JP2940459B2 - Node / link search device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a node / link search apparatus for finding a range of nodes to be searched under a condition specified in a database composed of nodes and links, and more particularly to a distributed hypermedia system. The present invention relates to a node link search device used in a database in which nodes are distributed and stored in a plurality of servers on a network.

[0002]

2. Description of the Related Art As one of databases, there is a hypertext system for searching a plurality of nodes storing various information such as a document file by following a link indicating a dependency between the nodes. Such a search operation using hypertext can proceed smoothly by holding the index of the information stored in the nodes and the link relation between the nodes in advance.

Japanese Patent Laid-Open Publication No. Hei 4-321144 discloses a hypertext system capable of displaying a relationship between nodes so that the relationship can be easily grasped. Since each node can be linked to an arbitrary node, a link may be formed in a loop between a plurality of nodes.
In this system, a display screen is created based on a table in which link relationships between nodes are registered in advance. Then, the link in a loop is separated, and the link relation between the nodes is displayed as a tree structure starting from a certain node.

Conventionally, the hypertext system has been constructed on a local workstation or the like, but with the recent development of communication technology, a system in which nodes are distributed and stored in a plurality of servers connected via a network has appeared. doing. Such a system is called a distributed hypermedia system. For example, the World Wide Web (WWW) has attracted attention as a means of transmitting information on the Internet. Hypermedia is hypertext that can handle not only text data such as characters and tables, but also multimedia data such as moving images and audio.

[0005] In the distributed hypermedia system, a table representing a link relationship between nodes is created by acquiring information on nodes stored in each server via a network. Information indicating the link relationship between nodes is called a node / link information database. Also, an apparatus that creates a node link information database will be referred to as a node link search apparatus.

In the World Wide Web (WWW), nodes are distributed among a plurality of server machines on the Internet, and node data stored in each server is stored in a hypertext transfer protocol (HyperText). Transfer Protocol (hereinafter referred to as HTTP). Multimedia documents that are nodes are hypertext markup languages.
It is expressed as HTML below. ) Is described in a hypermedia description language format.

The node / link searching device analyzes the contents of the node described in HTML acquired via the network, and extracts a hyperlink indicating the storage location of the next linked node. Hyperlink is HT
By searching for a predetermined character string (referred to as a tag) indicating the start position specified by the ML and a tag indicating the end position, the character string is extracted as a character string between them.

[0008] The hyperlink representing the link destination node is a Uniform Resource Locators (Uniform R).
esource Locators Hereinafter referred to as URL. ). A node and a hyperlink can be searched for by finding a portion sandwiched between tags indicating hyperlinks from the acquired text of the node and extracting a character string described by a URL from the portion. The node link search device constructs a node link information database by repeating such a search. "WWW Wanderer", "WWW Robot", "WWW Spide"
There is something called "r".

The following configuration of a hyperlink representing a link destination node is defined in a URL. <Scheme>: <Scheme-Spec part
ific-Part)> In the case of the Hypertext Transfer Protocol (HTTP) scheme used as a document file transfer protocol in the World Wide Web (WWW), hyperlinks are represented as follows. http: // <host>: <port>: <path>? <searchpart>

Here, “<host>” represents the host name of the server machine where the document file exists, and “<port>” represents the port number for performing communication. The port number can be omitted, and if omitted, it is treated as if the standard value “80” was specified. “<Path>” indicates the location of the document file on the server machine. “<Searchpart>” is an area used when transferring data to the server machine.

FIG. 14 shows an example of nodes and hyperlinks connecting these nodes. The document files included in each of the nodes 1001, 1002, and 1003 are described in HTML. The contents of the document file are classified and identified by various tags. Of these, the tag indicating a hyperlink is <A HREF="URL">… </ </
A part represented by A> and corresponding to "URL" indicates a link destination node. In the document file of the node 1001, areas 1004 and 1005 surrounded by dotted lines represent hyperlinks, respectively.

The node 1001 uses the URL “http: // hos
tA / tale / TOC.html ”. The node 1002 is“ http: // hostB: 80 / tale ”.
/Introduction.html ”and node 1003 is“ http: // h
ostC: 8080 / tale / Birth.html ". Hyperlink 10 in node 1001
04 indicates the node 1002. The hyperlink 1005 in the node 1001 is
3 is indicated. In this way, by the hyperlink registered in the node's document file,
The link destination node is represented.

FIG. 15 shows an example of the relationship between documents of a node. The area surrounded by the dashed line represents the server where the node is stored. An area 1011 is a server machine having a host name “A”, an area 1012 is a server machine having a host name “B”, and an area 1
013 represents a server machine whose host name is “C”, respectively.

The nodes 1001 to 1003 shown in FIG.
1014 to 1016 of the document file included in the file, excluding tags and the like defined by HTML
Is an area 10 representing a server in which the node is stored.
It is represented in the 11-1013. Node 100
The link destination nodes 1 are nodes 1002 and 1003
Are represented by hyperlinks 1017 and 1018.

Next, the configuration of a node / link search device that constructs a node / link information database based on hyperlinks registered in each node will be described.

FIG. 16 shows an outline of the configuration of a conventionally used node / link searching device.
A plurality of server machines 1023 are connected to the node / link search device 1021 via a network 1022 such as the Internet. Each server machine 1023
Stores nodes storing document files. The node / link search device 1021 includes a file collection unit 1031 that collects node files from the server machine 1023, and a hyperlink extraction unit 1032 that extracts hyperlinks from node files acquired from the server machine 1023. .

Further, a node link information database 1033 for storing and managing various attribute information such as a storage location of the node and link information between the nodes based on the extracted hyperlink is provided. The search range control unit 1034
This is a part that limits the search range of the node to a range that matches the condition of the search range provided from an input terminal (not shown) or an external file for setting. System control unit 1035
Is a circuit part that controls the flow of operation of each unit in the node / link search device 1021 in an integrated manner.

