JP4503379B2 - Search device - Google Patents

Description

  The present invention relates to a search device.

Nowadays, various contents such as documents and images are converted into digital data and stored in a database. A search service such as a search engine is used as means for extracting target data from the enormous amount of accumulated data. The search service requires high-speed search performance for a large amount of data as well as an interactive search function.
There is a parallel search device as means for realizing this. In the processing method using the parallel search device, one database is divided into a plurality of computers and registered, and one partial search result obtained by searching in parallel on each computer (hereinafter referred to as a slave computer) is stored. Computer (hereinafter referred to as a master computer) collects and outputs a final search result.

For example, in the conventional parallel data search processing device disclosed in Patent Document 1, when the search terminal requests search processing or display processing of search results, the search coordinator (corresponding to the master computer) requests the search request or display processing. And a search processing device (corresponding to a slave computer) that requests processing is determined, and a search request or display processing request is issued. Upon receiving the request, the search processing device performs a process such as a search on the divided database assigned to each, and generates a processing result. The search coordinator receives the processing results from each search processing device, collects them, and provides them to the search terminal.
Non-Patent Document 1 displays a list of hits and hit target data only when the number of hits in the search result is equal to or less than a predetermined maximum number of retrievals, and hits when the number of hits exceeds the maximum number of retrievals. A method of displaying only the number of cases is disclosed.

However, in the parallel search device, the partial search result by the slave computer is transferred to the master computer via a communication line such as a network, so if the number of hits obtained by the slave computer increases, it is transferred to the master computer. There is a problem that the amount of data increases, and the search performance is degraded due to the bottleneck of the communication path.
In order to solve such a problem, in the parallel data search processing device disclosed in Patent Document 1 described above, the number of hits and hit target data are set so that the data transfer amount via the communication path is minimized in the search process. The data transfer is limited to the minimum necessary data (for example, titles) for specifying.

JP-A-10-154160 “PATOLIS full-text search service”, PATOLIS News, Japan Patent Information Organization, July 1999, No. 225

However, in the parallel data search processing device disclosed in Patent Document 1, when the search results are displayed, if the number of hits in the obtained search results is large, the amount of data to be transferred increases and the communication path becomes a bottleneck. There is a problem that the search performance deteriorates.
Furthermore, in the method disclosed in Non-Patent Document 1, since the hit data list is not displayed on the master computer when the number of hits exceeds the maximum number of fetches, the data specifying the hit target and the information on the hit data list are stored in the master computer. Although there is no need to transfer to a computer, the prior art has a problem in that useless data transfer is performed.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a search device that can efficiently transfer data between a master computer and a slave computer and can perform high-speed search processing.

The search device according to the present invention is based on a data input unit that acquires search target data in a certain data amount unit, and the number of matches and the number of hits in a matching process using a given search condition for the search target data. , A data matching unit that calculates the total number of matching times that is the total number of matching times, the total number of hit numbers that are the number of hits, and a predetermined calculation formula that uses the total number of matching times and the total number of hits obtained by the matching process as parameters If the predicted hit count is less than a certain threshold, a verification result generation section that generates a hit data list and a verification result transmission section that transmits the hit count and hit data list to the master computer are calculated. One or more slave computers, and a matching result receiving unit that receives the hit count and hit data list from each slave computer If the total number of hits received from each slave computer is less than a certain threshold, the search result generator generates a list of all hit data received from each slave computer, and a search that outputs the total number and all hit data list A master computer having a result output unit, a collation process by the data collation unit, a hit data list generation according to the number of hits by the collation result generation unit, and a hit computer and hit data list master computer by the collation result transmission unit Is repeatedly performed for each search target data in a data amount unit .

According to the present invention, the slave computer calculates the predicted hit number by the predetermined calculation formula using the total number of matching times and the total hit number obtained by the matching process as parameters, and based on the predicted hit number, the master computer Since it is determined whether or not to transmit the hit data list, it is possible to make the data transfer between the master computer and the slave computer more efficient and to enable high-speed search processing.

Hereinafter, various embodiments of the present invention will be described.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of a search system 100 according to Embodiment 1 of the present invention. As shown in the figure, the search system 100 includes a search device 10 and a search terminal device 20, and the search device 10 and the search terminal device 20 are connected via a network 30.
The search device 10 includes a master computer 11, slave computers 12a to 12c, and a LAN switch 13 that connects the master computer 11 and the slave computers 12a to 12c.

