KR102104496B1 - Method and apparatus of searching data - Google Patents

Abstract

An embodiment of the present invention relates to a data retrieval method and apparatus capable of reducing the time required for retrieving Resource Description Framework (RDF) data, a graph including a node and an edge ) Receiving at least one or more triple data corresponding to), generating an index for a subgraph of the graph using the triple data, receiving query triple data corresponding to a query graph to be searched for in the graph Providing a data retrieval method comprising the steps of: filtering triple data using query triple data and the generated index, retrieving query triple data using filtered triple data, and outputting the searched results. You can.

Description

Data retrieval method and device {METHOD AND APPARATUS OF SEARCHING DATA}

An embodiment of the present invention relates to a data retrieval method and apparatus, and more particularly, to a data retrieval method and apparatus capable of reducing time required for retrieving RDF (Resource Description Framework) data.

The Resource Description Framework (RDF) was introduced as a standard for the Semantic Web. RDF format data is widely used in fields where flexibility and schema-independent data description are required. Accordingly, RDF format data is widely used to describe large amounts of data in fields such as bioinformatics, metadata, Wikipedia, and social networks, which require graph-type data technology.

In addition, the SPARQL standard was introduced as a query language for RDF data accumulated in large amounts. Accordingly, interest in a method of processing a SPARQL query for RDF data is increasing.

Most RDF repositories give data and store triples including three information (Subject, Predicate, Object) corresponding to the verb and object as a unit. Accordingly, when a SPARQL query is input to the RDF storage, a plurality of join operations may be performed using data stored in triple units to process the query. Join operations can take longer than other types of operations. Therefore, the speed of the join operation can influence the speed of query processing in the RDF repository.

In order to reduce the execution time of the join operation performed when processing a query in graph data, the number of join operations is reduced, the efficiency of the join operation is improved, or the amount of data that is the object of the join operation is reduced. Reduction methods can be used.

The method of reducing the number of join operations is mainly used in Jena or Oracle. In the above method, a table generated as a result of joining the frequently used patterns can be stored separately in advance. However, it is very difficult to determine the range in which to perform the join in advance, and a null value and a multi value are generated, which may degrade the performance of the operation.

The method of improving the efficiency of join operation is a method mainly used in SW-Store, Hexastore, and RDF-3X. In the above method, for each triple, several indexes such as SPO (Subject-Predicate-Object), PSO (Predicate-Subject-Object), and OPS (Object-Predicate-Subject) can be previously generated. This can speed up access to triples. In addition, when a merge join is performed on them, the join operation can be performed more efficiently.

In a method of reducing the amount of data that is a target of a join operation, data not subject to a join operation may be excluded from input data by using a filter such as Ubiquitous Sideways Information Passing (U-SIP). This method is also used in RDF-3X.

In addition, SPARQL query processing systems such as GRIN index, DOGMA, PIG, gStore, etc., offer indexing techniques. However, the index in such a system serves to reduce the search range, and may be limited in application to reduce the amount of data that is the object of the join operation.

An embodiment of the present invention can provide a data retrieval method and apparatus that can reduce the time required to retrieve RDF (Resource Description Framework) data.

An embodiment of the present invention can provide a data retrieval method and apparatus capable of reducing execution time of a join operation performed when processing a query in graph form data.

An embodiment of the present invention can provide a data retrieval method and apparatus capable of reducing the amount of data that is a target of a join operation performed when processing a query in graph form data.

An embodiment of the present invention can provide a data retrieval method and apparatus capable of improving the speed at which join operations using graph data are performed.

An embodiment of the present invention is a data retrieval method and apparatus that can reduce the amount of data in the form of a graph that is a target of a join operation by using information on the structure of a graph including nodes and edges Can provide.

In the data retrieval method according to an embodiment of the present invention, receiving at least one or more triple data corresponding to a graph including a node and an edge, and using the triple data Generating an index for a subgraph of the graph, receiving query triple data corresponding to a query graph to be searched in the graph, and filtering the triple data using the query triple data and the generated index The method may include searching the query triple data using the filtered triple data, and outputting the searched results.

In addition, generating an index for a subgraph of the graph using the triple data may include regenerating the graph using the triple data, selecting an arbitrary subgraph included in the graph, and It may include the step of generating the index for the selected subgraph, and repeating the step of selecting any subgraph included in the graph to the step of generating an index for the selected subgraph.

In addition, the step of generating an index for the selected subgraph may include calculating a number of times the selected subgraph is matched in the graph, and when the calculated number of times is greater than or equal to a reference number, for the selected subgraph. And generating the index.

Further, the repeating may include repeating the steps of selecting the subgraph or generating the index for subgraphs included in the graph and different from the selected subgraph.

In addition, the step of generating an index for a subgraph of the graph using the triple data generates a list including nodes corresponding to each node included in the subgraph among the nodes included in the graph. It may include steps.

