JP2009187224A - Information processor and information processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information processor for performing the retrieval of a structure pattern appearing several times in a plurality of tree structures by recursive processing for every place to which a node is added. <P>SOLUTION: The first retrieval means of an information processor performs the retrieval of a structure pattern appearing several times in a plurality of tree structures to nodes of lower rank than a node as the object of current processing in the tree structures, and a second retrieval means performs the retrieval of the structure pattern to nodes in higher rank than the node as the object of current processing in the tree structures for every non-retrieved node in lower rank than the node in high rank, and the first retrieval means and the second retrieval means return to the original node starting the retrieval when there exist no node as the object of retrieval. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an information processing apparatus and an information processing program.

In addition to dramatic increases in computer processing capacity and storage device capacity, IT and networking have made it possible to easily collect large amounts of information. There is an increasing expectation for early discovery of information on market opportunities and risks from the collected information and discovery of hidden knowledge.
However, the amount of information collected often exceeds human processing capabilities. For this reason, it has been difficult to find a risk from information gathered in a large amount of information or extract knowledge and use it.

  On the other hand, with advances in technology such as pattern mining, it has become possible to extract information such as patterns of products purchased simultaneously from such a large amount of information. Technology for extracting patterns of products purchased at the same time and patterns of order of purchase has been researched and developed with attention from demands such as analysis of customer purchasing behavior. With the progress of structuring, pattern mining techniques for extracting patterns having a structure like a tree structure have attracted attention. Among the pattern mining techniques for extracting structure information, the tree structure is especially used for structuring various information such as document structuring and knowledge representation including XML (extensible Markup Language). large.

The technology for extracting subtree patterns from a tree-structured data group is broadly divided into an extracted subtree mining technology that extracts only the structure in which the parent-child relationship is exactly the same, and an ancestor-descendant even if the parent-child relationship is somewhat disturbed. There is an embedded subtree mining technique for extracting the structure if there is a relationship.
For data that occurs in the real world, such as those that record human operations such as document operation history, even if a person intends to work in the same way, the operation history data does not necessarily have the parent-child relationship of the data. Inconsistencies often occur. In such a case, it is desirable to apply an embedded subtree mining technique. Since real-world data often fluctuates in the same way, there is high expectation for the technique of embedded subtree mining. Examples of disclosed technologies for realizing embedded subtree mining include technologies such as TreeMiner, Dryade, and MB3-Miner.

  As a technique related to these, for example, Patent Document 1 has an object to provide a method and system for detecting an important pattern included in a set of data, and data represented by tree structure data. A system for detecting a frequent pattern from a database including a set using candidate patterns to be aggregated, (1) means for aggregating patterns matching the candidate pattern from the database, and (2) appearing by the aggregation It is disclosed that the apparatus includes a means for detecting a pattern having a high frequency and (3) means for generating a candidate pattern to be a next aggregation target from the detected pattern.

In addition, for example, Patent Document 2 has an object to provide an extraction device or the like suitable for extracting a pattern that frequently appears in an ordered tree, and an input reception unit of the extracting device inputs one or more ordered trees. The conversion unit converts each of the ordered trees that have received the input into a sequence expression, and the extraction unit extracts a pattern that appears at a predetermined frequency or more from the patterns included in each converted sequence expression The sequence representation can be obtained by performing a depth-first search of the ordered tree, placing a mark indicating the name of the clause that passes through the branch, and a backtrack mark when returning the branch. From the first mark to the last mark when the projection is repeated zero or more times, with one of the marks representing the names in the sequence of marks as a series representation as the first mark Refers to sequence of marks meet, whether projection is satisfied, the sequence context of the columns of the mark, be determined by the value of the projection context is disclosed.
JP 2001-134575 A JP 2004-355457 A

  It is an object of the present invention to provide an information processing apparatus and an information processing program that perform a recursive search for each place where a node is added in a search for a structure pattern that appears multiple times in a plurality of tree structures. Yes.

The gist of the present invention for achieving the object lies in the inventions of the following items.
The information processing apparatus according to claim 1, wherein a search for a structure pattern that appears multiple times in a plurality of tree structures is performed on a node lower than a node currently being processed in the tree structure. And a search for each of the unsearched nodes that are higher than the current processing target node in the tree structure and lower than the higher level node. Comprising two search means, wherein the first search means and the second search means return to the original node that started the search when there are no more nodes to be searched. To do.

  An information processing device according to claim 2 is the information processing device according to claim 1, wherein the information processing device in the tree structure coincides with a current processing target node in the structure pattern. The first search range determining means for determining a range to search for descendants based on the highest-order node among the nodes having a hierarchical relationship, wherein the first search means includes the first search means The search is performed based on the range determined by the search range determining means.

  An information processing apparatus according to claim 3 is the information processing apparatus according to claim 1, wherein the nodes in the tree structure that coincide with the parent node in the structure pattern have the highest relationship. Second search range determination means for determining a search range based on the lowest-order node with respect to an upper node and a node in the tree structure that coincides with a child node in the structure pattern. The second search means performs a search based on the range determined by the second search range determination means.

  An information processing apparatus according to claim 4 is the information processing apparatus according to claim 3, wherein the current information is selected from among the appearances of ancestor nodes of the current processing target node in the structure pattern. The image processing apparatus further includes holding means for holding an appearance location corresponding to a node that includes a node at the appearance location of the node to be processed as a descendant.

  The information processing apparatus according to claim 5 is the information processing apparatus according to claim 4, wherein deletion means for deleting from the holding means appearances other than the highest and lowest positions in a range where there is no branch in the tree structure Is further provided.

  An information processing apparatus according to claim 6 is the information processing apparatus according to claim 4, wherein the appearance portion to be held by the holding unit is targeted, and other than the highest and lowest positions in the tree structure where there is no branch The image processing apparatus is further characterized by further comprising selection means for selecting items other than the appearance locations of.

  The information processing program according to claim 7, wherein the computer searches for a structure pattern appearing a plurality of times in a plurality of tree structures with respect to a node lower than the current processing target node in the tree structure. 1 search means and a search for the structure pattern are performed for each unsearched node that is higher than the current processing target node in the tree structure and lower than the higher node. The first search means and the second search means return to the original node that started the search when there are no more nodes to be searched. It is characterized by.

  According to the information processing apparatus of the first aspect, it is possible to perform a recursive search for each place where nodes are added in a search for a structure pattern that appears multiple times in a plurality of tree structures.

  According to the information processing apparatus of the second aspect, it is possible to suppress an increase in storage capacity and processing time in the search process.

