JP5488271B2 - Search device - Google Patents

Description

  The present invention relates to a search device that searches for a document that matches a search condition input by a user and that the user has access rights to.

  Unlike Internet searches that search public information such as HTML and PDF, a search device that searches documents in the company manages access to groupware and business systems scattered on each server. There is a need to search across various types of documents. In particular, access management is strictly performed for many documents in a company, and it is necessary to avoid displaying documents without access authority in the search result list from the viewpoint of security. Therefore, a search device in a company needs a function of searching for a document that matches a search condition and has an access right to the user.

  A search apparatus that searches documents in consideration of both search conditions and access rights, for example, as described in Patent Document 1, is a method for providing access control information in a search index (hereinafter referred to as a prior confirmation method). For example, as described in Patent Document 2, there are two methods: a method of confirming the access right of each document when a search result is obtained (hereinafter referred to as a real-time confirmation method). .

  Among these, the prior confirmation method has a problem that when the access right of a document is changed, the contents of the changed access right are not reflected in the search result until the search index is updated. This problem is often unacceptable in organizations where access management is strictly performed, such as in-company documents, because documents that cannot be browsed may be displayed in search results even temporarily. For this reason, real-time confirmation methods are exclusively used by companies that place importance on security. The present invention relates to an improvement of this real-time confirmation method.

JP 2005-24258 A JP 2004-185095 A

  In the real-time confirmation method, the presence or absence of the access right is confirmed after a list of documents that match the search conditions is obtained, so that the search response is greatly reduced compared to the prior confirmation method. For example, if it takes 50 ms per document for the process of confirming whether or not there is an access right, if 200 documents are searched, it takes 10 seconds only for the access right confirmation process.

  In Japanese Patent Laid-Open No. 2004-228688, when a predetermined number of documents with access rights are found, a halfway result is returned to the user to improve the experiential search response. However, since the presence or absence of authority is investigated in order from the first document in the search result list, it is necessary to investigate the presence or absence of access rights for a considerable number of documents only by finding a predetermined number of documents. For example, if the probability of a document having an access right is 10%, the access right is confirmed for 100 documents in the search result list in order to find, for example, 10 documents in which the user has the access right. This requires 5 seconds of investigation time.

  The object of the present invention is to solve the above-described problem, that is, in the real-time confirmation method for investigating the presence or absence of authority in order from the first document in the search result list, it takes a long time to find a predetermined number of documents. It is an object of the present invention to provide a search device that solves the problem that it is difficult to improve the general search response.

  A search device according to an aspect of the present invention includes a document search unit that searches for a document that matches a search condition included in a search request input from a search requester, and a first access right check that checks whether or not the access right exists. Means, second access right check means that is less reliable than the first access right check means, but can check the existence of the access right at a higher speed, and the search request for the searched document. Whether or not the user has the access right is examined by the second access right check means, and based on the result of the investigation, the searched document is preliminarily determined according to the possibility of having the access right. Search result classification means for classifying into a plurality of determined groups, and sentences belonging to that group in order from the group with the highest possibility of having access rights Whether or not the search requester has the access right by the first access right check means, and the intermediate result including the document for which the search requester is determined to have the access right And a search result returning means for outputting to.

  According to the present invention, it is possible to reduce the number of documents that have to be investigated before returning the first halfway result in order to confirm the presence or absence of the access right in order from the document having the highest possibility of having the access right. As a result, the experiential search response is improved.

It is a block diagram of a 1st embodiment of the present invention. It is a flowchart which shows an example of the process of the 1st Embodiment of this invention. It is a block diagram of the 2nd Embodiment of this invention. It is a flowchart which shows an example of the process of the 2nd Embodiment of this invention. It is operation | movement explanatory drawing of the search result division | segmentation process part in the 2nd Embodiment of this invention. It is a figure which shows the example of calculation of search response time in the case of performing authority confirmation in an order from the head of a search result list | wrist in the search device of the real-time confirmation system relevant to this invention. It is a figure which shows the example of calculation of the search response time in the case of performing authority confirmation from the group with high possibility of having authority in the search device concerning the 2nd Embodiment of this invention. It is a figure which shows the other example of calculation of the search response time in the case of performing authority confirmation from the group with high possibility of having authority in the search device concerning the 2nd Embodiment of this invention. It is a block diagram of the 5th Embodiment of this invention. It is a block diagram of the 6th Embodiment of this invention.

