JP2005301430A - Distributed retrieval system, retrieval node device and distributed retrieval method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the traffic of a retrieval result to be transferred between retrieval nodes, and to reduce a processing load due to rating or the like in a final retrieval result reception node. <P>SOLUTION: At the time of retrieving resources stored in a plurality of retrieval nodes 30A to 30E connected to a communication network, a value requested by a user at the number of resources as an end batch acquisition result is set in a retrieval request message, and when the retrieval request is received by respective retrieval nodes 30A to 30E, the resource information matched with the retrieval conditions of the retrieval request is retrieved. Furthermore, the retrieval request is transferred through the communication network to another retrieval node, and the rating of the retrieval result received by the other retrieval node and the result retrieved in its own node is operated, and the resource information of the number of resources set in the retrieval request message is returned to the origin of retrieval request as the retrieval result from the order of higher evaluation. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a distributed search system, a search node apparatus, and a distributed search method for searching information such as contents according to a user request by using search nodes distributed and connected to each other on a communication network. About.

2. Description of the Related Art Conventionally, a peer-to-peer (P2P) type search system in which a plurality of search nodes distributed on a network are connected to each other via a communication line and the search nodes cooperate to search for content is known ( For example, Patent Document 1 and Non-Patent Document 1). This P2P type search system is shown in FIG.
In such an architecture, there is no need to install a server because the search nodes directly exchange information without using a server. In addition, since resource information (also simply referred to as resources) that increases explosively is distributed and processed, there is an advantage that there is scalability compared to a client-server type search system.

In the P2P type search system, when a search request occurs, the search request messages are sequentially transferred between search nodes connected to the P2P network, and the resource information held in the nodes in each search node that has received the search request message. Thus, a search using a plurality of search nodes is realized by searching for resources to be searched.
As such a P2P type search method, there is a method of sequentially transferring search request messages between search nodes until a corresponding search result is obtained or another search node is exhausted, or there is a search request message on the P2P network. There is a method in which a search result is returned to a search request source from a search node holding a corresponding resource in the range.

A search method in such a P2P type search system will be described with reference to FIG.
As illustrated in FIG. 7, the node 10A that requests acquisition of a resource transmits a search request message to the adjacent nodes 10B and 10C. The nodes 10B and 10C that have received the search request message check whether or not the requested resource is held by the own node. As a result, when the corresponding resource is not found, the search request message is transmitted to the adjacent search nodes 10D, 10E, 10F, and 10G other than the search request message transmission source. The search nodes 10D, 10E, 10F, and 10G that have received the search request message perform processing of searching for the own node and transferring the search request message to an adjacent node in the same manner as the search nodes 10B and 10C.

  In this way, each search node that has received the search request message performs a search, and holds resources corresponding to the search request condition. For example, the search node 10D uses the search result as a search response message as a search request destination search node. Return to 10B. This search response message traces the path through which the search request message has propagated in the reverse direction and is delivered to the search node 10A that originated the original search request.

In this method, the search request message is transferred to a plurality of routes until a corresponding search result is obtained. Therefore, the maximum number of search request response messages transmitted to the search request source is the route through which the search request message propagates. This is the number of (search nodes).
As another search method, as shown in FIG. 8, regardless of whether or not a search result is obtained, the search request message is temporarily sent to all search nodes within the range specified by the number of hops or TTL (Time To Live). There is a method of transmitting a search response message to the transmission source of the search request message when each node transmitted and receiving the search request message finds a corresponding resource in its own node.

  That is, as shown in FIG. 8, the node 20A requesting resource acquisition transmits a search request message to the adjacent nodes 20B and 20C, and the nodes 20B and 20C receiving the search request message transmit search request messages. The search request message is transferred to the search nodes 20D, 20E, 20F, and 20G connected to other than the original, and finally search is performed for all the search nodes within the number of transfer hops specified as the reach range of the search request message. A request message arrives.

  Then, the search response message is received from the transfer destination of the search request message, the search result of the own node is added to the search result, and the search response message is returned to the transmission source of the search request message. Upon receiving this search response message, the search request source node 20A performs rating (evaluation) on the search results collected from all the nodes based on the policy of the search system, and presents the higher-level resources to the user as the search results.

