JP3328995B2 - Information retrieval system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-speed search method such as a keyword search for generating a search file in addition to an original information file and performing a high-speed search, and a normal search method for directly accessing and searching the original information file. And an information retrieval system having:

[0002]

2. Description of the Related Art A method of searching for information stored in a file can be roughly divided into a method of directly accessing an original information file to be searched and a method of searching for a keyword from the original information file. There is a method of generating the search file in advance and searching for the search file. Obviously, the search speed is faster when searching for a search file than when using a search method that directly accesses the original information. However, in order to generate a search file, the contents of the original information file are subjected to pattern extraction processing or the like to extract words that are keywords used in the search, and a search file is generated. There is a method in which a keyword dictionary storing a set of is prepared in advance, and a keyword is extracted from the file of the original information based on the keyword dictionary to generate a search file. However, it takes a considerable amount of time to generate a search file. Generally, the time required to generate the search file increases as the size of the original information file increases.
Further, when the original information file or the keyword dictionary is updated, it is necessary to recreate this search file.

[0003]

[Problems to be Solved by the Invention] Even if a high-speed search method is used, it takes a lot of time to generate a search file. Therefore, the search is performed by directly accessing the original information without generating the search file. May be more convenient for the operator.

For example, if the same original information file is searched many times, it is better to generate a search file. However, if the search is performed only once, the search file is not generated. It is better to directly access and search the original information file. However, this is not always the case. When performing a search two or three times, it was impossible to know which search method was better without actually searching.

[0005]

FIG. 1 is a block diagram showing the principle of the present invention. 1 is an input device, 2 is a display device, 3 is a disk device, and 4 is a processing device. 31 is an original information file to be searched, 32 is a search file used for high-speed search, and 33 is a calculation table used for calculating a search file generation time. Reference numeral 41 denotes a calculation table updating unit, which is activated every time a search file is generated, and updates the calculation table 33 using data on the generation time of the search file 32. 42 is a search method control unit, 43 is a generation time calculation unit that calculates the generation time of the search file 32, 44 is a search file generation unit, 45 is a high-speed search unit that searches using the search file 32, and 46 is It is a normal search means for performing a search by directly accessing the original information.

[0006] From the original information file 31 to the search file 3
Generation time calculation means 43 for calculating the time required for generation of
And the time required to generate the search file 32 obtained by the generation time calculation means 43 is output to the display device 2, and the search file 32 is generated according to the input from the input device 1, and the search file 32 is used. Normal search means 46 for directly accessing the high-speed search means 45 or the original information file 31
And a search method control unit 42 for executing any one of the above.

[0007]

The operator considers the time required for generating the search file 32 to be output to the display device 2 and the work to be performed by the operator himself / herself. Normal search means 46 for directly accessing information
Then, the information search system searches for information by the search method selected by the operator.

[0008]

FIG. 2 is a block diagram of one embodiment of the present invention. In the figure, the same components as those shown in FIG. 1 are denoted by the same symbols. Reference numeral 34 denotes a keyword dictionary, which stores keywords used for search, and is stored in the original information file 31.
It is used when generating the search file 32. 3
5 is a search information management table, and the original information file 31
, And whether the search file 32 is generated based on the keyword dictionary 34 or the keyword from the original information file 31 without using the keyword dictionary 34 for full-text search. Is extracted and stored as information on whether the data is generated. FIG. 3 shows the contents of the search information management table 35.

In the present invention, the search file 32 is searched by the search file generation means 44 from the original information file 31.
Is generated, information on the generation time of the search file 32 is stored in the generation time calculation table 33. Specifically, the size of the original information file 31 and the original information file 3
1 and the size ratio of the search file 32, and the time required to generate the search file 32 are stored.

How the generation time calculation table 33 is updated will be described with reference to FIGS. 4, 5, and 6. FIG. FIG. 4 shows, for each of the original information files 31 of (1) to (15), the size of the original information file 31, the size of the search file 32, the ratio of the size of the original information file 31 to the size of the search file 32, It is the result of actually measuring the time required to generate the search file 32. However, this is a search file 32 for full-text search generated without using the keyword dictionary 34.