The hyperlink extracting unit 1032 includes a hypermedia description language analyzing unit 1036 that analyzes the contents of the multimedia document file of the node written in the hypermedia description language. The hypermedia description language analysis unit 1036 extracts a hyperlink and a link destination node by detecting various tags. The search range is the server machine 10
23 is specified by the host name as an identifier. The search range control unit 1034 includes a host name comparison unit 1037 that compares the host name of the server machine where the node is stored with the host name as the specified search range. By comparing the host name as a part of the information indicating the storage location of the node, it is determined whether or not the node is within the search range.

The file collection unit 1031 of the node / link search device 1021 sends the network 1 from the server machine 1023 that matches the host name specified as the search range.
022, the multimedia document file of the node is read. Then, the hyperlink extraction unit 10
Then, the information of the hyperlink and the link destination node is extracted from the file read in step S32. Search range control unit 103
No. 4 continues the search for the linked node if the host name of the server machine where the extracted linked node is stored is within the search range. On the other hand, when the server machine at the storage destination of the link destination node is out of the search range, the search to the node after that is stopped. Various attributes of nodes and hyperlinks within the search range are registered in the node / link information database 1033.

As described above, in addition to the node whose host name is specified as a search range, a node or node whose search range cannot be specified at all.
There is also a link search device.

[0021]

A node / link search device that cannot specify a search range makes the entire network such as the Internet a search range, and searches for all nodes existing in all server machines on the network. . For this reason, unnecessary information on nodes other than those which should be originally set as a search range is also accumulated in the node link information database, which not only wastes resources but also reduces work efficiency in searching. . further,
There is also a problem that the time for accessing the server machine or the network where the node exists becomes long, and the use of network resources of other users is restricted.

Conventionally, even a device that can specify a search range can only limit the search range in units of host names. This means that the search range is limited by the physical position of the node. However, since hypertexts are searched sequentially based on the semantic relationship between nodes, an appropriate range cannot be specified by specifying a search range in units of server machines.

For example, in the example shown in FIG. 15, a node 1014 whose table of contents is registered exists on server "A" 1011 and server "B" 1012 and server "C" 1
013, nodes 1015 and 1016 in which sentences having contents corresponding to the table of contents are registered, respectively. In such a case, if the search range is to include not only the table of contents but also the contents of the written document, the server “A”,
All of the servers "B" and "C" must be specified as host names in the search range. As a result, there is a problem that a search range extends to many other nodes having no meaningful connection with the table of contents existing on the server “B” and the server “C”, and an unnecessary node is searched. .

It is an object of the present invention to provide a node / link search device which can limit the search for unnecessary nodes having no meaningful connection.

[0025]

According to the first aspect of the present invention, link information indicating the name of the link destination node included in each node of the hypertext and the storage position of the link destination node and the host of the search range are provided. restriction of the search range when sequentially performing the search to the linked node from any node to name a based, is the number of nodes existing between the node as a starting point of the search until the search destination node Search condition setting means for setting the maximum value of the number of layers and a character string for specifying the server machine to be searched, and searching for reading the contents of the linked node indicated by the link information from the server storing it File collection means that repeats in order from the node serving as the starting point of the file, and each time the file collection means reads the contents of one node, Link information storage means for storing link information and information indicating the correspondence between the link destination node indicated by the link information and the node read this time, and each time the content of one node is read by the file collection means, Means for calculating the number of layers of the node at the link destination indicated by the included link information, and this reading when the number of layers obtained by the means for calculating number of layers is larger than the maximum value of the number of layers set by the search condition setting means Search range restricting means for stopping the reading of the contents of the nodes linked after the node by the file collecting means, and the number of layers obtained by the search range restricting means is greater than the maximum value of the number of layers set by the search condition setting means. If less than or equal to this, the string specified as the host name in the search range and the node Host name partial match determining means for determining whether or not the host name partially matches only a specified character string from the beginning; and if the host name partial match determining means determines that the partial match occurs, this node sets the search range. Is provided in the node / link search device.

That is, according to the first aspect of the present invention, the number of nodes existing between the search starting point node and the search destination node is set as the number of layers, and the condition for limiting the search range is set to the maximum number of layers. Specify by value. By following the link from the starting node, the contents of the nodes are sequentially read from the server, and the names and storage positions of the linked nodes of the read nodes and the correspondence between the link source nodes and the link destination nodes are registered. Also, each time a node is read from the server, the number of layers for the node to which the read node is linked is obtained. When the number of hierarchies becomes equal to or greater than the set maximum value, the link is traversed further and reading of the node is stopped.

As a result, the range of the search is limited by the number of layers from the starting node. As described above, since the search range is limited by the number of layers from the starting node, the searched nodes are only those having a strong semantic connection with the starting node. Further, since the search range can be limited by the number of layers, it is possible to collect only information in a necessary range. In addition, it was obtained by the search range limiting means.
The number of layers is the maximum number of layers set by the search condition setting means.
If less than or equal to the value, the search range
Character string specified as the enclosed host name and the host of the node
Whether the name partially matches only the specified character string from the beginning
Is determined by the host name partial match determination means.
If the partial match determination means determines that partial
Is determined to exist in the search range.
You .