FIG. 2 is a block diagram showing a configuration of slave computers 12a-12c according to the first embodiment of the present invention. As shown in the figure, the slave computers 12 a to 12 c include a storage device 121, a data input unit 122, a data collation unit 123, a collation result generation unit 124, and a collation result transmission unit 125. The data input unit 122, the data collation unit 123, the collation result generation unit 124, and the collation result transmission unit 125 represent program modules that operate the processors of the slave computers 12a to 12c. The processor of the slave computers 12a to 12c is configured.
The storage device 121 is a memory of the slave computers 12a to 12c, or an external storage device connected to the slave computers 12a to 12c. Search target data is divided and stored in the storage devices 121 of the slave computers 12a to 12c.
FIG. 3 is a block diagram showing the configuration of the master computer 11 according to the first embodiment. As shown in the figure, the master computer 11 includes a collation result receiving unit 111, a search result generating unit 112, and a search result output unit 113. The collation result receiving unit 111, the search result generating unit 112, and the search result output unit 113 represent program modules that operate the processor of the master computer 11, and these are actually integrated with the processor of the master computer 11. Constitute.

Next, the operation will be described.
When a search inquiry is notified from the search terminal device 20, the master computer 11 of the search device 10 notifies the slave computers 12 a to 12 c via the LAN switch 13 of the inquiry content and the maximum number M of retrieval results (a certain threshold). To do. Here, the master computer 11 analyzes the inquiry content from the search terminal device 20, and the slave computers 12a to 12c are notified of the search conditions converted into information necessary for executing the search process. The maximum number of retrievals M represents the maximum number of retrieval results displayed by the master computer 11.

When the slave computers 12a to 12c receive the search condition and the maximum number M of retrieval from the master computer 11, the slave computers 12a to 12c start the search process. FIG. 4 is a flowchart of the operation of the slave computers 12a to 12c according to the first embodiment.
First, the slave computers 12a to 12c initialize a counter for the total number of hits K in the slave computer (step ST100). Next, the data input unit 122 acquires a predetermined amount (a certain data amount unit) of search target data including one or more verification units from the storage device 121 (step ST101). Here, the collation unit represents, for example, one record of a relational database, and indicates a data unit for performing collation processing. At this time, if it is determined that the data in the storage device 121 has been completed, a completion notification is sent to the master computer 11 (step ST102, step ST112), and the processing of the slave computers 12a to 12c is terminated. If it is not the end of data, a counter of the number of hits NS for the acquired search target data is initialized (step ST103).

  Next, the data collation unit 123 executes collation processing based on the retrieval condition notified from the master computer 11 on the collation unit included in the acquired search target data (step ST104). As a result of the collation processing, when the search condition is hit, 1 is added to the hit number NS counter (step ST105, step ST106), and 1 is added to the total hit number K counter (step ST107). If the search condition is not hit in step ST105, the process proceeds to step ST110.

Next, the collation result generation unit 124 determines whether or not the counter for the total hit number K is equal to or less than the maximum number M (step ST108). Here, if the counter for the total hit number K is equal to or less than the maximum number M, the collation result generation unit 124 generates a list of hit search target data (hereinafter referred to as a hit data list) (step ST109). If the total hit count K counter exceeds the maximum fetch count M, no hit data list is generated and the process proceeds to step ST110.
FIG. 5 is an example of the contents of a search query given to the search device 10, and FIG. 6 is an example of a hit data list generated by the matching result generation unit 124 for this search query.
The search query shown in FIG. 5 is for outputting a document ID satisfying the search condition of address = “Kanagawa” and a name indicating its author to the search target data table. In response to this search query, the verification result generation unit 124 generates a list of document IDs and names (author names) of verification units (records) that satisfy the search conditions, as shown in FIG. Here, three cases of L1, L2, and L3 satisfy the search condition.

Next, collation result transmission section 125 determines whether or not there is a next collation unit (step ST110). If there is a next collation unit, the process returns to step ST104, and collation processing based on the search condition is executed for the collation unit. If there is no next collation unit, the collation result transmission unit 125 transmits the collation result to the master computer 11 (step ST111).
The collation result includes the number of hits NS and the hit data list generated in step ST109. As described above, the hit data list is only when the counter of all hits number K is equal to or less than the maximum number of fetches M. It is created and transmitted to the master computer 11 only in that case. If the counter for the total hit count K exceeds the maximum fetch count M, only the hit count NS is transmitted.
Next, the process returns to step ST101, and the search target data is read from the storage device 121. The process is continued until it is determined in step ST102 that the data is finished.
As described above, the slave computers 12a to 12c repeat the process of performing the collation process in multiple times until the search target data in the storage device 121 is completed, and transmitting the collation result to the master computer 11 each time ( Step ST101 to Step ST111).