In any of the above cases, the search results obtained from each search node are collected at the search request source node, and after a plurality of resources obtained as search results are rated, the search results are presented to the user in descending order. To do.
In addition, there are Patent Documents 2 and 3 as distributed search systems. The contents of Patent Document 2 are as follows. When searching a plurality of information distributed on a network consisting of a center that is a search subject and a plurality of nodes connected to the center, the search condition is transmitted from the center together with an information request for all the nodes. After executing the search for each node, only the node having information that matches the search condition is caused to respond to the center.

The contents of Patent Document 3 are as follows. In a distributed system that communicates between nodes connected to a network, this is a distributed processing method that obtains a final processing result by repeatedly combining the results of partial processes executed in a distributed manner at each node. Thus, the partial processing is executed while circulating the nodes between the nodes using the procedure information of the joining processing and the intermediate result information as a transfer information unit. Each node holding this result acquires a transfer information unit from the network, and based on the procedure information of this transfer information unit, the partial processing result held by the node and the intermediate result information of the acquired transfer information unit Using this, a part of the combining process is performed, and according to this result, the previously acquired intermediate result information is updated and transmitted to the network together with the previously acquired procedure information.
Japanese Patent Laid-Open No. 2004-21770 JP-A-5-235594 JP-A-10-27164 Gnutella <URL http: // www. gnutella. com />

By the way, in the P2P type search system of the above-mentioned patent document 1 and non-patent document 1, which are conventional distributed search systems, the search results are all gathered at the node that originally transmitted the search request, and the search results are originally rated. The node that sent the search request.
However, when the number of resources corresponding to the search request condition is very large, there is a problem that it takes a very large load on the node to receive all and process the rating. Also, if the user does n’t need a lot of search results, only the top results of the rated results will be used and the lower results will not be used in many cases. Since all search results are transferred over the network, there is a problem that the traffic becomes high.

In the systems of Patent Documents 2 and 3, ratings cannot be performed. In Patent Document 3, the result of partial processing executed in a distributed manner at each node is transferred and coupled between nodes. Because it repeats, there is a problem that traffic becomes high.
The present invention has been made in view of such a problem, and can reduce the traffic of search results transferred between the search nodes, and can also perform processing by ratings and the like in the final search result receiving node. An object of the present invention is to provide a distributed search system, a search node device, and a distributed search method that can reduce the load.

  To achieve the above object, a distributed search system according to claim 1 of the present invention is a distributed search system for searching resource information stored in a plurality of search node devices distributed and connected to a communication network. A resource to be returned as a response to the search request source from among the search conditions for the resource information, the transfer range of the search request information, and the search results that match the search conditions, to the search request information for making a search request for the resource information The search node device that sets the number of information and receives the search request information searches for resource information that matches the search conditions of the search request information, and also searches the received search request information according to the transfer range to other search nodes. Transfer to the device, and after this transfer, rate the search results received from other search node devices and the search results in the local search node device Thus with the evaluation rank, the resource information of the search results in descending order of the ranking to select only the number is characterized by the return as a search result the selected resource information to the search request source.

  According to this configuration, only the necessary information can be sent to the communication network by performing the rating processing of the search results in each search node device that has performed the search, and returning only the upper search results to the search request source. The traffic of the communication network can be reduced. Furthermore, since the load of the search processing rating processing is distributed to each node device, the concentration of processing load on the final search result receiving node device can be reduced.

  According to a second aspect of the present invention, there is provided a search node device that is distributedly connected to a communication network and stores resource information to be searched in response to a search request. First receiving means for receiving search request information in which a transfer range of request information and the number of resource information to be returned as a response to a search request source from search results that match the search condition are set; A first transmission means for transferring the search request information received by one receiving means to another search node device according to a transfer range set in the search request information; and a search request received by the first reception means Search means for searching for resource information that matches the search conditions set in the information, and search results searched by other search node devices based on the search request information sent from the first sending means A second receiving means for receiving, an evaluation means for rating a search result in another search node device received by the second receiving means and a search result in the own search node device, and a rating in the evaluation means The search result processing means for selecting the resource information of the search results in the descending order according to the number of resource information set in the search request information, and searching the resource information selected by the search result processing means to the search request source And a second transmission means for returning the result.