According to the second aspect of the present invention, any information is provided based on the link information indicating the name of the link destination node included in each node of the hypertext and the storage position of the link destination node and the host name of the search range. Conditions for search range when searching sequentially from a given node to a linked node
And the character string for identifying the maximum and exploration <br/> search to the server machine in the hierarchy number is the number of links existing between from the starting point of the search node to the search destination node <br A search condition setting unit to be set; a file collection unit for repeatedly reading out the contents of the link destination node indicated by the link information from the server storing the same in order from the node serving as the search start point; Node / link information storage means for storing link information included in a node every time the contents of one node is read by the means, and information indicating the correspondence between the link destination node indicated by the node and the currently read node; Each time the content of one node is read by the collection means, the number of layers of the link to the link destination node represented by the link information included in the node Means for calculating the number of layers, and when the number of layers obtained by the number of layers calculation means is greater than the maximum value of the number of layers set by the search condition setting means, the file of the contents of the nodes linked after the node read this time A search range limiting means for stopping reading by the collecting means, and a host name of the search range when the number of layers obtained by the search range limiting means is smaller than or equal to the maximum value of the number of layers set by the search condition setting means. A host name partial match determining unit that determines whether or not a character string specified as a part of the node and a host name of the node partially match only the character string specified from the beginning; and that the host name partial match determining unit partially matches. A search range determining means for determining that this node exists in the search range when the determination is made;
A link search device is provided.

That is, in the invention according to the second aspect, the number of links existing from the search starting point node to the search destination node is defined as the number of layers, and the search range is limited by the maximum value of the number of layers. ing. As a result, it is possible to collect only information about nodes that have a strong semantic connection with the originating node. Further, only for nodes within the search range specified by the number of layers, their names, storage positions, and link relationships between nodes can be collected. Also, search range
The number of hierarchies determined by the limiting means is
Less than or equal to the maximum number of layers
If they are equal, the specified host name in the search range
Character with the character string and the host name of the node specified from the beginning
Host name partial match determination means to determine whether only a column partially matches
And the host name partial match determination means
If it is determined that this node is in the search range
Is determined.

According to the third aspect of the present invention, the nodes to be searched are distributed and stored in a plurality of servers connected to the network.

That is, according to the third aspect of the present invention, each node to be searched is distributed and stored in a plurality of servers connected via a network. Since the search range is limited by the number of layers, even if nodes are distributed to a plurality of servers, only necessary node information can be collected from many servers. For example, if the search range can be specified only for each server, not only the nodes exceeding the required number of layers from the starting node,
In some cases, irrelevant nodes that do not have a link relationship with the originating node are collected. By limiting the search range according to the number of layers, even if nodes in the search range are distributed to multiple servers, it is possible to collect only information about nodes in the necessary range that have a semantic connection from the starting node. it can.

According to the fourth aspect of the present invention, the file collection means determines whether or not the link destination node indicated by the link information included in the currently read node is the same as the already read node. Determining means; and multiplex reading stopping means for stopping the rereading of the node when the same node determining means determines that the read node is the same as the already read node.

That is, in the invention according to claim 4, re-reading of a node which has been read once is prevented. This makes it possible to avoid repeatedly searching the looped range.

[0034]

BEST MODE FOR CARRYING OUT THE INVENTION

[0035]

FIG. 1 is a block diagram of a node according to an embodiment of the present invention.
1 shows an outline of a configuration of a link search device. The search procedure on the hypermedia structure can be viewed as a tree structure with the node that started the search as the root. Therefore, the depth of a tree at each node is set as the number of layers with the root as the starting point, and the search range is limited by the number of layers from the starting point.

A plurality of server machines 13 storing nodes are connected to the node / link search device 11 via a network 12 such as the Internet. The node / link search device 11 includes a file collection unit 21 that collects node files from the server machine 13 and a hyperlink extraction unit 22 that extracts hyperlinks and link destination nodes from the node files acquired from the server machine 13. It has.

The node / link search device 11 has a node / link information database 23 for storing and managing extracted hyperlinks and various types of attribute information of link destination nodes. The search range control unit 24 is a part that limits the search range of the node so as to meet a condition for specifying a search range input from an input terminal (not shown) or a setting file. The system control unit 25 is a circuit unit that controls the operation flow of each unit in the node / link search device 11 in a comprehensive manner.

The hyperlink extracting unit 22 includes a hypermedia description language analyzing unit 26 that analyzes the contents of the multimedia document file of the node written in the hypermedia description language. The hypermedia description language analysis unit 26 extracts a hyperlink and a link destination node by detecting various tags. The hyperlink extraction unit 22 includes a hierarchy number calculation unit 27 that calculates the hierarchy number of the extracted link destination node. The number-of-layers calculation unit 27 obtains the number of layers of the link destination node by adding “1” to the number of layers of the link destination node.

The search range is specified by the host name as an identifier of the server machine 13 and the maximum number of layers. The search range control unit 24 compares the host name of the storage destination of the node with the host name specified as the search range, the number of layers of the node, and the maximum number of layers as the limit value of the search range. And a number-of-layers comparing section 29 for comparing the number of layers. The search range control unit 24 limits the search range according to the number of nodes and the host name of the storage destination. Also, in order to prevent multiple readings of the same node, the range of the search is limited by the elapsed time from the previous search.

The file collection unit 21 is connected to the network 12
The multimedia document file of the node is read from the server machine 13 through the server. The hyperlink extraction unit 22 extracts a hyperlink and a link destination node from the multimedia document file based on the analysis result of the hypermedia description language analysis unit 26. Next, the hierarchy number calculation unit 27 calculates the number of layers of the link destination node or the hyperlink by adding “1” to the link source node.

The search range control unit 24 determines the search range based on whether the number of layers of nodes or hyperlinks is equal to or less than the specified maximum number of layers, or whether the host name partially matches the specified number. Restrict. For the nodes and hyperlinks within the search range, their attribute information is stored in the node / link information database 23. In the node / link information database 23, a table for registering attributes of nodes is called a node table, and a table for registering attributes of links is called a link table. In this way, by limiting the search range by the number of layers, collection of attribute information on unnecessary nodes and registration in the database are prevented.

FIG. 2 shows an example of the registered contents of the node table. The node table 31 includes, from the left of the figure, a node identifier 32, a scheme (Scheme) 33, a host name 34, a port number 35, and a scheme-specific part (Scheme-
Specific) 36, the number of layers 37, the registration date and time 38, the search date and time 39, and the last update date and time 41 are registered. The node identifier 32 is an identifier that is automatically generated when the attribute information of the node is registered in the node table by the database management system (DBMS) not shown in FIG. The scheme 33, the host name 34, the port number 35, and the scheme specific part 36 correspond to each part of the character string of the URL indicating the storage position of the node. The number of layers 37 is an item indicating the number of layers of the node from the search start position.