Next, embodiments of the present invention will be described in detail with reference to the drawings.
[First embodiment]
Referring to FIG. 1, a search device 100 according to the first embodiment of the present invention receives a search request 101 from a search requester, matches a search condition included in the search request 101, and is a search requester. Has a function of searching for a document having an access right from a search target document group (not shown) and outputting the search result 102 to a search requester. The search apparatus 100 includes a document search unit 111, a first access right check unit 112, a second access right check unit 113, a search result classification unit 114, and a search result return unit 115.

  The document search unit 111 has a function of searching for a document that matches a search condition included in the search request 101 input from the search requester. The document search unit 111 notifies the search result classification unit 114 of a list of documents that match the search condition.

  The first access right checking means 112 has a function of checking whether a certain user has an access right for a certain document. The first access right checking unit 112 acquires, for example, the latest access control list (ACL) of the document set in the server storing the document from the server, and acquires the acquired ACL, the user attribute, and the like. The user may be configured to determine whether or not the user can access the document.

  The second access right check unit 113 is a method of determining whether or not a certain user has an access right for a certain document, although it is less reliable than the first access right check unit 112, but is a faster method. It has a function to check with.

  For example, the second access right checking unit 113 collects ACLs of documents to be searched from each server in advance and stores them in a local memory, and based on the stored ACLs and user attributes. The user may be configured to determine whether or not the user can access the document. Even if the ACL is updated on the collection source server, it is not reflected in the ACL on the local memory, so there is a possibility that the access right is checked based on the old ACL. For this reason, reliability becomes low. However, it is not necessary to access the ACL of each server via the network, and the ACL on the local memory is accessed, so that high-speed processing is possible.

  Further, the second access right check unit 113 includes, for example, a cache that holds the past investigation history by the first access right check unit 112, and checks the access right based on the investigation history held in the cache. It may be configured to do. For example, when the search by the first access right check unit 112 reveals that the search requester A has the access right for the document X, for example, data indicating that the search requester A has the access right for the document X is cached. Saved in. The second access right check unit 113 determines that the access right is present based on the data on the cache when searching whether the search requester A has the access right for the document X later. Even if the ACL of the document X is updated, the data on the cache is not updated. Therefore, although the reliability is low, since the data on the cache is accessed, high-speed processing is possible.

  The search result classification unit 114 investigates whether or not the search requester has the access right for the document searched by the document search unit 111 by the second access right check unit 113, and based on the investigation result, It has a function of classifying retrieved documents into a plurality of groups determined in advance according to the degree of possibility of having access rights. For example, the search result classification unit 114 classifies the retrieved documents into either a group of documents determined to have access right by the second access right check unit 113 or a group of other documents. be able to. Moreover, you may make it classify | categorize in any of three or more groups.

  The search result return means 115 determines whether or not the search requester has the access right for the documents belonging to the group in descending order of the possibility of having the access right among the groups classified by the search result classification means 114. Is searched by the first access right checking means 112, and a midway result including a document for which the search requester is determined to have the access right is output to the search requester. For example, when the search requester finds a predetermined number (for example, 10) of documents for which the search requester has access rights, the search result return means 115 returns the search results 102 including those documents to the search requester. Good. In addition, the search result return unit 115 determines that one or more documents having the access right have been stored when the first access right check unit 112 has finished the investigation for all the documents belonging to the group having a higher possibility of having the access right. If found, the search result 102 including those documents may be returned to the search requester.

  Next, the operation of this embodiment will be described.

  When a search request 101 specifying a search condition is input from a search requester through a terminal or input device (not shown) (S11 in FIG. 2), the document search unit 111 of the search device 100 searches a search index (not shown). A document matching the search condition is searched (S12). At this time, the search device 100 having a ranking function assigns a score to the searched document.

  Next, the search result classification unit 114 uses the second access right check unit 113 to check whether or not the search requester has the access right for the document in order from, for example, a document with a good score. Based on the above, the retrieved documents are classified into a plurality of groups determined in advance according to the degree of possibility of having an access right (S13).