  According to the information processing apparatus of the third aspect, it is possible to suppress an increase in storage capacity and processing time in the search process.

  According to the information processing apparatus of the fourth aspect, it is possible to hold the search state and execute recursive processing in the search.

  According to the information processing apparatus of the fifth aspect, the storage capacity in the search process can be further reduced.

  According to the information processing apparatus of the sixth aspect, the storage capacity in the search process can be further reduced.

  According to the information processing program of the seventh aspect, it is possible to perform a recursive search for each place where a node is added in a search for a structural pattern that appears multiple times in a plurality of tree structures.

First, the techniques of the above-mentioned TreeMiner, Dryade, and MB3-Miner will be described.
Dryade has a function restriction that sibling nodes do not include the same thing, and is not suitable for mining such as an operation history of a document in which such a scene frequently appears.
MB3-Miner is a breadth-first search, and the number of pattern candidates prepared for each processing hierarchy becomes enormous, so there is a limit to application to large-scale data.
Although TreeMiner is claimed to be a depth-first search, in reality, all the branch candidates that occur in the path direction connecting leaf nodes from the root of the tree (tree) structure are the depth of recursion processing. To feed in the direction. For this reason, the same problem as the breadth-first search occurs, and when applied to large-scale data, there is a problem that the number of pattern candidates becomes enormous and the processing cannot be performed.

In MB3-Miner and TreeMiner, the appearance position of a tree (correspondence between each node in the pattern tree and a node in the data tree) is managed. In the case of embedded subtree mining, There was a problem that the number of descendant nodes expanded in the depth direction expanded exponentially. In TreeMiner, a countermeasure against this is taken in TreeMinerD, but in reality it does not function sufficiently.
A similar situation does not occur only in the operation history of a document. For example, even in protein structure data, nodes having the same label often appear in a series of times. Regarding the management of the combination of appearance positions, a large amount of information must be managed in such a case, and there arises a problem that the processing capacity increases due to an increase in storage capacity necessary for processing and data to be processed. This makes it difficult to process large-scale data with realistic time and resources.
That is, conventionally, a search is performed by converting a tree structure into a string format or converting it into an equivalent one. In the present embodiment, the tree structure itself is searched.

In this embodiment, a relational structure that appears multiple times (a subtree pattern, hereinafter also referred to as a structure pattern, a structure pattern tree, or a pattern) is extracted from a data group that can handle the relationship set between elements as a tree structure. It is related to the technology.
This embodiment uses data of a path from a root node (root) to a leaf node (leaf) (hereinafter also referred to as “Epath”) and an Epath range to which each node belongs (hereinafter also referred to as “EpRange”), and a search step This is an information extraction device that performs a search process by a depth-first search using a node appearance information management mechanism that is managed in accordance with.
Furthermore, in order to increase the efficiency of pattern extraction, a tree data management mechanism using the EpRange is provided.

Hereinafter, an example of a preferred embodiment for realizing the present invention will be described with reference to the drawings.
FIG. 1 is a conceptual configuration diagram of a system suitable for applying the present embodiment. This system has a structure information DB 110, an information collection device 120, an information extraction device 130, and an extraction information management device 140. These are connected via a communication line. Note that all these device groups may be built in one device, or some of these devices may be built in one device.

The structure information DB 110 is connected to the information collection device 120 and the information extraction device 130, and is data received from the information collection device 120. The structure information DB 110 stores and manages tree structure data from which information is to be extracted. Accessed from the extractor 130.
The information collection device 120 is connected to the structure information DB 110, collects information from other devices (not shown), or receives information transmitted from other devices (not shown), and shapes the information (information extraction device) if necessary. (Converted into tree-structured data that can be handled by 130) and stored in the structure information DB 110.
The information extraction device 130 is connected to the structure information DB 110 and the extraction information management device 140, accesses the structure information DB 110, extracts frequent information from the tree structure data, and transmits it to the extraction information management device 140.
The extracted information management device 140 is connected to the information extracting device 130, and receives and accumulates frequent information transmitted from the information extracting device 130 or transmits it to other devices (not shown) such as a display device and a printing device. .

FIG. 2 is a conceptual module configuration diagram of a configuration example in the information extraction apparatus 130 according to the present embodiment.
The module generally refers to components such as software (computer program) and hardware that can be logically separated. Therefore, the module in the present embodiment indicates not only a module in a computer program but also a module in a hardware configuration. Therefore, the present embodiment also serves as an explanation of a computer program, a system, and a method. However, for the sake of explanation, the words “store”, “store”, and equivalents thereof are used. However, when the embodiment is a computer program, these words are stored in a storage device or stored in memory. It is the control to be stored in the device. In addition, the modules correspond almost one-to-one with the functions. However, in mounting, one module may be composed of one program, or a plurality of modules may be composed of one program. A plurality of programs may be used. The plurality of modules may be executed by one computer, or one module may be executed by a plurality of computers in a distributed or parallel environment. Note that one module may include other modules. In the following, “connection” includes not only physical connection but also logical connection (data exchange, instruction, reference relationship between data, etc.).
In addition, the system or device is configured by connecting a plurality of computers, hardware, devices, and the like by communication means such as a network (including one-to-one correspondence communication connection), etc., and one computer, hardware, device. The case where it implement | achieves by etc. is also included. “Apparatus” and “system” are used as synonymous terms.

  As shown in FIG. 2, the information extraction device 130 shown in FIG. 1 includes a structure information management module 210, an appearance information selection module 220, an appearance information management module 230, a survey range processing module 240, a search processing module 250, and a search state management module. 260 and an extraction information processing module 270.

  The structure information management module 210 is connected to the appearance information selection module 220, the appearance information management module 230, and the investigation range processing module 240, and investigates the structure information in the structure information DB 110 according to the designation of the investigation range processing module 240. The appearance of nodes in the tree structure data is totaled for each label. Depending on the configuration, appearance position information may be collected for each label. Appearance information that is a result of the aggregation is transmitted to the appearance information management module 230. It should be noted that the appearance information includes both information on the appearance location and a total value indicating how many trees have appeared. Further, the structure information management module 210 may include a storage unit (not shown) if necessary, and may convert the structure information in the structure information DB 110 into a data structure suitable for processing and store it.

  The appearance information selection module 220 is connected to the structure information management module 210 and the appearance information management module 230, and selects a target label based on the appearance information generated by the structure information management module 210. In other words, the appearance information for each label of the node whose appearance is confirmed by the structure information management module 210 is received, and an element that satisfies a predetermined criterion (such as one that appears more than a preset number of times) and an element that does not satisfy it are selected. Alternatively, the appearance information is arranged by discarding appearance information that does not meet the conditions specified by the user using an input device (not shown).