The registration date and time 38 indicates the date and time when the node was first registered in the database. The search date and time 39 represents the latest date and time when this node was searched. The last update date / time 41 represents the last update date / time of the multimedia document file of the node. In the figure, items of a search date and time 39 and a last update date and time 41 surrounded by a dotted line 42 are attributes updated when registered as a link source node, and items enclosed by a dotted line 43 are items of the link source node. This attribute is updated when registered as a link destination node indicated by a hyperlink.

FIG. 3 shows an example of the registered contents of the link table. The link table 51 includes a link identifier 52, a link source node identifier 53,
The link destination node identifier 54 is registered. In these items, the identification numbers of the nodes automatically assigned by the DBMS are registered. The node identifier of the source node of the hyperlink indicated by the link identifier 52 is registered as the link source node identifier 53, and the node identifier of the node pointed to by the hyperlink is registered as the link destination node identifier 54.

In the node table 31 of the node / link information database, at least one node information serving as a search start point is registered in advance. At this time, the items to be registered are the node identifier 32, the scheme 33, the host name 34, the port number 35, the scheme specific part 36, the number of layers 37, and the registration date 38. "0" is set to the number of layers of the node serving as the starting point of the search.

FIG. 4 shows the flow of the operation performed by the node / link searching device. The processing of the steps surrounded by the dashed line 61 is performed by the system control unit 25 of the node / link device. The processing of the steps surrounded by the dashed line 62 is performed by the search range control unit 24, and the processing of the steps surrounded by the dashed line 63 is performed by the file collection unit 21 and the hyperlink extraction unit 22. First, prior to the search, the system control unit initializes the system such as inputting search conditions (step S101). The search condition is input from an input terminal or an external file not shown in FIG. Here, as the search conditions, the maximum value of the number of layers (the maximum number of layers) of the nodes to be set as the search range, the elapsed time for determining that the previous search result is old, and the server machine to be searched are specified. (A part or all of the host name to be searched) is set as a condition.

The system control unit 25 includes the node table 3
1 to check whether there is an unsearched node (step S102). When the search date and time 39 registered in the node table 31 is undefined (unregistered),
When the search date / time 39 corresponds to one of the cases where the current time is earlier than the current time by the elapsed time set as the search condition, it is determined that the node has not been searched. When there is an unsearched node (step S102; Y), the search range control unit 24 extracts from the node table 31 the attribute information of one unsearched node. At this time, the node is preferentially selected from the one having the smaller number of layers (step S103). Therefore, at first, the attribute information of the node at the search start point whose hierarchical level is “0” is extracted.

Next, it is determined whether or not the selected node is a node existing within the designated search range (step S104). The process of determining whether or not it is within the search range will be described later in detail. If the node exists within the search range (step S104; Y), the document file of the node is read through the network, and the table information is updated such that the link destination node described therein is additionally registered in the node table. The processing (step S105) is performed. Details of the table update processing will be described later. After the update processing for one node (step S105) is completed, step S10 is performed again.
Returning to step 2, the search process for the unsearched nodes is repeated.

If the selected node is outside the designated search range (step S104; N), the search time 39 for the selected node in the node table 31 is changed to the current time (step S106). Then, the process returns to step S102. When the unsearched node no longer exists in the node table (step S102; N), a termination process for terminating the search is performed (step S10).
7) End the processing (END).

FIG. 5 shows the flow of processing for determining whether a node exists within the search range. Here, it is determined whether or not the area is within the search range based only on the number of layers. First, the search range control unit 24 compares the number of layers of the selected node with the maximum number of layers set as a search condition (step S201). If the number of layers of the node is larger than the maximum number of layers (step S20
1; N), it is determined that this node does not exist in the search range (step S202). If the number of layers of the node is smaller than or equal to the maximum number of layers (step S
201; Y), it is determined that this node exists in the search range (step S203).

FIG. 6 shows the flow of another example of the process for determining whether or not a node exists within the search range. Here, in addition to the determination based on the number of layers, the determination based on the host name is added. The search range control unit 24 compares the number of layers of the selected node with the maximum number of layers (step S301), and when the number of layers of the node is larger than the maximum number of layers (step S301; N), this node It is determined that it does not exist in the search range (step S30)
2). When the number of layers of the node is smaller or equal to the maximum number of layers (step S301; Y), it is determined whether or not the character string specified as the host name in the search range partially matches the host name of the node. (Step S30
3).

For example, when "AB" is specified as the search character string, all characters including the character string "AB" specified from the beginning of the host name such as "ABC" or "ABD" are partially matched. Is determined. If the partial match does not occur, such as “CAB” (step S30
3; N), it is determined that this node does not exist in the search range (step S302). If they partially match (step S303; Y), it is determined that this node exists in the search range (step S304). Here, the partial number matching of the host name is determined after the number of layers is determined, but the order may be interchanged.

FIG. 7 shows the flow of the table updating process shown in FIG. This process is performed by the file collection unit 21 and the hyperlink extraction unit 22. First, the content of the node determined to be within the search range in step S104 of FIG. 4 is read out from the server machine where it is stored via the network. Therefore, the file collection unit 21 reads the attribute information of the corresponding node from the node table 31 of the node / link information database 23 (Step S401).
Here, the number of layers of the read node is assumed to be n here.
Next, the file collection unit 21 initializes a communication buffer and a temporary buffer (not shown) used for communication with the server machine, and then connects to the server machine 13 to prepare for transfer (reading) of the document file of the node. (Step S402).