  According to this configuration, only the necessary information can be sent to the communication network by performing the rating processing of the search results in each search node device that has performed the search, and returning only the upper search results to the search request source. The traffic of the communication network can be reduced. Furthermore, since the load of the search processing rating processing is distributed to each node device, the concentration of processing load on the final search result receiving node device can be reduced.

  The search node device according to claim 3 of the present invention further comprises storage means for storing the search results subjected to rating processing by the search result processing means according to claim 2, wherein the search result processing means includes the first search result processing means. When a search result addition request information is received from the search request source with the same search condition by the receiving means, a search for other search node devices within the transfer range set in the received additional search request information is performed. When the result is stored in the storage unit, the stored search result is processed so as to be returned from the second transmission unit to the search request source, and the search result of the other search node device is stored in the storage unit. If it is not stored, the acquisition range of the number of resource information that is insufficient is set, and a search request is made to another search node device again using the search request information that has been set. To have.

  According to this configuration, the search result once rated is stored, and when there is a search result addition request under the same search condition from the search request source, it is set in the additional search request information. If the search result of another search node device within the transfer range is stored, this search result is returned to the search request source. If it is not stored, the acquisition range of the number of the missing resource information is set in the search request information, and a search request is made again to another search node device using this search request information. Since the search result obtained by performing the rating process using the search result obtained by this request is returned to the search request source, transmission of useless information can be eliminated.

  According to a fourth aspect of the present invention, there is provided a distributed search method for searching for resource information stored in a plurality of search node devices distributed and connected to a communication network. Set the search conditions for resource information, the transfer range of the search request information, and the number of resource information to be returned as a response to the search request source from the search results that match the search conditions The first step of performing, the second step of receiving the search request information set in the first step, and the search request information received in the second step are set in the search request information A third step of transferring to another search node device according to the transfer range, and searching for resource information that matches the search conditions set in the search request information received in the second step A fourth step, a fifth step for receiving a search result searched by another search node device based on the search request information transferred in the third step, and another received in the fifth step. A sixth step for rating a search result in the search node device and a search result in the own search node device, and resource information of the search result in descending order according to the rating in the sixth step, the search request information And a seventh step of selecting the number of resource information set in (8) and an eighth step of returning the resource information selected in the seventh step as a search result to the search request source. .

  According to this method, only the necessary information can be sent to the communication network by performing the rating processing of the search results in each search node device that has performed the search, and returning only the upper search results to the search request source. The traffic of the communication network can be reduced. Furthermore, since the load of the search processing rating processing is distributed to each node device, the concentration of processing load on the final search result receiving node device can be reduced.

  As described above, according to the present invention, it is possible to reduce the traffic of search results transferred between search nodes, and to reduce the processing load due to ratings and the like in the final search result receiving node. There is an effect that can be done.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(Embodiment)
FIG. 1 is a diagram showing a configuration of a distributed search system according to an embodiment of the present invention.
A distributed search system 1 shown in FIG. 1 is a system that performs a search using search nodes (search node devices) 30 (30A to 30G) that are distributed and connected to each other on a communication network such as the Internet. In each drawing of this specification, a search node is simply described as a node, and may be expressed as a node in the text.

FIG. 2 is a block diagram showing a configuration of a search node represented by the search node 30B.
As shown in FIG. 2, the search node 30B includes a search request message reception unit (first reception unit) 301, a search request message transmission unit (first transmission unit) 302, and a search processing unit (search unit). 303, a search result evaluation unit (evaluation unit) 304, a resource database 305, a search result processing unit (search result processing unit) 306, a search result cache unit (storage unit) 307, and a search response message transmission unit (first 2 transmission means) 308 and a search response message reception unit (second reception means) 309.