  The appearance information management module 230 is connected to the appearance information selection module 220, the search processing module 250, and the structure information management module 210, and receives and manages the label and appearance information of the node selected by the appearance information selection module 220. This information is sequentially transmitted to the search processing module 250 at the request of the search processing module 250.

  The search range processing module 240 is connected to the structure information management module 210, the search processing module 250, and the search state management module 260, and is generated by the search state and structure information management module 210 stored in the search state management module 260. Based on the appearance information stored in the search state management module 260 after copying or moving from the appearance information management module 230 according to the instruction of the search processing module 250, the search range of the structure pattern in the tree structure data is determined. To do.

  The search processing module 250 is connected to the appearance information management module 230, the investigation range processing module 240, the search state management module 260, and the extracted information processing module 270. The appearance information management module 230, the investigation range processing module 240, and the search state management. The control by the module 260 and the extraction information processing module 270 is controlled, the structure information management module 210 takes out based on the search range determined by the survey range processing module 240, and the label and appearance selected by the appearance information selection module 220 Information is obtained from the appearance information management module 230, a structure pattern that appears multiple times in the tree structure data is searched, and the search result is passed to the extraction information processing module 270. This search process is performed recursively by updating the search state of the search state management module 260 in a timely manner.

The search state management module 260 is connected to the search range processing module 240 and the search processing module 250, and stores and manages the intermediate state of the search process. The stored intermediate state is used for recursive search processing by the search processing module 250. Responsible for pattern candidate information during search processing by the search processing module 250, retention of appearance information in each structure information of nodes in the structure pattern, and updating / recovering appearance information in the process.
The extracted information processing module 270 is connected to the search processing module 250, receives the pattern extracted from the search processing module 250, stores it in a storage device (not shown), and takes charge of output processing to an output device (not shown).

FIG. 3 is a flowchart showing an example of processing according to this embodiment.
In step S302 (preparation of scanning information), the structure information management module 210 converts the structure information stored in the structure information DB 110 to prepare for efficiently executing the subsequent processing, and is not shown. Store in storage means.

FIG. 8 is an example of tree structure data used for explanation.
The three tree structure data Tr1 (FIG. 8A), Tr2 (FIG. 8B), and Tr3 (FIG. 8C) illustrated have 16 nodes, 16 nodes, and 11 nodes, respectively. For simplification of explanation, it is assumed that there is an order relationship between siblings, and identifiers such as v0, v1,... Are set to the nodes based on the order when the nodes are traced by the depth-first search. Thereafter, when it is not necessary to indicate the tree structure, the node is simply indicated by a notation such as v0 or v2, and when it is necessary to indicate which tree data, each node is indicated as v0 _Tr1 or v2 _Tr2. It is described with the identifier of the tree data to which the belongs.
In the example shown in FIG. 8, the labels of the nodes are A, B, C,. The labels of the nodes appearing in the example are A, B, C, D, E, F, G, H, I, J, and K. The alphabet in the circle in FIG. 8 is the label of the node.

Also, the pattern extraction condition will be described as a pattern appearing in two or more tree structure data.
The example shown in FIG. 9 is a diagram showing a path from each root node to a leaf node for each tree structure data. In the figure, identifiers Ep1 to Ep6 are attached to the respective paths. Similarly, when it is necessary to indicate a tree, it is described with an identifier of the tree data, such as Ep1 _Tr1 . For example, Ep4 _Tr1 (Ep4 of _Tr1 ) is a path that passes through v0, v10, v11, v12, and v13, and Ep2 _Tr3 (Ep2 of Tr3) is a path that passes through v0, v1, v2, v3, v4, and v6. .
By setting Epath in this way, each node is on at least one Epath, and some nodes are on multiple Epaths. For example, v12 _Tr1 is on Ep4 _Tr1 and Ep5 _Tr1 , and v7 _Tr2 is on Ep3 _Tr2 , Ep4 _Tr2, and Ep5 _Tr2 .

Information EpRange indicating on which EP Path each node is located is defined as a function of each node. For _example, a _{EpRange (v12 Tr1) = {Ep4} Tr1, Ep5 Tr1}, EpRange (v7 Tr2) = a _{{Ep3 Tr2, Ep4 Tr2, Ep5} Tr2}.
Here, since the target tree structure assumes an ordered tree, the EPath numbers appearing in EpRange are consecutive. Therefore, for simplification of description, EpRange is simply indicated by the smallest and largest EPPath numbers. That is, it is expressed as follows. For example, EpRange (v12 _Tr1 ) = [4, 5] and EpRange (v7 _Tr2 ) = [3, 5].
At this time, the number on the smaller side of EpRange is represented by EpRangeL, and the number on the larger side is represented by EpRangeR. For _{example, EpRangeL (v12 Tr1) = 4} , EpRangeL (v7 Tr2) = a _{3, EpRangeR (v12 Tr1) =} 5, EpRangeR (v7 Tr2) = 5.

In addition, EpRange is not referred to as a function of a node but simply as a range of EPath. For _example, when you want a _{{Ep2 Tr3, Ep3 Tr3, Ep4} Tr3}, simply performed also indicate a _EpRange Tr3 [2,4] of the Tr3. Similarly, EpRangeL is used as a number with a small EpRange, and EpRangeR is used as a number with a large EpRange.

  FIG. 10 shows a set of (node identifier, label, EpRangeL, EpRangeR) for each node in Tr1, Tr2, and Tr3 in the example shown in FIG. In FIG. 10A, a node identifier 1001a, a node label 1002a, EpRangeL1003a, and EpRangeR1004a correspond to each. For example, the node identifier V0 of Tr1 shown in FIG. 10A is a label of the node A, EpRangeL is 1, and EpRangeR is 6.

  Since the structure information management module 210 has a function of scanning a node by designating a node existence range or EpRange range for each tree structure data, necessary information in each tree information is stored in an internal storage means (not shown). Can be stored in a separate data structure.

  FIG. 11, FIG. 12, and FIG. 13 show an example in which tree information is stored inside. 11, 12, and 13 are examples in which each node is stored so that it can be easily traced from the value of EpRangeL for each of Tr1, Tr2, and Tr3. It can be configured by adding. That is, for example, the Ep information 1100 of Tr1 shown in FIG. 11 includes link information to Ep1 1101, Ep2 1102, Ep3 1103, Ep4 1104, Ep5 1105, and Ep6 1106. In addition, for convenience of illustration, 1101, 1102, 1103, 1104, 1105, 1106 are individually shown, but these may be arranged in a continuous area in the order shown in FIG. 10.