Subsequently, the file collection unit 21 starts transferring the document file according to the transfer protocol of the distributed hypermedia system. File collection unit 21
Reads the transfer protocol message header (step S403) and checks the format such as the description language of the document file (step S404). If the read file is not in the hypermedia description language format (step S404; N), the file is read up to the end of the document file (step S405). Then, post-processing such as releasing the communication buffer and the temporary buffer and disconnecting the connection with the server machine is performed (step S406), and the processing ends (END).

If the read file is in the hypermedia description language format (step S404; Y), the file collection unit 21 reads the document file of the body of the transfer protocol message into the communication buffer (step S407). Next, the hypermediad description language analysis unit 26 moves the start character to the end character of the tag representing the hyperlink portion to the temporary buffer (step S408). The portion describing the link destination node is extracted from the hyperlink stored in the temporary buffer (step S409). The hierarchy number calculation unit adds “1” to the hierarchy number “n” of the link source node to “n +
1 "is obtained as the number of layers of the link destination node (step S410).

The attribute information of the nodes and hyperlinks thus obtained is stored in the node / link information database 23.
It is written into the node table 31 and the link table 51 in (step S411). At this time, each part of the attribute information of the link source node and the attribute information of the link destination node and all the items of the attribute information (link table) of the hyperlink are written. As a result, the attribute information of the one-level advanced link destination node linked to the currently read node and one hyperlink from the read node to the link destination node are additionally registered in each table. Become. However, if the same node has already been registered, no additional registration to the table is performed, and only the update of the search date and time is performed. This avoids multiple registrations of the same node or hyperlink.

Next, it is checked whether or not processing has been performed up to the end of the document file of the node read this time (step S 412).
412; N), and return to step S407. Each node is
Since there may be a plurality of link destination nodes, such processing is repeated until the end of the file, whereby all the link destination nodes of the currently read node are registered in the node table and the link table.
When processing is performed up to the end of the file (step S41)
2; Y), the table update processing for this node ends (END).

Here, the contents of the attribute information registered in the node table and the link table in step S411 of FIG. 7 will be described. A node read from the server machine is defined as a link source node, and a node described in a hyperlink included in the node is defined as a link destination node. First, for the link source node, the search date and time 39
Is registered. Thus, the latest time information indicating when the node was searched can be left. For example, in the node table 31 shown in FIG. 2, it is assumed that a node having a node identifier of "1" is read from the server machine, and a node having a node identifier of "2" is described as a destination pointed by the hyperlink.
In this case, the node with the node identifier “1” is the link source node, and the node with the node identifier “2” is the link destination node.

In FIG. 2, the search date and time and the last update date and time surrounded by a dotted line 42 are updated as attribute information of the link source node. What is registered as the attribute information of the link destination node is a portion surrounded by a dotted line 44 in FIG. That is, the node identifier and the URL indicating the storage location of the node
, A scheme name, a host name, a port number, a scheme-specific part, the number of layers obtained by the number-of-layers calculation unit, and a registration date and time registered in the database. As the node identifier, one automatically generated by the DBMS is registered. As for the link destination node, the search date and time and the last update date and time remain unregistered since the hyperlink included in the linked node has not been actually read from the server machine 13 yet.

In the link table 51, a link identifier 52 automatically generated by the DBMS, a link source node identifier 53, and a link destination node identifier 54 are registered. This will be described with reference to FIG. It is assumed that the node having the node identifier “1” is read, and the node having the node identifier “2” is a link destination node by the hyperlink. First, since the node identifier of the source node of the hyperlink (55) whose link identifier is "1" is "1", this value is registered as the link source node identifier (56). Since the node identifier of the node pointed to by the hyperlink is “2”, this value is registered as the link destination node identifier (57). If other hyperlinks of the read node are registered, a link identifier is also assigned to those links, and a link source node identifier and a link destination node identifier are registered. If there are two hyperlinks, the dotted line 58 in FIG.
Is registered in the link table 51.

The updating of the node table 31 and the link table 51 is performed each time one node is read from the server machine 13 and a new link destination is found. Therefore, in the flowchart of FIG. 4, when the table update processing (step S105) is completed for one node and the process returns to step S102, the node newly registered in the table update processing is also searched. As a result, the search for the link destination proceeds one after another. However, the search range is limited by conditions such as the number of layers.

Further, the fact that the search date and time is undefined or a time earlier than the elapsed time is used as an unsearch determination criterion, and a time near the current time is registered as the search date and time of the node once searched. The link destination is not repeatedly examined for the node. Thus, for example, even when the link destination of the node “A” is the node “B” and the link destination of the node “B” is a loop formed by links like the node “A”, the node “A” "
Is not checked again, and the loop is not repeated up to the maximum number of layers.

As described above, since the range of the search is limited by the number of layers of the nodes, it is possible to search only the nodes having strong semantic connection with the starting node. Further, unnecessary search can be reduced as compared with the case where the search range can be limited only on a server basis.

Modification

In the above-described embodiment, the number of layers is given as the attribute information of the node. In the modified example, it is determined whether or not the node is within the search range based on the number of layers of the hyperlink. I have. The configuration of the device is the same as that shown in FIG. 1, and a description thereof will be omitted.

FIG. 8 shows an example of the registered contents of the node table used in the node / link search device of the modified example. The same items as those in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted as appropriate. The node table 71 is different from the node table 31 shown in FIG. 2 in that items for registering the number of layers and the search date and time are deleted.

FIG. 9 shows an example of the registered contents of the link table used in the node / link search device of the modified example. Items that are the same as those in FIG. Link table 81
Has an item 82 of the number of layers and an item 83 of the search date and time in addition to the link table 51 shown in FIG. These are registered in the node table in the embodiment. In the modification, it is necessary to register at least one hyperlink serving as a search start point in the link table 81 prior to the search. The hyperlink serving as a search start point registered here is virtual, and has no link source node (unknown). In addition, attribute information about a link destination node indicated by a hyperlink serving as a search start point must be registered in the node table 71 in advance.