  The search request message receiving unit 301 is a module connected to the communication interface, and a search request addressed to the own node 30B transmitted via the communication network from another search node (here, the search node 30A) that requests the search. Receive a message. In the received search request message, a search condition, a range in which the search request message is transferred (number of transfer hops), and a range of the number of resources requested by the user as a search result are set.

Also, the search request message receiving unit 301, when the search range of the received search request message includes a search node ahead of its own node 30B starting from the search request message transmission source, The message is sent to the message transmitter 302.
The search request message transmission unit 302 transfers the search request message to the search nodes (in this case, the search nodes 30D and 30E) to which the own node 30B is connected other than the search request message transmission source search node.

The search processing unit 303 outputs a search result obtained by searching the resource database 305 based on the search condition described in the search request message from the search request message receiving unit 301 to the search result evaluating unit 304.
The search result evaluation unit 304 evaluates (ratings) the search result resource obtained by the search processing unit 303 according to the degree of conformity to the search condition that is the policy of the distributed search system 1 and the popularity of the resource. And the evaluation value of the resource is output to the search result processing unit 306 together with information about the resource (for example, URL).

The search response message receiving unit 309 receives the search result for the search request message transferred to the adjacent node as a search response message, and outputs the search result based on this search response message to the search result processing unit 306.
The search result processing unit 306 merges the search result by the search response message from the search response message receiving unit 309 with the search result in its own node, and sets the order based on the evaluation value. The higher-order search results that fall within the specified range are output to the search response message transmission unit 308. The merged search results are stored in the search result cache unit 307 for a certain period.
The search response message transmission unit 308 transmits a search result within the range requested in the search request message to the search node that is the search request source as a search response message.

Next, an operation when information search processing according to a request is performed by the search node having such a configuration will be described with reference to a flowchart shown in FIG.
In step S1, for example, when the search node 30B receives a search request message from the search node 30A in the search request message receiving unit 301, in step S2, the search request message is displayed in the search node 30B under the same conditions. It is determined whether or not it is a search request message when an additional request is made. This determination is No in the first search.

  In addition, the search request message reception unit 301 confirms the transfer range set in the received search request message, so that the search node ahead of the own node 30A is included in the transfer range from the search request message transmission source. In step S3, the search request message is transferred to the adjacent nodes 30D and 30E other than the search request message transmission source.

Next, in step S4, the search processing unit 303 searches the resource database 305 for a resource corresponding to the search request condition in the node 30A, and outputs the search result resource to the search result evaluation unit 304.
In step S5, the search result evaluation unit 304 performs a rating on the resource of the search result according to the degree of conformity to the search condition that is the policy of the distributed search system 1, the popularity of the resource, and the like. Is output to the search result processing unit 306 together with the information of the resource.

  Since the search request message is transferred to the adjacent nodes 30D and 30E in step S3, the search response message receiving unit 309 receives the search response message from the adjacent nodes 30D and 30E in step S6. The received search response message is output to the search result processing unit 306. In step S7, the search result processing unit 306 merges the search result by the received search response message with the search result of the own node 30A.

  The merged search results are ranked based on the evaluation values, and stored in the search result cache unit 307 for a certain period in step S8. In step S9, the ranked search results are transmitted to the search request source by the search response message transmission unit 308 as a search response message in order from the upper search result within the range specified by the search request message. .

An example of the search result cached in the search result cache unit 307 in step S8 is shown in FIG. As shown in FIG. 4, in the search result, each resource name is ranked and arranged, and the evaluation value is associated with the acquisition destination and the rank at the acquisition destination.
If it is determined in step S2 that the search node 30B has received a search request message that additionally requests a search result under the same conditions, the search result processing unit 306 searches the search result cache unit 307 in step S10. As a result, the search response message is returned to the search request source using the cached search result.

  When this is done, in step S11, the search result processing unit 306 refers to the cached search results and whether the lowest rank of the search results acquired from the other nodes 30D and 30E is included in the requested rank. Judge whether or not. To further explain, there is a possibility that the result of highly evaluated resources may be obtained from the other nodes 30D and 30E. Judge whether it is higher than lower.