FIG. 7 is a flowchart of the operation of the master computer 11 according to the first embodiment.
Master computer 11 initializes a counter for the total number of hits N stored in master computer 11 (step ST200). Next, the collation result receiving unit 111 receives the collation result transmitted from the collation result transmitting unit 125 of the slave computers 12a to 12c in step ST111 of FIG. 4 (step ST201). As described above, the collation result may include only the hit number NS or may include the hit number NS and the hit data list. The verification result receiving unit 111 determines whether or not the received notification is an end notification (step ST202).
If it is determined in step ST202 that the notification is not an end notification, the notified hit count NS is added to the total hit count N counter (step ST203). Next, the search result generation unit 112 determines whether or not the counter for the total number of hits N is equal to or less than the maximum number of extractions M (step ST204). If the counter for the total hit count N is equal to or less than the maximum fetch count M, a hit data list is generated (step ST205). If the total hit count N counter exceeds the maximum fetch count M, no hit data list is generated. Next, the process returns to step ST201, and the same processing is repeated for the next collation result transmitted from the slave computers 12a to 12c.

  If it is determined in step ST202 that it is an end notification, the process proceeds to step ST206, and it is determined whether or not the end notification is received from all the slave computers 12a to 12c. If the end notification has not been received from all the slave computers 12a to 12c, the process proceeds to step ST201 and the process is repeated. If it is determined in step ST206 that the end notification has been received from all the slave computers 12a to 12c, the search result output unit 113 displays the hit data list created by the search result generation unit 112 and the total number of hits N. As a search result (step ST207).

  As described above, according to the first embodiment, the slave computers 12a to 12c perform the transmission of the collation result to the master computer 11 in a plurality of times, and the total hit number counter K exceeds the maximum number of extractions M. After that, since only the hit number NS is transmitted and the hit data list is not transmitted, the data transfer amount to the master computer 11 can be reduced, and the bottleneck caused by the data transfer can be eliminated and the search can be performed. Processing can be speeded up.

Embodiment 2. FIG.
In the first embodiment, the slave computer 12a to 12c determines whether or not to include the hit data list in the collation result transmitted from the slave computers 12a to 12c to the master computer 11.
However, even if the total hit count K of the slave computers 12a to 12c does not exceed the maximum fetch count M, if the total hit count N of the master computer 11 exceeds the maximum fetch count M, the master computer 11 hits Since the data list is not generated, even if the hit data list is transmitted from the slave computers 12a to 12c, the transfer data is wasted.
Therefore, in the second embodiment, the master computer 11 notifies the slave computers 12a to 12c whether or not the hit data list needs to be transmitted.

The operation of the search device 10 according to the second embodiment will be described. The configurations of the search system 100 and the search device 10 according to the second embodiment are the same as those shown in FIGS.
FIG. 8 is a flowchart of the operation of the slave computers 12a to 12c according to the second embodiment of the present invention.
The slave computers 12a to 12c initialize a counter for the total number of hits K in each slave computer (step ST100). Next, the data input unit 122 acquires a certain amount of search target data including one or more verification units from the storage device 121 (step ST101). Here, when it is determined that the data in the storage device 121 is terminated, a termination notification is sent to the master computer 11 (step ST102, step ST112), and the processing of the slave computers 12a to 12c is terminated. If it is not the end of data, a counter of the number of hits NS for the acquired search target data is initialized (step ST103).
Next, the data collation unit 123 executes collation processing based on the retrieval condition notified from the master computer 11 on the collation unit included in the acquired search target data (step ST104). As a result of the collation processing, when the search condition is hit, 1 is added to the hit number NS counter (step ST105, step ST106), and 1 is added to the total hit number K counter (step ST107). If the search condition is not hit in step ST105, the process proceeds to step ST110.