As contents to be registered in the node table 71 prior to the search, first, there is a node identifier of a link destination node indicated by a hyperlink at the starting point of the search. This value is DB
Assigned automatically by MS. Furthermore, a scheme, a host name, a port number, and a scheme specific part are initially registered as items corresponding to the URL character string indicating the storage location of the node. Also, a registration date and time indicating the time when this node was registered in the database is initially registered. In the example of FIG. 8, items in a range surrounded by a dotted line 72 are initially registered.

The contents initially registered in the link table 81 include a link identifier which is an identifier of a hyperlink serving as a search starting point. This value is automatically generated by the DBMS. Since the link source of the hyperlink serving as the starting point of the search is unknown, the initial value of the link source node identifier is set to “0” representing undefined. As the link destination node identifier, the same value as the node identifier assigned by the DBMS to the link destination node is registered. For the number of layers, “0”, which is the starting point of the search, is initially registered. The search date and time registers the date and time when the search for the hyperlink was performed, and is left undefined as the initial value of the search start. In the example of FIG. 9, the items enclosed by the dotted line 84 are initially registered.

FIG. 10 shows a flow of a process performed by the node / link searching device according to the modification. The steps surrounded by the dashed line 101 are the system controller 25
Is performed. Steps surrounded by the dashed line 102 are performed by the search range control unit 24 by the dashed line 10.
Steps surrounded by 3 represent processes performed by the file collection unit 21 and the hyperlink extraction unit 22.

First, the system controller 25 initializes the system (step S501). At this time, the maximum number of layers as the maximum value of the number of layers of the hyperlink to be the search range and the elapsed time serving as a reference for treating the previous search result as an old one are set. When setting a server range by a host name, a character string for limiting the host name is input. These are taken from an input terminal (not shown) or an external file. Next, it is determined whether or not a hyperlink that has not been searched exists in the link table 81 (step S502). A hyperlink that has not been searched is one whose search date and time is undefined (unregistered), or one whose search date and time is older than the current time by an elapsed time or more.

When a hyperlink that has not been searched exists in the link table 81 (step S502;
Y), one of the unsearched hyperlinks is selected and its attribute information is extracted (step S503).
Next, the search range control unit 24 determines whether or not the selected one hyperlink is within the search range (step S504). The determination is made based on the number of layers or both the number of layers and the host name, and details thereof will be described later.

If it is within the search range (step S504; Y), the table is updated (step S50).
Perform 5). In this process, the attribute information of the link destination node indicated by the hyperlink is extracted from the node table 71, the document file of this node is read from the server machine, and the attribute information of the link destination node registered therein is additionally registered. . The detailed flow of the processing will be described later. After finishing the table updating process (step S505) for the link destination node indicated by the hyperlink, the process returns to step S502 again, and the search process for the unsearched hyperlink is repeated.

If the selected hyperlink is out of the search range (step S504; N), the link table 81
The search date and time 39 for the selected hyperlink is changed to the current time (step S506), and step S5 is performed.
Return to 02. When an unsearched hyperlink no longer exists in the link table 81 (step S502;
N), end processing for ending the search is performed (step S507), and the processing is ended (END).

FIG. 11 shows the flow of processing when determining whether a hyperlink is within the search range.
Here, it is determined whether or not the area is within the search range only by the number of layers of the hyperlink. First, the search range control unit 24
Compares the number of layers of the selected hyperlink with the maximum number of layers set as the search condition (step S6).
01). If the number of layers of the hyperlink is larger than the maximum number of layers (step S601; N), it is determined that the hyperlink does not exist in the search range (step S602). When the number of layers of the hyperlink is smaller than or equal to the maximum number of layers (step S60)
1; Y), it is determined that this hyperlink exists in the search range (step S603).

FIG. 12 shows the flow of another example of the process for determining whether or not the hyperlink is within the search range. Here, in addition to the determination based on the number of layers of the hyperlink, the determination based on the host name is added. The search range control unit 24 compares the number of layers of the selected hyperlink with the maximum number of layers (step S701), and when the number of layers of the hyperlink is larger than the maximum number of layers (step S701).
701; N), it is determined that this hyperlink does not exist in the search range (step S702). When the number of layers of the hyperlink is smaller or equal to the maximum number of layers (step S701; Y), the attribute information about the link destination node indicated by the hyperlink is extracted from the node table 71 (step S703).

It is checked whether or not the host name included in the attribute information of the extracted link destination node partially matches the character string specified as the host name in the search range (step S7).
04). If they partially match (step S704;
Y), it is determined that the preceding hyperlink exists within the search range (step S705). If the partial links do not match (step S705; N), it is determined that the hyperlink does not exist within the search range (step S702).
Here, the partial number matching of the host name is determined after the number of layers is determined, but the order may be interchanged.

FIG. 13 shows the flow of the table updating process shown in FIG. This process is performed by the file collection unit 21 and the hyperlink extraction unit 22. First, the attribute information of the hyperlink determined to be within the search range in step S504 of FIG. 10 is read from the link table 81 (step S801).
The number of layers of the read hyperlink is assumed to be n here.
And Next, the attribute information of the node corresponding to the link destination node identifier of this hyperlink is stored in the node table 71.
(Step S802).

The file collection unit 21 connects to the server machine after initializing a communication buffer (not shown) and a temporary buffer used for communication with the server machine, and prepares to read the document file of the link destination node indicated by the hyperlink. Is performed (step S803). Subsequently, the file collection unit 21 starts transferring the document file according to the transfer protocol of the distributed hypermedia system.

The file collection unit 21 reads the transfer protocol message header (step S804). If the description language of the document file is not in the hypermedia description language format (step S805; N), the file collection unit 21 reads up to the end of the document file. (Step S806). Then, post-processing such as releasing the communication buffer and the temporary buffer and disconnecting the connection with the server machine is performed (step S807), and the processing ends (end).