[0011] For example, the Paso-tsu conference room (1)
The original information file 31 is 481 KB, and when the search file generating means 44 is activated, the 1164 KB search file 32 is generated in 10 minutes, and the original information file 3
This indicates that the ratio of the size of 1 to the size of the search file 32 was 2.4. Also, from this figure, the original information file 31 having similar contents
This indicates that the ratio of the size of the search file 32 to the size of the search file 32 is substantially equal.

FIG. 5 shows a generation time calculation table 33 which stores information relating to the generation of the search files (1) to (15) shown in FIG. As described above, this is the full-text search file 32 generated without using the keyword dictionary 34. The generation time calculation table 33 shown in FIG. 32
It is for. When the search file 32 is generated using the keyword dictionary 34 and when the keyword dictionary 34 is not used, the generation time of the search file 32 is different, and the generation time of the search file 32 using the keyword dictionary is different. In order to obtain the value, it is necessary to prepare another generation time calculation table 33.

The generation time calculation table 33 shown in FIG.
Will be described using the example of FIG. For example,
The original information file 31 in FIG. 4A has a file size of 481 KB, and the ratio of the size of the original information file 31 to the size of the search file 32 is 2.4. The time (10 minutes) required for generating the search file 32 is stored in the column (d) where the ratio of the file size is 2.3 to 2.4. In this column (d), the maximum value and the minimum value of the time required are stored. In this case, since only one example is used, the maximum value,
The minimum value stores 10 minutes. Update of each column of (a) to (j) of the generation time calculation table 33 of FIG. 5 is performed by updating column (a) in FIG. 4 (4), column (b) in (11),
Column (c) is (7), column (d) is (1), column (e) is (2), (12) and (14), column (f) is (6), and column (g) is ( 5), column (h) is (13), column (i) is (1)
5), column (j) is based on (3). The time required to generate the search file 32 is stored in the generation time calculation table 33 according to the following rules.

When the generation time calculation table 33 is updated with the obtained generation time, if no value is stored in the column to be updated, the obtained generation time is stored as the minimum value and the maximum value. If the obtained generation time is the minimum value in the column, the minimum value is updated. If the obtained generation time is the maximum value in the column, the maximum value is updated.

If the obtained generation time is within the range between the maximum value and the minimum value in the column, no update process is performed.

FIG. 6 is a flowchart of a process of generating the search file 32 according to one embodiment of the present invention. In synchronization with the generation of the search file 32 by the search file generation unit 44, a process of obtaining data relating to the generation time of the search file 32 and a process of updating the generation time calculation table 33 based on the data are performed.

First, immediately before generating the search file 32, a process of obtaining a start time of the generation process is performed.
After the process of generating the search file 32, a process of obtaining the end time of the process of generating the search file 32 is performed. The time required to generate the search file is obtained from the difference between the start time and the end time of the generation processing. Regarding the generation of the search file 32, in the case of the keyword search method, the search file 32 of the original information file 31 is generated based on the keyword dictionary. In the case of the full-text search method, a keyword is extracted from the data stored in the original information file 31 through pattern extraction and the like, and a search file 32 is generated.

In this embodiment, the search file 3
If the generation time of No. 2 is less than 3 minutes, the subsequent process of updating the generation time calculation table 33 is not performed.

If the generation time is 3 minutes or more, the sizes of the original information file 31 and the search file 32 and the ratio thereof (search file size / original information file size)
Is performed. Then, the original information file 31
The corresponding column of the generation time calculation table 33 is specified based on the size of the data and the above ratio. This column is for search file 3
2 has an area for storing data of the minimum value and the maximum value of the generation time. If no data is stored in the column obtained above, the generated generation time is stored in the minimum value and maximum value columns. If the calculated generation time is smaller than the already stored minimum value, the minimum value is updated, and if it is larger than the stored maximum value, the maximum value is updated.