The data corresponding to each node in the tree information examples shown in FIGS. 11, 12, and 13 is added to the rightmost number in addition to the set of (node identifier, label, EpRangeL, EpRangeR) shown in FIG. ing. This number is a number indicating the position of each data in the same EpRangeL. This number is for use when searching for the next lower data with the same EpRangeL.
For example, for Tr2, if it is desired to enumerate the nodes above EPath3 and EPath4, it can be realized simply by tracing the nodes whose EpRangeL is 3 or 4 and searching for and listing those whose EpRangeR is not greater than 5. However, this is not the only node on Ep3 and Ep4, and v0 and v1 whose EpRangeL is Ep1 are left. However, nodes such as v0 and v1 can also be easily enumerated by an additional process described later.
In the present embodiment, description will be made using the data structure shown in FIGS. 11, 12, and 13 as an example of a method for efficiently enumerating nodes by specifying EpRange. However, other data management methods such as buckets are used. A method using division or a method using a search tree for searching may be applied.

In step S304 (aggregation of element occurrences), the structure information management module 210 aggregates how many trees each label has appeared. This process can also be executed at the same time as the above-described step S302. As a result of processing, information that can determine how many tree-structured data each label appears is obtained, and a list of frequent elements (node labels in this embodiment) that leaves only those that meet the pattern extraction conditions create. Depending on the configuration, the appearance position information (hereinafter also simply referred to as appearance information) of each frequent element is also processed.
In this example, A, B, C, D, and F remain in the frequent element list. In accordance with the list of frequent elements, a process of extracting a structure pattern having each frequent element in the list as a root node is performed in the subsequent iterative process.

  In step S306 (Frequent element processing end?), It is checked that the processing has been repeated for each of the labels (A, B, C, D, and F shown in the example) of the nodes in the frequent element list. Control the flow. When the processing is completed for all the labels in the frequent element list, the processing ends (step S314).

  In step S308 (processing target element selection), the appearance information selection module 220 selects one from the unprocessed elements in the frequent element list. In the example, one unprocessed item is selected from the node labels A, B, C, D, and F. Since the process flow is the same regardless of which label is selected, the case where label A is selected will be described below.

In step S310 (search state creation), the search state management module 260 prepares information indicating where the selected label appears in each structure data.
In the case where the appearance position information of each label is also stored in the appearance information management module 230 in step S304, this step S310 is a process of only extracting the information. Depending on the configuration, the appearance position information may not be stored in the appearance information management module 230 due to the storage capacity and the like. In such a case, the structure information is scanned at this stage to obtain the appearance position information. You may make it extract.
For example, as shown in FIG. 14, the appearance information of the label A is v1 _Tr2 , v3 _Tr2 , v9 _Tr2 in the appearance information 1420 of v0 _Tr1 , v2 _Tr1 , v4 _Tr1 , v12 _Tr1 , Tr2 in the appearance information 1410 of _Tr1 . , V11 _Tr2 , and Tr3 appearance information 1430 are v0 _Tr3 and v8 _Tr3 . Here, the appearance information shown in FIG. 14 is the same as the information stored in FIG. 11, FIG. 12, and FIG. That is, (1) “node identifier”, (2) “label”, (3) “EpRangeL”, (4) “EpRangeR”, (5) “EpRangeL” in the information stored as the same number It is shown as a pair.

  In addition to the appearance information, a structure pattern tree having the selected label as a root is created, a current processing position in the structure pattern tree is set, and a search state node shown in the example of FIG. 15 is created. That is, the search state node has a structure pattern tree 1510, appearance information 1520, and change information stack 1530. The structure pattern tree 1510 stores the structure pattern tree having the label node selected in step S308 as a root and the current processing position in the structure pattern tree. Tr1 1521, Tr2 1522, and Tr3 1523 in the appearance information 1520 are the same as the appearance information 1410, the appearance information 1420, and the appearance information 1430 shown in FIG. The change information stack 1530 is used in processing described later.

Using this search state node as a root node, a search state is created as shown in the example of FIG. 16 and stored in the search state management module 260. FIG. 16 shows a search state node (A) 1610 that shows the structure of the pattern corresponding to the structure pattern tree stored in the search state node as a character string.
Here, the notation of the character string of the structure pattern tree is shown by recursively applying the notation {parent} [{child 1}, {child 2},...]. However, [] is omitted for nodes having no children.

In step S312, the search processing module 250 performs a frequent structure pattern search. Step S312 will be described with reference to the flowchart shown in FIG.
Note that step S310 (search state creation) is performed here before step S312 (frequent structure pattern search), but step S310 can be incorporated into the processing of step S312 depending on the configuration.

FIG. 4 is a flowchart showing a processing example of a frequent structure pattern search according to this embodiment. Two types of searches are performed here. That is, a lower search (step S404) that is the first search and a lateral search (step S410) that is the second search.
In the lower search process (step S404), a structure pattern that appears multiple times in a plurality of tree structure data is searched for a node lower than the current processing target node in the tree structure data. .
In the lateral branch search process (step S410), the search for the structure pattern is a node that is higher than the current processing target node in the tree structure data and is lower than the higher node. And for each unsearched node.
In the lower search process and the lateral branch search process, when there are no more nodes to be searched, the process returns to the original node that started the search.

  In step S402 (structure pattern information output), the structure pattern tree data of the search state node at that time is output. It may be determined whether or not to output in accordance with a predetermined standard before this output.

In step S404 (subordinate search), a partial structure that frequently appears in the descendant direction is searched. When this sub-search is performed, the node in the tree structure data that matches the current processing target node in the structure pattern (the highest level for nodes having a vertical relationship between the nodes) The sub-search range is determined based on the determined range, and the sub-search process is performed based on the determined range. Step S404 will be described with reference to the flowchart shown in FIG.
In addition to the subordinate search, a side branch search (step S410) repetitive process for searching for a partial structure in the horizontal direction from the parent node and the ancestor node is performed. When this traverse branch search is performed, the node in the tree structure data that matches the parent node in the structure pattern (the highest-order node if there is a vertical relationship between the nodes) and the child in the structure pattern The search range is determined based on the node in the tree structure data that matches the node (the lowest node when there is a vertical relationship between the nodes), and the lateral branch search processing is performed in the determined range. Search based on
It should be noted that step S406 (there is a horizontal branch search location remaining?) And step S408 (horizontal branch search location selection) during this iterative process are easier to explain if there are multiple nodes in the pattern tree. The explanation will be given later. Step S410 (horizontal branch search) will also be described later.