  Next, the search result returning unit 115 first determines whether the search requester has the access right for the documents belonging to the group in order from the group having the highest possibility of having the access right. For example, when a predetermined number of documents for which the search requester is determined to have access rights are found, the search result 102 including these documents is output to the search requester as an intermediate result (S14). Subsequent processing is arbitrary. For example, when acquisition of the remaining search results is requested by the search requester, the search result return means 115 includes a document to which the search requester has access right among the remaining search results. Return the search results to the search requester.

  As described above, according to the present embodiment, since the presence or absence of the access right is confirmed in order from the document having the highest possibility of having the access right, the number of documents that must be investigated before returning the first halfway result is reduced. As a result, the search response can be improved.

[Second Embodiment]
[Overview]
In this embodiment, in a system for cross-searching documents in an enterprise where access rights scattered across each server are set, when checking whether there is access rights to the document list of the search results, By using the attribute attribute unit cache (per division, job title, etc.) and checking the authority first for documents that are likely to be authorized, the search response can be improved and the ACL of each document can be immediately reflected. secure.

  For this reason, in this embodiment, the document list of the search result is divided into groups using “possibility of authority” as an index, and authority is checked first from the group having the highest possibility of authority. Then, by returning the authority check result of the document list of the group having a high “possibility of authority” first and displaying it on the screen, the search response is improved. In addition, the authority check result of the document list with a low “possibility of authority” is executed in the background, returned sequentially, and sorted and updated in order of score each time on the screen, gradually refining the ranking accuracy. Go.

  The “possibility of authority” in the present embodiment is a combination of “document ID + authorized (O)”, “document ID + unauthorized (×)” in the cache of the searcher attribute (affiliated department, position, etc.). Information is held in a hash map, the cache for each attribute is confirmed, and it is determined that a document having more cases in which the authority becomes ◯ has a higher “possibility of authority”. Note that an attribute that is a cache unit is an item that can be manually set in advance for the entire system (organization), and is effective when set in a unit in which an access right is likely to be set. For example, in a case where there are many cases in which a document access right is set in units of officers, general managers, general managers, and general employees in a certain company, “title” is added to the attribute that is a cache unit.

[Constitution]
Referring to FIG. 3, the second embodiment of the present invention is an in-company search system for searching for various documents managed by a company in an employee of a specific company, and transmits a document search request. The client 1 for displaying the search result, the search server 2 for receiving a search request and searching for a document, and various servers 3 in the company in which the document and ACL are stored.

  The client 1 is an information processing apparatus used by an employee, for example, a personal computer provided to each employee of a company. Here, the client 1 is assumed to be used by an employee named A. In FIG. 3, only one client 1 is depicted, but when there are a total of n employees, there are n clients corresponding to each employee on a one-to-one basis.

  The client 1 includes a search screen / result display unit 4 and a search request processing unit 5. The client 1 is connected to an input device composed of a keyboard, a mouse, etc., a display device composed of a liquid crystal display, etc., and a storage device (not shown) composed of a memory.

  On the other hand, the search server 2 is constituted by a personal computer or the like, and includes a client request reception / return processing unit 6, a document search unit 7, an index (search index) 8, a search result division processing unit 9, an ACL cache 10, and an ACL check unit 11. The information collecting unit 12 is included. Here, the ACL cache 10 and the search result division processing unit 9 correspond to the second access right check unit 113 and the search result classification unit 114 of FIG. Further, the ACL check unit 11 and the client request reception / return processing unit 6 correspond to the first access right check unit 112 and the search result return unit 115 in FIG.

  Each of these operates as follows.

  The client 1 inputs a search keyword designated by the employee A from the search screen displayed on the search screen / result display unit 4, and transmits a request to the search server 2 via the search request processing unit 5. This request may include information on employee A (for example, employee ID). Thereafter, the search request processing unit 5 returns the document information list, which is the search result from the search server 2, divided into several times with a time difference, so that it is received and transmitted to the search screen / result display unit 4 each time. . Each time the search screen / result display unit 4 receives a document information list as a response from the search request processing unit 5, the search screen / result display unit 4 updates the search result document list (resorts in order of score) and displays it on the screen.