FIG. 7 shows a flow of processing for calculating the generation time of the search file 32. Search file 32
Before the generation time calculation processing of the search information management table 3, the original information file 31 to be searched by the operator is specified.
A check is made to see if the information has been registered in 5. Not being registered means that the search file 32 corresponding to the original information file 31 has not been generated. When the search file 32 is registered, the original information file 31 and the creation date and time of the search file 32 are compared. If the original information file 31 is newer than the creation date of the search file 32, the original information file 31 has been updated after the search file 32 was generated. , You will not be able to do the right search. Also, a search file 32 generated based on the keyword dictionary 34
About the search file 32 and the keyword dictionary 34
Also, if the keyword dictionary 34 is newer than the creation date of the search file 32, the keyword dictionary 34 is updated after the search file 32 is generated. Must be done.

When it is determined that the search file 32 needs to be generated, first, the generation time calculation table 33 is read. Then, the file size of the original information file 31 is read, and it is checked whether it is larger than 100 KB. 10
If it is smaller than 0 KB, generation of the search file 32 does not take much time. Therefore, the generation time calculation process is not performed, and the operator is notified that generation of the search file 32 does not take much time. For example, as shown in FIG. 4, in the case of an original information file (9) “journal reference list (part 3)” of about 100 KB, the generation time of the search file 32 is one minute, and when will it end? It is a time when the operator is not so anxious.

The file size of the original information file 31 is 1
If the data exceeds 00 KB, only the first 100 KB of the original information file 31 is used as sampling data to generate a search file 32. Then, the file size of the generated search file 32 and 100 KB
A process is performed to obtain a comparison ratio which is a ratio with respect to (the sampling data of the original information file 31).

Based on the comparison ratio obtained above and the file size of the original information file 31, a corresponding column of the generation time calculation table 33 is determined. If data of the generation time is stored in that column, The data is notified to the operator as the generation time of the search file 32. If the data regarding the generation time is not stored in the obtained field, the generation time calculation table 33 is compared with the file size of the original information file 31 at the same comparison ratio in the range of ± 100 KB. It is checked whether or not there is any data related to the generation time, and if so, the value is notified to the operator as the generation time of the search file 32. If there is still no corresponding data, the comparison ratio is expanded to a range of ± 0.5, and it is checked whether there is a column in which data relating to the generation time is stored. Is notified. If not, the operator is notified that calculation is impossible due to insufficient data.

In the above, the original information file 31 to be searched is
Are sampled to obtain a comparison ratio which is a ratio of the file size of the generated search file 32 to the original information file 31. However, from the original information file 31 having similar contents, Since the search file 32 is generated, by defining the type of the original information file 31 in advance, it is possible to omit the process of sampling and obtaining the comparison ratio.

Hereinafter, the above flow will be described using a specific example with reference to FIG. (1) When there is a corresponding column in the generation time calculation table 33 Example 1: File size of original information file 31 = 500K
B, ratio for comparison = 2.4 (d) column is the corresponding column, and both the minimum value and the maximum value are 10
Minute, the notification is “10 minutes”.

(2) When there is no corresponding column in the generation time calculation table 33 A column in the range of ± 100 KB of the file size of the original information file 31 is checked at the same comparison ratio. Example 2: File size of original information file 31 = 700K
B, comparison ratio = 2.4-100 KB range: 17-18 minutes [file size =
621 to 640 KB] +100 KB range: 24 to 24 minutes [file size =
741 to 760 KB] As described above, the information notified to the operator is “18 to 24 minutes”.
Becomes Example 3: File size of original information file 31 = 430K
B, ratio for comparison = 2.3−100 KB range: none + 100 KB range: 10 to 10 minutes [file size =
481 to 500 KB] As described above, the information notified to the operator is “10 minutes or less”. Example 4: File size of original information file 31 = 350K
B, comparison ratio = 2.4-100 KB range: 5 to 5 minutes [File size =
301-320 KB] +100 KB range: none As described above, the information notified to the operator is “5 minutes or more”.