FIG. 5 is a flowchart showing an example of a low-order search process according to this embodiment.
In step S502 (preparation of lower search range information), a search range of descendant nodes is prepared. In this step, specifically, a range for searching for descendant nodes is designated for each structure data. Here, an example in which the designation is performed by EpRange is shown.
A node in the pattern is selected by label A, and the search state node at that time is as shown in FIG. At this time, for example, the candidate of the offspring of label A in Tr1 is a descendant of v0 _Tr1, a descendant of v2 _Tr1, a descendant of v4 _Tr1 , or a descendant of v12 _Tr1 based on Tr1 1521.

The descendants of v0 _Tr1 are in the range EpRange (v0 _Tr1 ) = [1,6], the descendants of v2 _Tr1 are in the range EpRange (v2 _Tr1 ) = [1,3], and the descendants of v4 _Tr1 are EpRange (v4 in the range of _{Tr1) = [1,2], v12} Tr1 progeny is present in the range of _{EpRange (v12 Tr1) = [4,5} ].

Nodes that satisfy all of the following conditions ((1) and (2)) have a node with label A as an ancestor.
_{(1) EpRange (v0 Tr1)} = [1,6], EpRange (v2 Tr1) = [1,3], EpRange (v4 Tr1) = [1,2], EpRange (v12 Tr1) = [4,5] It is in the range of any EpRange.
(2) If none of the EpRanges of v0 _Tr1 , v2 _Tr1 , v4 _Tr1 , and v12 _Tr1 is wider than the EpRange of the node, a higher one exists having the same EpRange.
However, the above-mentioned two conditions are conditions for one method when enumerating descendant nodes without duplication in the structure information management method used in the present embodiment. Other methods may be used as long as all descendant node candidates can be enumerated from the appearance information corresponding to the pattern.

One method for enumerating descendant nodes according to the above conditions is shown.
First, for each structure data (Tr1, Tr2, Tr3 in the example), the EpRange inclusion relation of the appearance data is checked, and only the data not included in other EpRanges is left. In addition, if EpRange is the same, the thing with the smaller number assigned between things with the same EpRangeL is left. By this process, v0 _Tr1 remains in _Tr1 , v1 _Tr2 remains in _Tr2 , and v0 _Tr3 remains in _Tr3 .

The descendant node is searched based on the appearance information.
The search example of descendant nodes from node A is all nodes below v1 _Tr1 in Tr1, all nodes below v2 _Tr2 in Tr2, and all nodes below v3 _{Tr3 in Tr3} .
In this scanning process, appearance information is totaled and collected for each label in step S504 (aggregation of element occurrences).
At this stage, labels that are determined to appear frequently are A, B, C, D, and F.

Next, the process of selecting each of these frequent labels as a process target element is repeated in step S506 (end frequent element process?).
For example, description will be made assuming that B is selected in step S508 (selection of processing target element). As shown in FIG. 17, the appearance information of B at this time is v2 _Tr2 , v4 _Tr2 in the appearance information 1720 of v1 _Tr1 , v3 _Tr1 , v6 _Tr1 , v11 _Tr1 , v13 _Tr1 , Tr2 in the appearance information 1710 of _Tr1 . , V7 _Tr2 , v12 _Tr2 , Tr3 appear in the appearance information 1730 as v2 _Tr3 , v10 _Tr3 .

Then, the process of step S510 (search state creation / update) is performed. In other words, here, from the appearance location of the ancestor node of the current processing target node in the structure pattern, the node corresponding to the node including the node of the current processing target node appearance as the descendant The appearance location is kept.
A pattern with B as a child node is registered under the root node A of the structure pattern tree, and the current position information in the structure pattern tree indicates the position of B. In addition, appearance information 1710, appearance information 1720, and appearance information 1730 shown in FIG. 17 are registered as appearance information, and a search state node is created and registered in the search state management module 260 as shown in the example of FIG. That is, the search state node shown in the example of FIG. 18 includes a structure pattern tree having a label B node below a label A node in the structure pattern tree 1810 and current position information in the structure pattern tree ( (Node of label B) is stored, Tr1 1821, Tr2 1822, Tr3 1823 similar to those shown in FIG. 17 are stored in appearance information 1820, and change information 1831 is stored in change information stack 1830 is doing.

  The search state here is as shown in the example of FIG. That is, in the search state shown in the example of FIG. 19, the search state node (A [B]) 1920 is connected under the search state node (A) 1910.

Then, the search state is updated. Specifically, the upper search state node is traced in order from the registered search state node, and the appearance information that does not include any appearance position registered in the lower search state node is temporarily deleted and stored in the change information stack 1830. To do.
In the state of this example, the lower search state node is A [B], the upper search state node is A, and the search state node A is one that does not include any appearance information of the search state node A [B]. Is stored in the change information stack 1830. Here, since there is no appearance information corresponding to this condition in the search state node A, nothing is done. If the state of the ancestor direction tree in the structure pattern tree is updated, the updated appearance state node information is used as the search state node (in this case, the search state node A [B ]) In the change information stack 1830.

Then, the process recursively enters step S512 (frequent structure pattern search).
In the frequent structure pattern search process, the structure pattern tree information A [B] is output in accordance with a predetermined condition in the structure pattern information output process. Next, the lower search is similarly performed.
In the next lower search process, the search range from the appearance information of B searches for descendants of the lower node. In other words, Tr1 is a search for descendants of v1 _Tr1 and v11 _Tr1 , Tr2 is a descendant of v2 _Tr2 and v7 _Tr2 , and Tr3 is a search for descendants of v2 _Tr3 .
When searching in this range, for example, v10 _Tr1 , v15 _Tr1 , v6 _Tr2 , v11 _Tr3, etc. will not enter the search range, the search range will be narrowed, but at this stage the frequent nodes are still A, B, C , D, F.