  If the determination result is Yes, it is determined in Step S12 whether or not the adjacent nodes 30D and 30E further hold the corresponding result. If the result is Yes, the adjacent nodes 30D and 30E are determined in Step S13. Send a search request message to. In step S14, a search response message corresponding to the search request message is received from the adjacent nodes 30D and 30E. In step S15, the search results are merged, and the process proceeds to step S8.

That is, a search request message is transmitted by setting a request range of search results that may be included again as resource information for the nodes 30D and 30E that have returned the search results, and additionally acquiring search results from other nodes. .
In the case of the example shown in FIG. 4, among the results returned as a search response, there is a possibility that a better evaluation result can be obtained than the node 30D as a resource having a rank of 22nd or higher. Set to and send search request.

FIG. 5 shows an example of a search procedure exchanged between the search nodes 30A to 30E in a situation where a plurality of search nodes 30A to 30E are distributed and arranged in the configuration of the distributed search system 1 shown in FIG. .
FIG. 5 shows a state in which the node 30A is instructed to return the search results for the top 10 out of resources under a certain condition from the user in step S21.

In step S22, the node 30A transmits a search request message to the adjacent nodes 30B and 30C. Also in the nodes 30B and 30C, when the search request message is received in step S23, the search request message is transferred to the adjacent nodes 30D and 30E.
In each of the nodes 30A to 30E, in steps S24 to S28, a resource that matches the search request is searched from the resource information stored in the own node, and in steps S29 to S33, how much is determined based on the searched resource. Whether each node is suitable is rated according to the same standard for each node, and the resource order is set.

In this example, it is assumed that the numbers of corresponding resources held in the nodes 30A to 30E are 40, 5, 20, 30, and 70, respectively.
First, in step S34, the node 30C returns the top 10 search results as a search response message together with the evaluation value to the node 30A together with the search result of the node itself. Similarly, in step S35, the node 30E also returns the top 10 search results as a search response message to the node 30B together with the evaluation value from the search result of its own node. In step S35, the top 10 search results are returned from the node 30D.

  The node 30B merges the search results obtained from the node 30D and the search results obtained from the nodes 30D and 30E in step S37, and returns 10 highly evaluated items to the node 30A as search responses in step S38. The node 30A also performs the same merge process in step S39, and finally, in step S40, the top 10 search results are presented to the user from the node 30A.

FIG. 6 shows an example of a procedure in which the user performs additional acquisition of search results under the same conditions again following the series of searches in FIG.
In step S41, if there is an additional request for search requests 11 to 20 from the user, in step S42 the node 30A checks the cached results, and the rank of the search results within the requested range is all nodes. If the result is reflected, in step S43, the cached result is presented to the user.

  Next, in step S44, when a request for additional presentation of search results 21 to 30th is generated again, in the cache confirmation of the node 30A in step S45, the resource of the search result from the node 30B is further requested. If there is a possibility that the results in the range remain, in step S46, an additional search request for the search results (11th to 20th ranks) is made to the node 30B.

  Next, in step S47, the node 30B similarly checks the cache in its own node, and if there is a possibility that the search result remains in the range further requested by the node 30D, in step S48, the node 30D An additional search request for the search results (11th to 20th) is made. In step S49, the node 30D confirms the cache, and if the search results 11 to 20 remain in the range further requested by the node 30D, the search results (11 to 20) are stored in step S50. To the node 30B.

The node 30B merges the search result in its own node in step S51 with the search results (11th to 20th) acquired from the node 30D, and in step S52, the 10 highly evaluated (11th to 20th) It returns to 30A as a search response. The node 30A also performs the same merge process in step S53, and finally, in step S54, the search result (21st to 30th) is presented to the user from the node 30A.
In this way, the cache is checked, and if it is determined that the results of all the nodes 30A to 30E are not reflected as a result, the adjacent node is inquired again.

  As described above, according to the distributed search system 1 of the present embodiment, it is possible to perform a search in a distributed environment, propagate the result, and minimize unnecessary traffic. Also, when the search result rating process is performed at the search node that lastly obtained the search result, the traffic flowing on the network is reduced and the load of the search result rating process is distributed in a distributed manner to each search node. Thus, it is possible to eliminate the concentration of processing load at the final search result receiving node.