Next, the collation result generation unit 124 checks whether or not a notification (hit data list unnecessary notification) indicating that transmission of the hit data list is unnecessary is received from the master computer 11 (step ST113). If a hit data list unnecessary notification has been received, the process proceeds to step ST110. If the hit data list unnecessary notification has not been received, it is determined whether or not the counter for the total hit count K is equal to or less than the maximum fetch count M (step ST108). Here, if the counter for the total number of hits K is equal to or less than the maximum number of fetches M, the matching result generation unit 124 generates a hit data list (step ST109). If the total hit count K counter exceeds the maximum fetch count M, no hit data list is generated and the process proceeds to step ST110.
In step ST110, the collation result transmission unit 125 determines whether there is a next collation unit. If there is a next collation unit, the process returns to step ST104, and collation processing based on the search condition is executed for the collation unit. If there is no next collation unit, the collation result transmitting unit 125 transmits the collation result to the master computer 11 (step ST111).
Next, the process returns to step ST101, and the search target data is read out. The process is continued until it is determined in step ST102 that the data has ended.

As described above, the slave computers 12a to 12c repeat the process of performing the collation process in a plurality of times until the search target data is completed, and transmitting the collation result to the master computer 11 each time (step ST101 to step ST111). ).
Further, the collation result transmitted from the collation result transmitting unit 125 to the master computer 11 includes the hit count NS and the hit data list generated in step ST109. As described above, the hit data list includes the master computer. 11 is generated only when there is no notification of unnecessary hit data list from 11 and the counter of the total number of hits K is equal to or less than the maximum number of fetches M, and only in that case is transmitted to the master computer 11. When the hit data list unnecessary notification is received from the master computer 11 or when the counter of the total hit count K exceeds the maximum fetch count M, only the hit count NS is transmitted.

FIG. 9 is a flowchart of the operation of the master computer 11 according to the second embodiment.
Master computer 11 initializes a counter for the total number of hits N stored in master computer 11 (step ST200). Next, the collation result receiving unit 111 receives the collation result transmitted from the collation result transmitting unit 125 of the slave computers 12a to 12c in step ST111 of FIG. 4 (step ST201). As described above, the collation result may include only the hit number NS or may include the hit number NS and the hit data list. The verification result receiving unit 111 determines whether or not the received notification is an end notification (step ST202).
If it is determined in step ST202 that the notification is not an end notification, the notified hit count NS is added to the total hit count N counter (step ST203). Next, the search result generation unit 112 determines whether or not the counter for the total number of hits N is equal to or less than the maximum number of extractions M (step ST204). If the counter for the total hit count N is equal to or less than the maximum fetch count M, a hit data list is generated.
If the total hit count N counter exceeds the maximum fetch count M, a hit data list is not generated and a notification that the hit data list need not be transmitted to all slave computers 12a to 12c ( (Notifying hit data list unnecessary)) (step ST208). Next, the process returns to step ST201, and the same processing is repeated for the next collation result transmitted from the slave computers 12a to 12c.

  If it is determined in step ST202 that it is an end notification, the process proceeds to step ST206, and it is determined whether or not the end notification is received from all the slave computers 12a to 12c. If the end notification has not been received from all the slave computers 12a to 12c, the process proceeds to step ST201 and the process is repeated. If it is determined in step ST206 that the end notification has been received from all the slave computers 12a to 12c, the search result output unit 113 displays the hit data list created by the search result generation unit 112 and the total number of hits N. As a search result (step ST207).

  As described above, according to the second embodiment, the master computer 11 notifies the slave computers 12a to 12c of the hit data list unnecessary notification when the hit count N exceeds the maximum fetch count M, and the slave computer Upon receiving the hit data list unnecessary notification, 12a to 12c do not transmit the hit data list to the master computer 11. That is, the slave computers 12a to 12c receive not only the hit data list unnecessary notification from the master computer 11 but also the case where the counter for the total hit number K exceeds the maximum number of fetches M as in the first embodiment. Also, the hit data list is not transmitted to the master computer 11. As a result, the data transfer amount can be further reduced as compared with the first embodiment, and the search process can be further speeded up.

Embodiment 3 FIG.
The first and second embodiments include means for reducing the amount of data transferred from the slave computers 12a to 12c to the master computer 11. In the third embodiment, the processing in the slave computers 12a to 12c is made efficient. Specifically, the efficiency of input / output processing for the storage device 121 is increased.

  The operation of the search device 10 according to Embodiment 3 will be described. The configurations of the search system 100 and the search device 10 according to the third embodiment are the same as those shown in FIGS. The third embodiment is characterized by the operation of the data input unit 122 of the slave computers 12a to 12c. Other collation processing, collation result generation, transmission processing, and processing of the master computer 11 are the same as those of the first embodiment. It is the same.