  In the search server 2, the client request reception / return processing unit 6 receives a search request from the client 1 and passes the search request information to the document search unit 7. The document search unit 7 performs a search on a document index 8 scattered on the various servers 3 in the company created in advance, and acquires a document list that matches the search request. The acquired document list is transmitted to the search result division processing unit 9 as a search result list before the ACL check.

  The search result division processing unit 9 compares the ACL cache 10 in the attribute unit of the searcher A with the search result before the ACL check, and uses the “possibility of authority” as an index to determine the search result list before the ACL check. Divide into groups of stages. Details of this group division processing will be described later. Then, the document list of the divided group is transmitted to the ACL check unit 11.

  The ACL check unit 11 performs an ACL check on the various servers 3 in the company in order from the document list of the group having the highest possibility of authorization, and accepts a search result list including only authorized documents as a client request. / Send to return processing unit 6. At this time, for example, when the search result list is divided into three groups, the search result list is transmitted to the client request reception / return processing unit 6 three times. The ACL check unit 11 simultaneously stores the ACL check result in the ACL cache 10 for each attribute of the searcher, or updates the ACL check result with the current ACL check result if the ACL check result is already stored.

  The client request reception / return processing unit 6 transmits to the client 1 each time it receives an ACL checked search result list from the ACL checking unit 11.

  As a process for generating the index 8, the information collection processing unit 12 acquires and indexes the full text information and meta information of documents stored in advance in various servers 3 in the company separately from the search. . Since this process is a heavy process that takes a relatively long time, it is assumed that it is performed automatically every day or every week, such as once at night.

[Operation]
Next, the operation of this embodiment will be described in detail with reference to the block diagram showing the configuration of FIG. 3, the detailed processing flow of FIG. 4, and the group division example of FIG. In the following, the search result division processing unit 9 corresponding to the corresponding part is focused on how quickly the authority is checked in the document list searched by the document search unit 7 and only the authorized document is displayed as a search result on the screen. Details of the operations of the ACL cache 10 and the ACL check unit 11 will be described. Since the information collection unit 12 searches (crawls) various servers 3 to create the index 8 and the document search unit 7 searches for a document from the index 8, it is an existing technology. Omitted. Similarly, a process in which search results are transmitted a plurality of times between the search request processing unit 5 and the client request reception / return processing unit 6 and is updated (resorted in order of score) and reflected on the screen each time. It is already known that several technical possibilities such as Ajax communication by session storage and Cookie transmission method can be realized, and detailed description of this processing is also omitted.

  Further, as a premise of explanation, N is the number of documents in the search result list (total number of search hits) passed to the search result division processing unit 9, and an attribute that is preset as a setting of the entire system (access rights within the organization Let M be the number of units set. For the ACL cache 10, it is assumed that a hash map of “document ID + authority ○ / ×” exists for each attribute. Further, when processing the search request, the search server 2 is assumed to know the attribute of the search requester who issued the search request. Any method can be used for the search server 2 to know the attributes of the search requester. For example, there is a method of acquiring an attribute by searching an employee table holding a correspondence relationship between the employee ID and the employee attribute using the employee ID included in the search request as a key, but the method is not limited thereto.

  Now, referring to FIG. 4, the search result division processing unit 9 checks whether there is an authority from the cache corresponding to the searcher attribute value, one document at a time in descending order of the search result score. Next, as the number of times the authority is in the cache increases, the group is assigned with a higher possibility of authority. This process is repeated until the number of documents of the group with the highest “possibility of authority” reaches the number of search results (display number) that the user wants to display on the screen, or until all the documents having the total number of search hits N have been confirmed. (S21-S28). Finally, the search result division processing unit 9 passes the document list of each group as a result of the division to the ACL check unit 11. In addition, when the number of documents of the group with the highest “possibility of authority” reaches the number of search results (displayed number) that you want to display on the screen, the remaining documents that have not yet been checked for authority are displayed in the background. In this case, the classification process may be continued, or the group may be forcibly classified into any group other than the group having the highest “possibility of authority”.