Next, assume that the node of label C is selected here. Then, the appearance information of C becomes v8 _Tr1 , v14 _Tr1 , v5 _Tr2 , v8 _Tr2 , v13 _Tr2 , v3 _Tr3 , v9 _Tr3 . As a result, the search state node becomes as shown in FIG. 20, and the search state becomes as shown in FIG. That is, the search state node shown in the example of FIG. 20 has a structure pattern tree in which a structure pattern tree 2010 has a label B node under the label A node and a label C node under the label B node. And the current position information (node of label C) in the structure pattern tree, Tr1 2021, Tr2 2022, and Tr3 2023 are stored in the appearance information 2020, and the change information stack 2030 is changed. Information 2031 is stored. In the search state shown in FIG. 21, the search state node (A [B]) 1920 is connected under the search state node (A) 1910, and the search state node (A [B]) 1920 is connected under the search state node (A [B]) 1920. [B [C]]) 2130 is connected.

Then, the upper one of the search state nodes is updated. The appearance information 2020 of C shown in FIG. 20 is not included in the lower order, that is, EpRange is not greater than or equal to any of the appearance information of Tr1 2021, Tr2 2022, and Tr3 2023 of C of FIG. The larger one is added to the change information 2031 and stored in the change information stack 2030.
In this case, v6 _Tr1 , v13 _Tr1 , v4 _Tr2 , v12 _Tr2 , and v10 _Tr3 are registered in the change information 2031 and stored in the change information stack 2030. As a result, the search state node A [B] is updated as shown in FIG. That is, the search state node shown in the example of FIG. 22 includes a structure pattern tree having a label B node under a label A node in the structure pattern tree 2210 and current position information in the structure pattern tree ( Node B), Tr1 2221, Tr2 2222, and Tr3 2223 are stored in the appearance information 2220, and the change information 2231 and the change information 2232 are stored in the change information stack 2230. Then, the updated information is stored in the change information 2231 of the search state node A [B [C]] that has caused the update.

Then, since the lower search state node A [B] has been updated, an update check of the upper search state node A is performed. In this case, v4 _Tr1 , v12 _Tr1 , v3 _Tr2 , v9 _Tr2 , v11 _Tr2 , and v8 _Tr3 out of the appearance information do not have B corresponding to the lower order, and thus are removed from the appearance information and moved to the change information stack. Then, the updated information is stored in the change information of the search state node A [B [C]] that has caused the update. The update result of the search state node A is shown in FIG. That is, the search state node shown in the example of FIG. 23 stores the structure pattern tree having the node of label A in the structure pattern tree 2310 and the current position information (node of label A) in the structure pattern tree. Tr1 2321, Tr2 2322, and Tr3 2323 are stored in the appearance information 2320, and the change information 2331 and the change information 2332 are stored in the change information stack 2330. Then, the updated information is stored in the change information 2231 of the search state node A [B [C]] that has caused the update.

  Next, the frequent structure pattern search is continued (step S512). Even at this stage, frequent labels can be found under the node having label C, but the same processing flow is repeated, so that the recursive execution of the sub-search process has proceeded to the next process. Will be explained.

The processing (step S406 to step S410) after step S406 (there is a horizontal branch search portion remaining?) In the flowchart shown in FIG. 4 is processing for tracing the search state node in the upper direction of the search state. Tracing in the direction from the newest registered search state node to the parent search state node, the side branch search process (step S410) is executed for each parent-child relationship pair.
In this example, the search state is composed of three search state nodes as shown in FIG. 21, and a search state node (A [B]) 1920 is connected under the search state node (A) 1910. A search state node (A [B [C]]) 2130 is connected under the search state node (A [B]) 1920, and a two-stage parent-child relationship is configured. Therefore, this loop is executed for the parent-child relationship between the search state node A [B] and the search state node A [B [C]] and the parent-child relationship between the search state node A and the search state node A [B]. Is done.
This processing is preferably performed in order from the lower order to the higher order of the search state node. Here, the process for the parent-child relationship between the search state node A [B] and the search state node A [B [C]] will be described first. In other words, it is assumed that the parent-child relationship of the search state node is selected as the horizontal branch search location (step S408).

Then, the process enters step S410 (horizontal branch search).
FIG. 7 shows a flowchart of a process example of the lateral branch search.
In the horizontal branch search process, first, the process of step S702 (preparation of horizontal branch search range information) is executed. Specifically, this processing is performed between the parent-child relationship of the selected search state node, and in this example, the traverse search between the search state node A [B] and the search state node A [B [C]]. Prepare range information.

  This process selects, from the appearance information of the upper search state node, one that EpRange is not included in the EpRange of the other appearance information, and for each EpRange, from the appearance information of the lower search state node to the upper EpRange, EpRange Is selected (hereinafter referred to as the lower leftmost EpRange), and the selected subordinate EpRange EpRangeR (the lowermost EpRange EpRange) and the upper EpRange EpRange Calculate the range.

For example, in Tr1, EpRange of v3 _Tr1 is included in EpRange of v1 _Tr1 among the appearance information of the upper search state node (in the case of the same EpRange, the one with the smallest right number includes the one with the larger number) However, since EpRange is the same, the reverse will not affect the result). Therefore, the EpRange of the upper search state node is the EpRange of v1 _Tr1 and v11 _Tr1 . The appearance information of the lower search state node corresponding to v1 _Tr1 is EpRange of v8 _Tr1 . Accordingly, the upper EpRange in this case is EpRange (v1 _Tr1 ) = [1, 3], and the lower leftmost EpRange is EpRange (v8 _Tr1 ) = [2, 2]. As a result, the search range for v1 _Tr1 is [2 + 1, 3], that is, EpRange [3, 3].

For the EpRange of the other v11 _Tr1 , the appearance information of the lower search state node is the EpRange of v14 _Tr1 . The upper EpRange is EpRange (v11 _Tr1 ) = [4, 5], and the lower leftmost EpRange is EpRange (v14 _Tr1 ) = [5, 5]. As a result, the search range for v11 _Tr1 is [5 + 1, 5], which is an inconsistent range, so there is no corresponding search range. As a result, the search range for Tr1 is only EpRange [3, 3]. Similarly, for Tr2, EpRange [5: 5] is the search range of the horizontal branch between the upper, v7 _Tr2 EpRange, and the corresponding lower leftmost EpRange, v13 _Tr2 EpRange. For Tr3, EpRange [4: 4] is the search range of the horizontal branch between the upper, v2 _Tr3 EpRange, and the corresponding lower leftmost EpRange, v9 _Tr3 EpRange.

  Each tree structure data is searched according to these search ranges (steps S704 to S714). In the present embodiment, a method using the data structure of FIGS. 11, 12, and 13 is shown as one method for efficiently performing this search process. This search process may be performed using other data structures and processing methods. However, by using at least this method, it is possible to easily realize a search for a node specifying the range of EPath from each tree structure data. it can.