FIG. 10 is a diagram illustrating an example of search target data stored in the storage device 121. As shown in the figure, the storage device 121 stores tabular data whose column items are document ID, name, address, and gender.
FIG. 11 is a flowchart of the operation of the data input unit 122 according to the third embodiment, which corresponds to the process of step ST101 shown in FIG.
First, the data input unit 122 checks whether or not the counter for the total number of hits K is equal to or less than the maximum number of extractions M (step ST1011).
When the counter of the total hit count K is equal to or less than the maximum fetch count M, the slave computers 12a to 12c need to create a hit data list, so all items necessary for executing the search query are stored from the storage device 121. Obtain (step ST1012). For example, in the case where a search query shown in FIG. 6 is executed as an example, it is necessary to acquire column items of document ID, name, and address.
On the other hand, if the counter of the total number of hits K exceeds the maximum number of fetches M, the slave computers 12a to 12c only need to notify the master computer 11 of the number of hits NS, so items necessary for calculating the number of hits That is, only items necessary for determining the search condition indicated in the search inquiry are acquired from the storage device 121 (step ST1013). For example, in the example of the search query shown in FIG. 6, the column item necessary for determining the search condition is an address, and only this column item needs to be acquired.

  Since the format of data acquired from the storage device 121 differs depending on whether the total hit count K is less than or equal to the maximum fetch count M and the total hit count K exceeds the maximum fetch count M, the data matching unit 123 In step ST104 shown in FIG. 4, collation processing is executed according to this difference.

  As described above, according to the third embodiment, when the counter for the total hit count K exceeds the maximum fetch count M, the data input unit 122 stores only the search target data necessary for calculating the hit count from the storage device 121. Therefore, the amount of data exchanged between the storage device 121 and the data input unit 122 can be reduced, the efficiency of reading data from the storage device 121 is improved, and the processing throughput of the entire search device 10 is reduced. Improvements can be made.

Even if the storage destination of the search target data is not the storage device 121 of the slave computers 12a to 12c but data stored in, for example, a file server or SAN (Storage Area Network) on the network, the data input unit 122 is similarly used. It is possible to narrow down the data to be acquired with, and the improvement effect of search performance can be expected.
FIG. 12 is a block diagram illustrating a configuration example of such a search system 200. The same reference numerals as those in FIG. 1 represent the same components. The slave computers 12 a to 12 c of the search device 10 are connected to the network 40. The storage devices 41a and 41b are storage devices such as file servers connected to the network 40, and store search target data.
Note that the search device 10 of the first embodiment and the second embodiment can be applied to the search system 200.

The third embodiment can also be applied to the second embodiment. In this case, whether or not a hit data list unnecessary notification has been received from the master computer 11 is added as a condition to be determined in step ST1011.
In order to improve the processing throughput of the search device 10, it is also effective to increase the processing efficiency of the CPUs of the slave computers 12a to 12c. However, this can be dealt with by increasing the number of slave computers. Is possible.

Embodiment 4 FIG.
In the fourth embodiment, the slave computers 12a to 12c calculate the predicted hit number P based on the total hit number K sequentially counted, and transmit the hit data list to the master computer 11 using the predicted hit number P. Judge whether or not.

The operation of the search device 10 according to the fourth embodiment will be described. The configurations of the search system 100 and the search apparatus 10 according to the fourth embodiment are the same as those shown in FIGS. The flowchart of the operation of the master computer 11 is the same as that of the second embodiment shown in FIG.
FIG. 13 is a flowchart of operations of the slave computers 12a to 12c according to the fourth embodiment of the present invention. Steps denoted by the same reference numerals as those in FIG. 8 are the same as those in the second embodiment, and thus description thereof is omitted.
When confirming that the hit data list unnecessary notification has not been received from the master computer 11 in step ST113, the collation result generating unit 124 calculates the predicted hit number P (step ST114).
Here, an example of a method for calculating the predicted hit count P will be described. First, the search hit rate R is defined as follows using the number Q of collation processes executed by the slave computers 12a to 12c and the total number K of hits.
R = K ÷ Q (1)
Next, assuming that the total number of search target data is C, the predicted hit number P can be calculated by the following equation.
P = R × C (2)
Note that other methods may be used for calculating the predicted hit count P.