  In the example of FIG. 5, two preset attributes are “affiliation department” and “position”. The attribute values of the searcher A are “◯ △ division” and “boss”, respectively. For this reason, the search result division processing unit 9 has two ACL caches 10, that is, an ACL cache of “XX division” of “affiliation department” attribute and an ACL cache of “boss” of “position” attribute. Get the cache. Then, the authority of the document is confirmed from the cache in the order of the score of the search result, such as “04856456456”, “49845645645”,..., The first group of “authority in both two caches”, “any one cache In the second group of “right ○, no other cache”, the third group “no cache, or right in one cache ○ but right in the other cache ×”, third group “right in any one cache × The other group is classified into five groups, that is, the fourth group of “no cache” and the fifth group of “authority x in both two caches”.

  Next, the ACL check unit 11 inquires the various servers 3 to check whether or not each document belonging to the group actually has authority in order from the first group having the highest “possibility of authority”. (S29). Processing for inquiring various servers 3 to confirm authority is an existing technology. As a result of the confirmation, the ACL check unit 11 removes unauthorized documents from the search result list, and stores authorized documents in the search result list (S30). The ACL checking unit 11 updates the ACL cache for each attribute if the content of the confirmed authority is different from the ACL cache for each attribute or does not exist in the ACL cache for each attribute (S30). When the authority confirmation and the cache update are executed for all the documents belonging to the group, the search result list is passed to the client request reception / return processing unit 6 (S31). If this process is executed for all groups, the process is terminated (S32).

  In the example of FIG. 5, the ACL check unit 11 confirms whether there is an authority of the document to which it belongs in order from the first group of “authority in both two caches” having the highest “possibility of authority”. Repeat up to the fifth group of “authority x in both two caches” with the lowest “possibility of authority”. The search result list to be returned to the client request reception / return processing unit 6 is returned in five divided times at the timing when the authority confirmation of all documents to which each group belongs is completed.

  Thereafter, as described at the beginning of the present embodiment, the search results are transmitted a plurality of times between the search request processing unit 5 and the client request reception / return processing unit 6, and the screen is re-sorted in order of score each time. The process to be reflected in is performed.

[effect]
According to this embodiment, the following effects can be obtained.

(1) In the method for checking authority in real time (at the time of retrieval), retrieval performance is improved when there is no change from the ACL cache, and retrieval results with a certain degree of order can be obtained even when there is a change from the ACL cache. And the sensation speed (the time from when the search is executed until the search result is displayed on the screen) is improved. The reason is that by checking the authority first from the document that seems to have authority based on the past authority confirmation result, the number of authority confirmation can be reduced compared to the method of simply performing authority confirmation in order of score. It is. In other words, the authority confirmation processing time has been a bottleneck so far, but it can be improved.

(2) Holding the cache in attribute units rather than individual units, and further checking the cache of all attributes of the searcher at the time of search increases the probability of hitting one of the caches (probability of being an authority ○) . For this reason, the case where a search response improves compared with the case where the cache of an individual unit is confirmed increases.

(3) While improving the search performance (sensory speed), even when the ACL is changed, it can be immediately reflected in the search results. The reason is that the authority is checked in real time (when searching).

  Hereinafter, the effect by this embodiment is demonstrated based on the specific example of FIGS. Note that FIG. 6 shows a method for checking authority in the order of scores as described in Patent Document 2, and FIGS. 7 and 8 show a method according to this embodiment.

  In FIG. 6 in which authority confirmation is performed in the order of score, the ACL check processing time is 50 ms per case (external communication via the network occurs, which may take several to several tens of times in some cases), searcher A Assuming that the ratio of the document having the authority of 10% is 10%, the ACL check is performed 100 times assuming a case where 10 search results are displayed. Therefore, it takes 50 ms × 100 times = 5000 ms (5 seconds) until the search result is displayed. A search response generally includes a 3 second rule as a time allowed by the user, which exceeds the index.