When searching from Ep1 with EpRange [3, 3], only v9 _Tr1 is the search range. The label that appears is F. When searching from Ep2 with EpRange [5, 5], only v14 _Tr2 and v15 _Tr2 are search ranges. Appearing labels are G and F. When searching from Ep3 with EpRange [4, 4], only v10 _Tr2 is the search range. The label that appears is B. The only frequent label at this time is F (step S704).

The label F is selected (step S708), and search state creation / update processing (step S710) is performed. Also, in step S710, from the appearance location of the ancestor node of the current processing target node in the structure pattern, the node including the node of the current processing target node appearance as a descendant It keeps the appearance part to do.
A search state node is created using the appearance information of F shown in FIG. 24 and registered in the search state. The search state at this time is shown in FIG. That is, the search state node shown in the example of FIG. 24 has a node of label B under the node of label A and a node of label C and a node of label F under the node of label B in the structure pattern tree 2410. And the current position information (the node of the label F) in the structure pattern tree are stored, Tr1 2421 and Tr2 2422 are stored in the appearance information 2420, and the change information stack 2430 is stored. The change information 2431 is stored. 25 includes four search state nodes. A search state node (A [B]) 1920 is connected under the search state node (A) 1910, and the search state node ( A search state node (A [B [C]]) 2130 and a search state node (A [B [CF]]) 2540 are connected under A [B]) 1920. At this time, the search state node A [B [CF]] is related as a child node of the search state node A [B] selected as the parent-side search state node during the search for the horizontal branch.

Next, the search state node is updated by examining the effect of registering the search state node A [B [CF]].
At this time, the search state node that checks whether or not the update is necessary is a search state node that is an ancestor of the search state node A [B [CF]]. Specifically, the search state node A [B] and the search state node A. In this case, the search state node is not updated because it is found that the search state node A [B] is not updated at the stage of inspection.

Then, the process proceeds to a frequent structure pattern search (step S712).
In the frequent structure pattern search, an output process is performed on the pattern candidate A [B [CF]] if necessary, and a low-order search process is performed.
In the lower search here, since there is no frequent label, the process immediately returns from the lower search process.
Next, the iterative search process is repeated. The parent-child relationship of the search state nodes that are candidates for the lateral branch search at this time is the relationship between the search state node A [B [CF]] and the search state node A [B], the search state node A and the search state node A There are two relationships between [B].
The flow of this process has already been described.
However, labels that appear frequently in any parent-child relationship of any search state node do not appear here.

  Then, the processing in the search state node A [B [CF]] is completed. As a result, the frequent structure pattern search process corresponding to the search state node A [B [CF]] is completed, and the process returns to the upper process flow. In this case, the process returns to the horizontal branch search process. Then, a search state recovery process (step S714) of the next step is performed.

The search state recovery process (steps S714 and S514) will be described.
The change information stack of the search state node A [B [CF]] is referred to. The updated search state nodes A [B] and A are recorded here.
For both search state nodes, search state node recovery processing is performed based on the change information at the top of the search state stack. Specifically, in the case of the search state node A [B], the appearance information v6 _Tr1 , v13 _Tr1 , v4 _Tr2 , v12 _Tr2 , v10 _Tr3 is returned as the appearance information. The returned result is the same as that shown in FIG. By performing the same processing, the search state node A becomes the same as that shown in FIG.
As described above, by executing the processing recursively, it is possible to extract all ancestor-descendant tree structures that are equal to or greater than the number of appearances specified in advance.

The descendant node scanning process will be described with reference to the flowchart shown in FIG.
In step S602, it is determined whether or not the upper appearance information has been processed. In this determination, if it is determined that the process is complete, the descendant node scanning process is terminated (step S612). If it is determined that the process is not complete, the process proceeds to step S604.
In step S604, one piece of appearance information is selected.

In step S606, EpRangeL and EpRangeR of the appearance information selected in step S604 are set to EpRangeLP and EpRangeRP, respectively.
In step S608, scanning is performed from the next number of the appearance information for EpRangeL of the appearance information that is the same as EpRangeLP.
In step S610, scanning is performed in a range from EpRangeLP + 1 to EpRangeRP where the value of EpRangeR is equal to or less than EpRangeRP.

  In the example shown in the present embodiment, all occurrence information with the same EpRange is handled equally, but there is one occurrence information with the same EpRange (that is, a range without a branch) (that is, the highest and lowest). Nodes) can be treated as representatives, and the same effect can be obtained. The amount of appearance information to be stored is reduced, the required storage capacity is reduced, and the cost of processing such as inspection is further reduced. . In other words, the appearance locations other than the highest and lowest positions in the range without branching in the tree structure may be deleted from the search state management module 260. Alternatively, it may be selected before being stored in the search state management module 260, and further, this selection may be performed at the time of storage in the structure information DB 110.

Further, a method is known in which a tree-structured node is depth-firstly searched, a node is numbered, and the range of descendant nodes of the node is indicated by the number of the node. For example, in the v2 of Tr2 shown in FIG. 7, this information called scope is expressed as [7, 15] using the number 7 of its own and the number of the node having the largest number among the descendant nodes.
In this embodiment, EpTree is used to search for a node by specifying a range from a tree structure. However, a configuration in which a search is performed using this scope information instead can be easily changed by those skilled in the art. . Since the scope information is information indicating the range of descendant nodes of each node, the range of descendants from a certain node can be determined. In addition, the scope information of the higher-order appearance information and the scope information of the lower-order appearance information can be calculated even when searching for the horizontal branch. If this is applied to the corresponding part of this embodiment, the search range of the horizontal branch is also It can be determined using the scope information.

  Note that the hardware configuration of a computer on which the program according to the present embodiment is executed is a general computer as illustrated in FIG. 26, specifically, a personal computer, a computer that can be a server, or the like. CPU 2601 for executing programs such as structure information management module 210, appearance information selection module 220, appearance information management module 230, survey range processing module 240, search processing module 250, search state management module 260, extraction information processing module 270, etc. A RAM 2602 for storing programs and data, a ROM 2603 for storing programs for starting up the computer, an HD 2604 as an auxiliary storage device (for example, a hard disk can be used), and data such as a keyboard and a mouse are stored. An input device 2606 for input, an output device 2605 such as a CRT or a liquid crystal display, and a communication line interface 2607 for connecting to a communication network (for example, using a network interface card) DOO can), and, and a bus 2608 for exchanging data by connecting them. A plurality of these computers may be connected to each other via a network.