In the following step ST115, the predicted hit number P is compared with the maximum extraction number M, and the total hit number K is compared with the hit number threshold Nt. If the number threshold is Nt or less, a hit data list is generated.
Here, the hit number threshold Nt can be defined by the following equation, for example. However, Nslave is the number of slave computers, and here Nslave = 3.
Nt = P ÷ Nslave (3)

  In addition, regarding the calculation of the predicted hit number P, it is conceivable that a large fluctuation occurs in the statistical accuracy when the number of matching times Q is small. Therefore, in practice, it is preferable not to calculate the predicted hit number P until the number Q of collations exceeds a preset value. Until the number Q of collations exceeds a preset value, the predicted hit number P may be always set to 0, and a hit data list may be generated without fail.

  As described above, according to the fourth embodiment, the slave computers 12a to 12c sequentially calculate the predicted hit number P based on the total hit number K, and the predicted hit number P exceeds the maximum number M of fetches. Since the hit data list is not created when the hit count K exceeds the hit count threshold Nt, it is possible to determine at an early stage whether or not to include the hit data list in the collation result transmitted to the master computer 11. Compared to the second embodiment, the amount of data transmitted to the master computer 11 can be further reduced. Thereby, since the data transfer efficiency is further improved, the search process can be further speeded up.

In the fourth embodiment, the timing of stopping the creation of the hit data list is different from that in the first to third embodiments, and therefore the final hit count and the number of hit data lists do not completely match. However, in the Internet search or the like, the searcher does not always refer to the entire hit data list, so it is considered that the number of hit data lists to be output should be approximately equal to the number of hits.
The fourth embodiment can also be applied to the first embodiment. Further, the third embodiment can be applied to the fourth embodiment.

It is a block diagram which shows the structure of the search system by Embodiment 1 of this invention. It is a block diagram which shows the structure of the slave computer by Embodiment 1 of this invention. It is a block diagram which shows the structure of the master computer by Embodiment 1 of this invention. It is a flowchart of operation | movement of the slave computer by Embodiment 1 of this invention. It is a figure which shows the example of the search inquiry content given to a search device. It is a figure which shows the example of the hit data list produced by the slave computer by Embodiment 1 of this invention. It is a flowchart of operation | movement of the master computer by Embodiment 1 of this invention. It is a flowchart of operation | movement of the slave computer by Embodiment 2 of this invention. It is a flowchart of operation | movement of the master computer by Embodiment 2 of this invention. It is a figure which shows an example of the search object data stored in the memory | storage device. It is a flowchart of operation | movement of the data input part by Embodiment 3 of this invention. It is a block diagram which shows the example of a structure of the search system by this invention. It is a flowchart of operation | movement of the slave computer by Embodiment 4 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 Retrieval device, 11 Master computer, 12a-12c Slave computer, 13 LAN switch, 20 Retrieval terminal device, 30, 40 Network, 41a, 41b Storage device, 100, 200 Retrieval system, 111 Collation result receiving part, 112 Retrieval result generation Unit, 113 search result output unit, 121 storage device, 122 data input unit, 123 data collation unit, 124 collation result generation unit, 125 collation result transmission unit.

Claims (3)

A data input unit for acquiring search target data in a certain data amount unit ;
Based on the number of matching and the number of hits in the matching process using the given search condition, the total number of matchings obtained by counting the number of matchings and the total number of hits obtained by counting the number of hits. A data verification unit to calculate ,
The number of predicted hits is calculated by a predetermined calculation formula using the total number of matching times obtained by the matching process and the total number of hits as parameters, and if the number of predicted hits is less than a certain threshold, a hit data list is calculated. A verification result generation unit to generate,
One or more slave computers having a matching result transmission unit for transmitting the hit number and the hit data list to a master computer;
A matching result receiving unit for receiving the number of hits and the hit data list from each slave computer;
If the total number of hits received from each slave computer is below a certain threshold, a search result generation unit that generates a list of all hit data received from each slave computer;
A master computer having a search result output unit for outputting the total number of records and the list of all hit data ;
The data collation processing, the collation result generation unit generates the hit data list according to the number of hits, the collation result transmission unit sends the hit number and the hit data list to the master computer, the data A search apparatus, which is repeatedly performed for each search target data in a quantity unit . The search result generation unit notifies each slave computer that a hit data list is unnecessary when the total number obtained by sequentially counting the number of hits received from the slave computer exceeds a certain threshold,
The verification result generation unit, when having received the notification, according to claim 1 Symbol placement search system characterized in that it does not generate a subsequent hit data list. If the data input unit is determined not to generate a hit data list in the collation result generation unit, thereafter, only the data items necessary for specifying the number of hits among the search target data are acquired. The search device according to claim 1 or 2 .

Images