  FIG. 7 shows a case where the ACL cache is not changed from the latest ACL. For convenience of explanation, there are two types of groups, “authority ○ for both two caches” and “authority x for both two caches”. Also, 1 ms is required to check one ACL cache (the ACL cache is on-memory hash map data, so high-speed reading is possible). Other conditions are the same as in FIG. The first response is the document list after the ACL check of the search result document of the “authority in both two caches” group. In this case, there is no change from the ACL cache. It becomes a document with. Also, the response time is 50 ms × 10 times + 1 ms × 10 times = 600 ms (0.6 seconds), and the search speed is nearly 10 times regardless of whether the same search results and ranking are obtained as compared with the case of FIG. It can be seen that it has improved. Note that the ACL check processing time of the search result document of the “authority with both two caches ×” group, which is the second response after the first response, is 600 ms + 50 ms × 90 times = 5100 ms (about 5 seconds). Since there are 0 documents with no, there is no update on the search result screen. As a result, the result of the first response becomes the final search result as it is, the 3-second rule can be cleared, and the search performance and the sensation (response) speed are greatly improved.

  FIG. 8 shows a case where the ACL cache has been changed five times compared to the latest ACL. Specifically, the ACL has an authority ○ but the latest ACL has two documents with authority ×, and the ACL cache has an authority. × However, in the latest ACL, there are three documents with authority ○, and other conditions are the same as those in FIG. The processing time of the first response and the second response is the same as in FIG. 7, but the first response results are 8 out of 10 cases, and the remaining 2 cases are displayed on the screen. In addition, 3 results are added in the second response after about 5 seconds, and the correct search result can be obtained by re-sorting and displaying the ranking. That is, although it is less than 10 in the first response, the search result document can be confirmed from the screen halfway, and the response speed is greatly improved, so that the sensory speed increases. Further, by displaying the result of the second response on the screen, it is possible to guarantee the number of final search result documents and the order.

[Third embodiment]
In the previous embodiment, after the search result list is divided into groups, the authorities are sequentially confirmed in the order of groups having a higher “possibility of authority”. On the other hand, in the third embodiment of the present invention, authority confirmation of each group is performed in parallel by multithread processing. In that case, when the CPU to be used is a multi-core that is resistant to parallel processing, further improvement in search performance is expected compared to the previous embodiment.

[Fourth Embodiment]
In the previous embodiment, the attribute that becomes the cache unit is manually designated in advance as the unit for which the access right is likely to be set in the system (organization). On the other hand, in the fourth embodiment of the present invention, this cache unit attribute is not set. In this embodiment, the cache is held in units of searchers (employee ID, etc.), and the divided groups are the first group of “authority in cache ○”, the second group of “not in cache”, and “authority in cache” It becomes a total of 3 groups of the 3rd group of x.

  In this embodiment, compared to the previous embodiment, the granularity of the cache unit becomes an individual and becomes smaller, so the number of cache hits is reduced. For this reason, the degree of improvement in search performance is lower than in the previous embodiment, but on the other hand, there is an advantage that there is no need to manually set an attribute that is a cache unit in advance.

[Fifth Embodiment]
Referring to FIG. 9, the fifth embodiment of the present invention is different from the second embodiment shown in FIG. 3 in that the ACL checked by the ACL check unit 11 does not exist in the various servers 3, and the active directory (Active Directory) It is different in that it is collectively managed on an authentication server 13 such as a server. It should be noted that the present invention can also be applied to a case where ACLs are mixed in various servers 3 and authentication servers 13.

[Sixth Embodiment]
Referring to FIG. 10, the sixth embodiment of the present invention includes an ACL index 10A instead of the ACL cache 10 as compared with the second embodiment shown in FIG. The difference is that the presence or absence of authority is confirmed by referring to the ACL index 10A. The information collecting unit 12 has a function of collecting and indexing ACLs of documents from each server 3 in advance and creating an ACL index 10A.

  The advantage of this embodiment is that, unlike the ACL cache, group splitting is possible even at the first search (before the ACL cache is created in the ACL cache method), so the cache hits in the ACL cache method. It is possible to stably realize a higher-speed search response than the case where there is no such response. On the other hand, the processing time for creating the ACL index 10A by the information collecting unit 12 becomes longer, and the group division processing (processing in FIG. 5) becomes slower than the ACL cache method. The reason is that the ACL cache 10 is hash map data on the memory, whereas the ACL index 10A takes time to access because of the actual file data.