Among the above-described embodiments, the computer program is a computer program that reads the computer program, which is software, in the hardware configuration system, and the software and hardware resources cooperate with each other. Is realized.
Note that the hardware configuration illustrated in FIG. 26 illustrates one configuration example, and the present embodiment is not limited to the configuration illustrated in FIG. 26, and is a configuration capable of executing the modules described in the present embodiment. I just need it. For example, some modules may be configured by dedicated hardware (for example, ASIC), and some modules may be in an external system and connected via a communication line. A plurality of systems shown in FIG. 5 may be connected to each other via communication lines so as to cooperate with each other. In particular, in addition to personal computers, information appliances, copiers, fax machines, scanners, printers, and multifunction machines (image processing apparatuses having two or more functions of scanners, printers, copiers, fax machines, etc.) Etc. may be incorporated.

The program described above may be provided by being stored in a recording medium, or the program may be provided by communication means. In that case, for example, the above-described program may be regarded as an invention of a “computer-readable recording medium recording the program”.
The “computer-readable recording medium on which a program is recorded” refers to a computer-readable recording medium on which a program is recorded, which is used for program installation, execution, program distribution, and the like.
The recording medium is, for example, a digital versatile disc (DVD), which is a standard established by the DVD Forum, such as “DVD-R, DVD-RW, DVD-RAM,” and DVD + RW. Standards such as “DVD + R, DVD + RW, etc.”, compact discs (CDs), read-only memory (CD-ROM), CD recordable (CD-R), CD rewritable (CD-RW), etc. MO), flexible disk (FD), magnetic tape, hard disk, read only memory (ROM), electrically erasable and rewritable read only memory (EEPROM), flash memory, random access memory (RAM), etc. It is.
The program or a part of the program may be recorded on the recording medium for storage or distribution. Also, by communication, for example, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a wired network used for the Internet, an intranet, an extranet, etc., or wireless communication It may be transmitted using a transmission medium such as a network or a combination of these, or may be carried on a carrier wave.
Furthermore, the program may be a part of another program, or may be recorded on a recording medium together with a separate program. Moreover, it may be divided and recorded on a plurality of recording media. Further, it may be recorded in any manner as long as it can be restored, such as compression or encryption.

It is explanatory drawing which shows the conceptual structural example of the system which applies this Embodiment suitably. It is a conceptual module block diagram about the structural example of this Embodiment. It is the flowchart which showed the example of a process by this Embodiment. It is the flowchart which showed the processing example of the frequent structure pattern search by this Embodiment. It is the flowchart which showed the example of the process of the low-order search by this Embodiment. It is the flowchart which showed the example of a scanning process of the descendant node by this Embodiment. It is the flowchart which showed the process example of the lateral branch search by this Embodiment. It is explanatory drawing which shows the example of the tree structure data made into object. It is explanatory drawing which shows the example of Epath. It is explanatory drawing which shows the example of EpRange in each node. It is explanatory drawing which shows the example of EpTree of tree structure data Tr1. It is explanatory drawing which shows the example of EpTree of tree structure data Tr2. It is explanatory drawing which shows the example of EpTree of tree structure data Tr3. It is explanatory drawing which shows the example of appearance of the label A in each tree structure data. It is explanatory drawing which shows the example of a search state node. It is explanatory drawing which shows the example of the search state at the time of selecting the label A for the root node of a pattern. It is explanatory drawing which shows the example of appearance of the label B in each tree structure data. It is explanatory drawing which shows the example of search state node A [B] at the time of selecting the label B as a descendant of the label A. It is explanatory drawing which shows the example of the search state at the time of selecting the label B as a descendant of the label A. It is explanatory drawing which shows the example of search state node A [B [C]] at the time of selecting the label C as the descendant of the label B. It is explanatory drawing which shows the example of the search state at the time of selecting the label C as a descendant of the label B. It is explanatory drawing which shows the example of search state node A [B]. It is explanatory drawing which shows the example of the search state node A. FIG. It is explanatory drawing which shows the example of search state node A [B [CF]]. It is explanatory drawing which shows the example of search state A [B [CF]]. It is a block diagram which shows the hardware structural example of the computer which implement | achieves this Embodiment.

Explanation of symbols

110 ... Structure information DB
DESCRIPTION OF SYMBOLS 120 ... Information collection device 130 ... Information extraction device 140 ... Extraction information management device 210 ... Structural information management module 220 ... Appearance information selection module 230 ... Appearance information management module 240 ... Investigation range processing module 250 ... Search processing module 260 ... Search state management Module 270 ... Extraction information processing module

Claims (7)

First search means for performing a search for a structure pattern that appears multiple times in a plurality of tree structures with respect to a node lower than the node currently being processed in the tree structure;
A second search in which the search for the structure pattern is performed for each unsearched node that is higher than the current processing target node in the tree structure and lower than the higher node. Means
The information processing apparatus according to claim 1, wherein the first search unit and the second search unit return to the original node that started the search when there is no node to be searched. For the nodes in the tree structure that match the current processing target node that is the current processing target in the structure pattern, the descendants are searched based on the highest-order node for those having a vertical relationship A first search range determining means for determining a range;
The information processing apparatus according to claim 1, wherein the first search unit performs a search based on a range determined by the first search range determination unit. Among the nodes in the tree structure that coincide with the parent node in the structure pattern, the nodes in the tree structure that coincide with the top node and the child nodes in the structure pattern with respect to those having a vertical relationship And a second search range determining means for determining a search range based on the lowest-order node for those having a vertical relationship to
The information processing apparatus according to claim 1, wherein the second search unit performs a search based on a range determined by the second search range determination unit. An appearance location corresponding to a node whose descendants include the node of the current processing target node among the appearance locations of the ancestor node of the current processing target node in the structure pattern The information processing apparatus according to claim 3, further comprising holding means that holds the information. 5. The information processing apparatus according to claim 4, further comprising: a deletion unit that deletes, from the holding unit, occurrence locations other than the highest and lowest levels in a range where there is no branch in the tree structure. And further comprising a selecting means for selecting an appearance location to be held by the holding means, and selecting items other than the appearance location other than the highest and lowest positions in the tree structure without branching. The information processing apparatus according to claim 4. Computer
First search means for performing a search for a structure pattern that appears multiple times in a plurality of tree structures with respect to a node lower than the node currently being processed in the tree structure;
A second search in which the search for the structure pattern is performed for each unsearched node that is higher than the current processing target node in the tree structure and lower than the higher node. Function as a means,
The information processing program characterized in that the first search means and the second search means return to the original node that started the search when there is no node to be searched.

Images