It is a figure which shows the structure of the distributed search system which concerns on embodiment of this invention. It is a block diagram which shows the structure of the search node of the distributed search system which concerns on the said embodiment. It is a flowchart for demonstrating the operation | movement at the time of performing the search process of the information according to a request | requirement by the search node of the distributed search system which concerns on the said embodiment. It is a figure which shows an example of the search result cached in the search result cache part of the search node of the distributed search system which concerns on the said embodiment. It is a sequence diagram for demonstrating the search procedure exchanged between each search node of the distributed search system which concerns on the said embodiment. It is a sequence diagram for demonstrating the procedure in which a user performs additional acquisition of a search result on the same conditions again by the search between each search node of the distributed search system which concerns on the said embodiment. It is a figure which shows the structure of the P2P type | mold search system which is the conventional distributed type search system. It is a figure which shows the structure of the other conventional distributed search system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Distributed type search system 30A, 30B, 30C, 30D, 30E, 30F, 30G Search node 301 Search request message receiving part 302 Search request message transmission part 303 Search processing part 304 Search result evaluation part 305 Resource database 306 Search result processing part 307 Search result cache unit 308 Search response message transmission unit 309 Search response message reception unit

Claims (4)

In a distributed search system for searching resource information stored in a plurality of search node devices distributed and connected to a communication network,
Resource information to be returned as a response to the search request source from the search conditions for the resource information, the transfer range of the search request information, and search results that match the search conditions, to the search request information for making a search request for the resource information And set the number of
The search node device that has received this search request information searches for resource information that matches the search conditions of the search request information, and forwards the received search request information to another search node device according to the transfer range,
After this transfer, the search results received from other search node devices and the search results in the self-search node device are rated to rank the evaluation, and the resource information of the search results is selected by the number in the descending order. A distributed search system characterized in that the selected resource information is returned as a search result to the search request source. In a search node device that is distributedly connected to a communication network and stores resource information searched according to a search request,
Receiving search request information in which a search condition for resource information, a transfer range of the search request information, and the number of resource information returned as a response to the search request source from search results that match the search condition are received. First receiving means;
First transmission means for transferring the search request information received by the first reception means to another search node device according to a transfer range set in the search request information;
Search means for searching for resource information that matches a search condition set in the search request information received by the first receiving means;
Second receiving means for receiving a search result searched by another search node device based on the search request information transmitted from the first transmitting means;
Evaluation means for rating the search result in the other search node device received by the second reception means and the search result in the own search node device;
Search result processing means for selecting the resource information of the search results in descending order according to the rating in the evaluation means, as many as the number of resource information set in the search request information;
A distributed search system comprising: second transmission means for returning the resource information selected by the search result processing means as a search result to a search request source. A storage means for storing the search results rated by the search result processing means;
The search result processing means is set in the received additional search request information when the first reception means receives information on a search result addition request from the search request source under the same search condition. When the search result of another search node device within the transfer range is stored in the storage unit, the stored search result is processed so as to be returned from the second transmission unit to the search request source. If the search result of the search node device is not stored in the storage means, set the acquisition range of the number of missing resource information, and search again to another search node device using the search request information for which this setting has been made The search node device according to claim 2, wherein a request is made. In a distributed search method for searching resource information stored in a plurality of search node devices distributed and connected to a communication network,
Resource information to be returned as a response to the search request source from the search conditions for the resource information, the transfer range of the search request information, and search results that match the search conditions, to the search request information for making a search request for the resource information A first step of setting the number of
A second step of receiving the search request information set in the first step;
A third step of transferring the search request information received in the second step to another search node device according to a transfer range set in the search request information;
A fourth step of searching for resource information that matches a search condition set in the search request information received in the second step;
A fifth step of receiving a search result searched by another search node device based on the search request information transferred in the third step;
A sixth step of rating the search result in the other search node device received in the fifth step and the search result in the own search node device;
A seventh step of selecting resource information as a search result in the descending order according to the rating in the sixth step by the number of resource information set in the search request information;
And an eighth step of returning the resource information selected in the seventh step as a search result to the search request source.

Images