. 3 is a keyword storage buffer 31 in this embodiment.
For explaining the structure of the.

FIG. 4 is a diagram for explaining a partial structure of a table management file storage buffer 33 in the present embodiment.

FIG. 5 is a diagram for explaining a partial structure of a table management file storage buffer 33 in the present embodiment.

FIG. 6 is a candidate document number storage buffer 3 in this embodiment.
4 is a view for explaining the structure of FIG.

FIG. 7 is a diagram for explaining the structure of a character appearance position table in this embodiment.

FIG. 8 is a diagram showing the definition of in-word position information in the present embodiment.

FIG. 9 is a diagram showing an example of word segmentation by morphological analysis on text data in the present embodiment.

FIG. 10 is a diagram showing an example of in-word position information given to each character in text data in the present embodiment.

FIG. 11 is a diagram for explaining the structure of a document file and a table management file stored in the external storage device 14 in this embodiment.

FIG. 12 is a flowchart showing the flow of overall processing according to the present embodiment.

FIG. 13 is a flowchart showing an outline of the flow of processing by the character string search unit 25 in the present embodiment.

FIG. 14 is a flowchart showing the flow of character presence determination processing in this embodiment.

FIG. 15 is a flowchart showing the flow of processing for character connection and word presence determination in this embodiment.

FIG. 16 is a flowchart showing the flow of word presence determination processing in this embodiment.

FIG. 17 is a diagram showing an example of a screen after displaying a list of titles in the present embodiment.

FIG. 18 is a diagram showing an example of a screen after displaying document data according to the present embodiment.

FIG. 19 is a diagram showing an example of a screen on which an icon for performing an ending process according to the present embodiment is displayed.

[Explanation of symbols]

10 ... Control device, 11 ... Input device, 12 ... Display device, 1
3 ... Memory device, 14 ... External storage device, 21 ... Main control unit, 22 ... Initialization unit, 23 ... Keyword input unit, 24
... Character appearance position table selection section, 25 ... Character string search section, 26 ... Character selection section, 27 ... Search result display section, 31 ...
Keyword character number storage buffer, 32 ... Keyword storage buffer, 33 ... Table management file storage buffer,
34 ... Candidate document number storage buffer, 35 ... Candidate document number storage buffer, 36 ... Displayed document number storage buffer.

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06F 17/30 JISST file (JOIS)

Claims (4)

(57) [Claims] 1. A document retrieval apparatus for retrieving a candidate document containing a keyword consisting of an arbitrarily given character string from among a plurality of documents containing text data consisting of a character code string, wherein each character code contained in the document A table management file for managing the character appearance position table that stores the information indicating the position of each character in the document and the position in the word that includes the same character for each document for each document A table management file storage means for storing a keyword, a keyword input means for inputting a character string as a keyword, a table management file corresponding to each character of the keyword input by the keyword input means, Table management function selected from the means
Whether the keyword input by the keyword input means is included in the text data of one document based on the file selection means and the character appearance position table of each table management file selected by the table management file selection means. A document search device comprising: a character string search unit for determining whether or not there is a search result display unit for displaying the content of a candidate document determined to contain a keyword by the character string search unit. 2. The information indicating the position of each character in the word in the character appearance position table is information indicating whether the character position is at the beginning or the end of the corresponding word. The document retrieval device according to claim 1, characterized in that 3. The information indicating the position of each character in the word stored in the character appearance position table is subjected to morphological analysis on the text data of each document to be searched and word-by-word units in the text. The document search device according to claim 1, wherein the document search device is set based on a result of division. 4. The character string search means refers to position information of each character in the word in the character appearance position table within a word, and the input keyword is not only used as a partial character string in the text of the document. 2. The document search apparatus according to claim 1, further comprising a word presence determination unit that determines whether or not the word exists.