As in Example 3, when there is no corresponding data in the range of -100 KB and there is data in the range of +100 KB, it is not "10 minutes" but "10 minutes or less".
This is because the larger the file size, the longer it takes to generate the search file 32.

Example 5: File size of original information file 31 = 350K
B: Comparative ratio = 3.0 -0.5 range: none +0.5 range: 9 to 9 minutes [Comparative ratio = 3.
As described above, the information notified to the operator is “9 minutes or less”.

As in Example 5, if there is no data corresponding to the range of -0.5 and there is data corresponding to the range of +0.5, it can be said that "9 minutes or less" is used instead of "9 minutes". The fact that the comparison ratio is large means that the number of words included in the original information file is large, and that it takes time to generate the search file 32.

After notifying the operator of the generation time of the search file 32, the search file 32 is generated and a high-speed search is performed. Alternatively, the search file 32 is not generated and the original information file 31 is generated. An inquiry is made to the operator as to whether to perform a normal search for direct access.
Then, a search method is selected according to the operator's input in response to the inquiry.

[0031]

Since the operator performing the information search is notified of the time until the search file is generated, whether the operator intends to perform the search once or changes the keyword many times. It is possible to determine whether a search file should be generated depending on whether a search is to be performed. Further, since the time when the generation of the search file is completed can be known, another effect can be obtained by that time.

[0032]

[Brief description of the drawings]

FIG. 1 is a block diagram of the principle of the present invention.

FIG. 2 is a configuration diagram of one embodiment of the present invention.

FIG. 3 shows the structure of a search information management table.

FIG. 4 shows an example of generation time of a search file.

FIG. 5 shows the structure of a generation time calculation table.

FIG. 6 is a flowchart of a search file generation process.

FIG. 7 is a flowchart of a process of calculating a generation time of a search file.

[Explanation of symbols]

 1: input device 2: display device 3: disk device 4: processing device 31: original information file 32: search file 33: generation time calculation table 41: calculation table updating means 42: search method control means 43: generation time calculation means 44: Search file generation means 45: High-speed search means 46: Normal search means

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G06F 17/30

Claims (4)

(57) [Claims] A search file generating means (44) converts a search file (3) from an original information file (31) to be searched.
2) high-speed search means (45) for generating and searching the original information file (31) using the search file (32)
And an ordinary search means (46) for directly accessing and searching the original information file (31). In the information search system, information on the time when the search file (32) is generated from the original information file (31) is stored. A generation time calculation means (33) for calculating a time required to generate a search file (32) of an arbitrary original information file (31) based on the stored generation time calculation table (33) and the generation time calculation table (33); 43) and the time required to generate the search file (32) obtained by the generation time calculation means (43) is output to the display device (2), and the normal search means (46) is output in accordance with the input from the input device (1). ) Or the search file generation means (44) generates a search file (32) and executes the high-speed search means (45). Information retrieval system characterized by comprising a Cormorant search method control unit (42). 2. The information search system according to claim 1, wherein when the search file generating means generates the search file, the search file is generated.
Calculation table updating means (41) for updating the generation time calculation table (33) using data on the generation time of
An information retrieval system comprising: 3. The information retrieval system according to claim 1, wherein the generation time calculation table (33) is used for generating the search information file (32) and the size of the original information file (31). The generation time calculation means (43) generates a search file (3) of the original information file (31) based on the size of an arbitrary original information file (31) and a generation time calculation table.
An information retrieval system characterized by calculating the generation time of 2). 4. The information retrieval system according to claim 1, wherein the generation time calculation table (33) is a ratio of a plurality of original information files (31) and a size of the retrieval file (32). And the size of the original information file (31) and the time required to generate the search file (32). The generation time calculating means (43) is a part of the arbitrary original information file (31). , A search file (32) is generated, a size ratio between a part of the original information file (31) and the generated search file (32) is calculated, and the calculated ratio and the original information file (31) are compared. An information retrieval system characterized by calculating a generation time of a search file (32) of the original information file (31) based on a size and a generation time calculation table.