If the read file is in the hypermedia description language format (step S805; Y), the file collection unit 21 reads the document file of the transfer protocol message body into the communication buffer (step S808). Next, the portion from the start character to the end character of the tag representing the hyperlink portion is moved to the temporary buffer (step S809), and the portion describing the link destination node is extracted from the hyperlink stored in the temporary buffer (step S809). Step S810). Layer number calculation unit 2
7 is “n +” as the number of layers of the extracted hyperlink.
1 "is set (step S812).

The attribute information of the nodes and hyperlinks thus obtained is stored in the node / link information database 23.
It is written into the node table 71 and the link table 81 in (step S812). At this time, each item of the attribute information of the link source node and the attribute information of the link destination node and all the items of the attribute information of the hyperlink are written. As a result, the attribute information about the link destination node of the currently read node and the hyperlink from the read node to the link destination node are registered.
However, if a hyperlink having the same link source and link destination has already been registered, additional registration to the table is not performed, and only update of the search date and time is performed. This prevents multiple registrations of the same node or hyperlink.

Next, it is checked whether or not processing has been performed up to the end of the document file of the currently read node (step S813).
813; N), and return to step S808. Each node is
Since there may be a plurality of link destination nodes, by repeating such processing until the end of the file, all the link destination nodes of the currently read node are registered in the node table and the link table. When processing is performed up to the end of the file (step S81
3; Y), end the process (END).

Here, the contents of the attributes registered in the node table 71 and the link table 81 in step S812 in FIG. 13 will be described. A node read from the server machine is defined as a link source node, and a node described in a hyperlink included in the node is defined as a link destination node. First, the last update date and time 41 is registered for the link source node. For example, it is assumed that, in the node table 71 shown in FIG. 8, a node having a node identifier "1" is read from the server machine, and a node having a node identifier "2" is described as a destination pointed by the hyperlink. In this case, the node identifier is “1”
Is the link source node, and the node with the node identifier “2” is the link destination node.

In FIG. 8, the last update date and time surrounded by a dotted line 73 is updated as attribute information of the link source node. What is registered as the attribute information of the link destination node is a portion surrounded by a dotted line 74 in FIG. That is, a scheme identifier, a host name, a port number, a scheme specific part, and a registration date and time, which is a date and time registered in the database, corresponding to the node identifier and the character string of the URL indicating the position of the node. As the node identifier, one automatically generated by the DBMS is registered.

In the link table 81, a link identifier automatically generated by the DBMS, a link source node identifier, a link destination node identifier, the number of layers, and a search date and time are registered. This will be described with reference to FIG. It is assumed that, based on the hyperlink having the link identifier “1”, the node having the link destination node identifier “1” is read. It is assumed that the node having the node identifier “2” and the node having the node identifier “3” are linked by the hyperlink described in the read node.

At this time, since the search is performed with reference to the hyperlink 84, the time is registered as the search date 85. Next, a link identifier of the newly obtained hyperlink is acquired from the DBMS, and registered in the column of a link identifier to be additionally registered. For example, the link identifier (8) of the hyperlink (86) whose link identifier is "2"
7) is registered as "2". This value is an ID number appropriately given by the DBMS based on the registration order of the table and the like.

The link source node of the hyperlink 86 is
Since the value of the node identifier is “1”, “1” is registered as the link source node identifier 88. In the link destination node identifier 89, the value of the node identifier given by the DBMS when the attribute of the link destination node is registered in the node table 71, that is, “2” is registered. Also, since the number of layers 91 is a hyperlink obtained from a node obtained by following the hyperlink with the number of layers “0”, a value “1” obtained by adding “1” to “0” is registered. Hyperlink 9 toward node with node identifier "3"
2 is registered by the same procedure. The information in the range indicated by the dotted line 93 in FIG. 9 is registered for the additionally registered hyperlink.

The updating of the node table 71 and the link table 81 is performed each time a link destination node is read from the server machine based on one hyperlink and a new link destination is found. Therefore, after the table update process (step S505) is completed for one node in the flowchart of FIG. 10, when returning to step S502, the hyperlink newly registered in the table update process is also searched. As a result, the search for the link destination proceeds one after another. However, the search range is limited by conditions such as the number of layers.

Further, since the search date and time is undefined or a time earlier than the elapsed time is used as an unsearched criterion, the hyperlink once searched is not checked again. Thus, for example, even when the link destination of the node “A” is the node “B” and the link destination of the node “B” is a loop formed by links like the node “A”, the node “A” Does not check the hyperlink from "" to node "B" again, and does not follow the loop up to the maximum number of layers.

As described above, even when the search range is limited based on the number of layers of hyperlinks, the same effect as when the search range is limited based on the number of layers of nodes can be obtained. However, when assigning the number of layers to a hyperlink, it is necessary to refer to the node table in order to check the link destination node of the hyperlink obtained by referring to the link table. The number of reference processes for the table increases as compared with the case where the number is given.

[0092]

As described above in detail, according to the first to third aspects of the present invention, the search range is limited by the number of layers from the starting node, so that the searched node is the starting node. Only those that have a strong semantic connection to the node. Further, since the search range can be limited by the number of layers, it is possible to collect only information in a necessary range.
In addition, since unnecessary nodes are not searched, the load on the server and the communication line between the servers can be reduced, and the time required for the search can be shortened. In addition,
According to the first and second aspects of the present invention, the search range restriction
The number of hierarchies determined by the level is set by the search condition setting means
Less than or equal to the maximum number of layers
When the string specified as the host name in the search range
Only the character string specified from the beginning with the host name of the host
Whether or not they match is determined by the host name partial match determination means,
The host name partial match determining means determines that partial match has occurred.
If this node exists in the search range
And For this reason, it seems to contain a different character at the beginning
Exclude nodes with strange host names from the search range
And efficient search becomes possible.