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and various other additions and modifications can be made. For example, in each of the above-described embodiments, the present invention is applied to an in-company search system for searching for documents held by an employee of a specific company, but a specific company group, a specific circle, or a specific organization The present invention can be applied to a search system for searching for documents held by an individual belonging to a specific group such as a specific group. Further, each functional means of the search server and the client constituting the present invention can be realized by a computer and a program in addition to being realized by hardware.

DESCRIPTION OF SYMBOLS 1 Client 2 Search server 3 Server 4 Search screen / result display part 5 Search request process part 6 Client request reception / return process part 7 Document search part 8 Index 9 Search result division | segmentation process part 10 ACL cache 10A ACL index 11 ACL check part 12 Information collection unit 13 Authentication server 100 Search device 101 Search request 102 Search result 111 Document search means 112 First access right check means 113 Second access right check means 114 Search result classification means 115 Search result return means

Claims (10)

A document search means for searching for a document that matches a search condition included in a search request input from a search requester;
A first access right checking means for checking presence or absence of access right;
A second access right check unit that is less reliable than the first access right check unit, but can check the presence or absence of the access right at a higher speed;
The second access right check means investigates whether or not the search requester has access right for the retrieved document, and the retrieved document has access right based on the investigation result. Search result classification means for classifying into a plurality of groups determined in advance according to the degree of possibility,
In order from the group having the highest possibility of having the access right, the first access right check means checks whether or not the search requester has the access right for the documents belonging to the group. A search apparatus comprising: search result return means for outputting a halfway result including a document determined to have an access right to the search requester. The second access right check means includes a cache that holds past investigation results by the first access right check means, and checks access rights based on the investigation results held in the cache. The search device according to claim 1. The second access right check means includes a cache that holds a past investigation result by the first access right check means in an attribute unit in which the access right is set, and for each attribute of the search requester, The search apparatus according to claim 1 or 2, wherein an access right is checked based on an investigation result held in a cache corresponding to the attribute. The search result classification means classifies the documents having access rights for all attributes into a group most likely to have access rights based on the access rights check results for each attribute of the search requester. 4. The document having no access right for all attributes is classified into a group having the lowest possibility of having the access right, and the other documents are classified into one or more intermediate groups. The search device described in 1. When the number of documents belonging to the group most likely to have access right reaches a predetermined number, the search result classifying means removes the remaining unclassified documents other than the group most likely to have access right. The search apparatus according to claim 1, wherein the search apparatus is forcibly classified into any one of the groups. The search result return means returns an intermediate result every time when the investigation by the first access right check means is completed for all documents belonging to one group. The described search device. The search device according to claim 1, wherein the second access authority check unit performs access authority check based on an ALC index collected and indexed in advance. 8. The first access right check means checks whether or not there is an access right based on the latest ALC of the document and the attribute of the search requester. Search device. Compared to the document search means, the first access right check means for checking the presence or absence of the access right, and the first access right check means, the presence or absence of the access right can be checked at a higher speed. A search method executed by a search device having a second access right check means, a search result classification means, and a search result return means,
The document search means searches for a document that matches a search condition included in a search request input from a search requester,
The search result classification means investigates whether or not the search requester has an access right for the searched document by the second access right check means, and the search result is classified based on the search result. Classify documents into groups that are pre-determined according to the degree of access
The search result return means investigates, by the first access right check means, whether or not the search requester has the access right for the documents belonging to the group in descending order of the possibility of having the access right. A search method comprising: outputting an intermediate result including a document determined to have the access right to the search requester to the search requester. Computer
A document search means for searching for a document that matches a search condition included in a search request input from a search requester;
A first access right checking means for checking presence or absence of access right;
A second access right check unit that is less reliable than the first access right check unit, but can check the presence or absence of the access right at a higher speed;
The second access right check means investigates whether or not the search requester has access right for the retrieved document, and the retrieved document has access right based on the investigation result. Search result classification means for classifying into a plurality of groups determined in advance according to the degree of possibility,
In order from the group having the highest possibility of having the access right, the first access right check means checks whether or not the search requester has the access right for the documents belonging to the group. A program for functioning as search result return means for outputting intermediate results including documents determined to have access rights to the search requester.

Images