According to the fourth aspect of the present invention, it is possible to prevent a node that has once been read from being read again, so that it is possible to avoid repeatedly searching a looped range. Thus, the search can be performed efficiently.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an outline of a configuration of a node / link searching device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of registered contents of a node table.

FIG. 3 is an explanatory diagram showing an example of registered contents of a link table.

FIG. 4 is a flowchart showing a flow of an operation performed by the node / link searching device.

FIG. 5 is a flowchart showing a processing flow when determining whether or not a node exists within a search range.

FIG. 6 is a flowchart illustrating another example of a process of determining whether a node exists within a search range;

FIG. 7 is a flowchart showing a flow of a table updating process shown in FIG. 4;

FIG. 8 is an explanatory diagram showing an example of registered contents of a node table used in a node / link search device of a modified example.

FIG. 9 is an explanatory diagram showing an example of registered contents of a link table used in a node / link search device of a modified example.

FIG. 10 is a flowchart illustrating a flow of an operation performed by a node / link search device according to a modification.

FIG. 11 is a flowchart showing the flow of processing when determining whether a hyperlink exists within a search range.

FIG. 12 is a flowchart illustrating another example of a process of determining whether a hyperlink exists within a search range.

FIG. 13 is a flowchart showing a flow of a table updating process shown in FIG. 10;

FIG. 14 is an explanatory diagram showing an example of nodes and hyperlinks connecting these nodes.

FIG. 15 is an explanatory diagram illustrating an example of a relationship between documents of a node.

FIG. 16 is a block diagram showing an outline of a configuration of a conventionally used node / link searching device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Node / link search device 12 Network 13 Server machine 21 File collection unit 22 Hyperlink extraction unit 23 Node / link information database 24 Search range control unit 25 System control unit 26 Hypermedia description language analysis unit 27 Layer number calculation unit 28 Host name Comparing unit 29 Hierarchical number comparing unit 31, 71 Node table 51, 81 Link table

────────────────────────────────────────────────── ─── Continuation of front page (56) References JP-A-8-305729 (JP, A) JP-A-6-110926 (JP, A) M. E. FIG. De Bra, R .; D.J. Post, "Information Retrieval in the World-Wide Web: Making Client-Based Searching Feasible", COMPUTER NETWORK KS AND ISDN SYSTEMS. 2 PP 183-192 (NOVEMBER 1994) Takeshi Sugai, Mitsunori Wada, "Information Filtering on the Internet (2)", Proc. -88 KIMMEL S, "Robot generated on the World Wide Web", Data base Vol. 19, No. 1, pp. 40-43, 46-49 (accepted by Japan Science and Technology Information Center on February 13, 1996) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30 G06F 12/00 545

Claims (4)

(57) [Claims] 1. A link destination from an arbitrary node based on link information indicating a name of a link destination node included in each node of the hypertext and a storage position of the link destination node and a host name in a search range. Specify the conditions for limiting the search range when searching for nodes sequentially, and specify the maximum value of the number of layers, which is the number of nodes that exist from the search start node to the search destination node, and the server machine to be searched. search condition setting means and repeatedly performs file contents of the destination node indicated by the link information from the node serving as a starting point for the search to be read from the server that stores it in the order set by a character string for Collection means, each time the contents of one node are read by the file collection means, the link information contained in the node and the link information A node / link information storage unit for storing information indicating the correspondence between the previous node and the node read this time; each time the content of one node is read by the file collection unit, the link information included in the node is indicated. A number-of-layers calculating means for calculating the number of layers of the link destination node; and if the number of layers determined by the number-of-layers calculating means is larger than the maximum value of the number of layers set by the search condition setting means, a link is provided after the node read this time Search range limiting means for stopping the reading of the contents of the node by the file collecting means, and the number of layers obtained by the search range limiting means is smaller than the maximum value of the number of layers set by the search condition setting means, or When equal to this, the character string specified as the host name of the search range and the host name of the node A host name partial match determining means for determining whether or not the character string specified from the beginning partially matches; and a determination that this node exists in the search range when the host name partial match determining means determines that the partial match has occurred. A search range discriminating means that performs
Link search device. 2. A method according to claim 1, further comprising the step of: searching for the link destination node name and the storage location of the link destination node included in each node of the hypertext and the host name of the search range from any node to the link destination. identify the server machine restriction of the search range when sequentially performing the search to the node, to a maximum value and search hierarchical number is the number of links existing between from the starting point of the search node to the search destination node search condition setting means and repeatedly performs file contents of the destination node indicated by the link information from the node serving as a starting point for the search to be read from the server that stores it in the order set by a character string for Collection means, each time the contents of one node are read by the file collection means, the link information contained in the node and the link information A node / link information storage unit for storing information indicating the correspondence between the previous node and the node read this time; and each time the contents of one node are read by the file collection unit, the link information included in the node is represented. Means for calculating the number of layers of the link to the link destination node; and a node read this time when the number of layers determined by the number of layers calculation means is greater than the maximum value of the number of layers set by the search condition setting means. A search range limiting means for stopping the reading of the contents of the linked nodes by the file collecting means; and the number of hierarchies obtained by the search range limiting means is greater than the maximum number of hierarchies set by the search condition setting means. If less than or equal to this, the string specified as the host name in the search range and the node Host name partial match determining means for determining whether or not the host name partially matches only a specified character string from the beginning; and, if the host name partial match determining means determines that the partial match has occurred, this node sets the search range. A search range determining means for determining that the node exists in the node.
Link search device. 3. The node / node according to claim 1, wherein the nodes to be searched are distributed and stored in a plurality of servers connected to a network.
Link search device. 4. The same node discriminating means for discriminating whether or not the link destination node indicated by the link information included in the currently read node is the same as the already read node. 3. The node according to claim 1, further comprising: a multiplex reading stop unit that stops reading of the node again when the same node determination unit determines that the read node is the same as the already read node. -Link search device.