In addition, the step of generating an index for a subgraph of the graph using the triple data is a list including a node corresponding to a node adjacent to each edge included in the subgraph among the nodes included in the graph It may include the step of generating.

In addition, the subgraph may include at least two or more edges among the edges included in the graph.

In addition, filtering the triple data using the query triple data and the generated index may include regenerating the query graph using the query triple data, and corresponding to each edge included in the query graph. The method may include determining an order of performing join operations on the triple data, and filtering the triple data targeted for each join operation using the generated index.

In addition, filtering the triple data that is the target of each join operation using the generated index may include each of the nodes included in the graph in the query graph using the generated index. Generating a list of at least one corresponding node including nodes corresponding to nodes adjacent to an edge of, generating a common node list including nodes commonly included in the respective corresponding node lists, and the query graph Filtering the remaining triple data except for the triple data corresponding to each node included in the common node list among the triple data corresponding to each edge included in the.

In addition, the step of retrieving the query triple data using the filtered triple data may include performing the join operation using the filtered triple data, and querying the query triple data from the result of the join operation. And searching.

The data retrieval apparatus according to an embodiment of the present invention receives at least one or more triple data corresponding to a graph including a node and an edge, and queries to search in the graph An input unit that receives query triple data corresponding to a graph, an index generator that generates an index for a subgraph of the graph using the triple data, and filters the triple data using the query triple data and the generated index It may include a filtering unit, a search unit for searching the query triple data using the filtered triple data, and an output unit for outputting the searched result.

In addition, the index generator regenerates the graph using the triple data, the index generator selects any subgraph included in the graph, and the index generator generates an index for the selected subgraph, The index generator may repeat the steps of selecting an arbitrary subgraph included in the graph or generating the index of the selected subgraph.

In addition, the index generator may calculate the number of times the selected subgraph matches in the graph, and the index generator may generate the index for the selected subgraph when the calculated number of times is greater than or equal to a reference number. .

Further, the index generator may repeat the steps of selecting the subgraph or generating the index for subgraphs included in the graph and different from the selected subgraph.

In addition, the index may include a list including nodes corresponding to each node included in the subgraph among the nodes included in the graph.

In addition, the index may include a list including nodes corresponding to nodes adjacent to each edge included in the subgraph among the nodes included in the graph.

In addition, the subgraph may include at least two or more edges among the edges included in the graph.

In addition, the filtering unit regenerates the query graph using the query triple data, and the filtering unit determines an order of performing a join operation on the triple data corresponding to each edge included in the query graph, and the The filtering unit may filter the triple data that is the target of each join operation using the generated index.

In addition, the filtering unit generates at least one corresponding node list including nodes corresponding to nodes adjacent to each edge included in the query graph among the nodes included in the graph, by using the generated index, The filtering unit generates a common node list including nodes commonly included in the respective corresponding node lists, and the filtering unit is selected from the triple data corresponding to each edge included in the query graph, in the common node list. The remaining triple data may be filtered except for the triple data corresponding to each node included.

Also, the search unit may perform the join operation using the filtered triple data, and the search unit may search the query triple data from the result of the join operation.

According to an embodiment of the present invention, it is possible to reduce the time required to search for RDF (Resource Description Framework) data.

According to an embodiment of the present invention, it is possible to reduce the execution time of a join operation performed when processing a query in graph form data.

According to an embodiment of the present invention, the amount of data that is a target of a join operation performed when processing a query in graph form data can be reduced.

According to an embodiment of the present invention, it is possible to improve the speed at which join operations using graph data are performed.

According to an embodiment of the present invention, by using information about a structure of a graph including a node and an edge, the amount of data in the form of a graph that is a target of a join operation can be reduced.

1 is a flowchart illustrating a process of performing a data search method according to an embodiment of the present invention.
2 is a view showing an overall graph according to an embodiment of the present invention.
3 is a flowchart illustrating a process in which the step of generating an index for a subgraph is performed according to an embodiment of the present invention.
4 is a diagram showing an index generated by using the graph shown in FIG. 2 according to an embodiment of the present invention.
5 is a diagram illustrating a query graph according to an embodiment of the present invention.
6 is a flowchart illustrating a process in which filtering of triple data is performed according to an embodiment of the present invention.
7 is a reference diagram for explaining a step of determining an order of performing a join operation on query triple data according to an embodiment of the present invention.
8 is a block diagram showing the configuration of a data retrieval apparatus according to an embodiment of the present invention.

Advantages and features of the present invention, and methods for achieving them will be clarified with reference to embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various different forms, and only the present embodiments allow the disclosure of the present invention to be complete, and the ordinary knowledge in the technical field to which the present invention pertains. It is provided to fully inform the holder of the scope of the invention, and the invention is only defined by the scope of the claims. The same reference numerals refer to the same components throughout the specification.

Although "first" or "second" are used to describe various components, these components are not limited by the terms above. The above terms may be used only to distinguish one component from another component. Accordingly, the first component mentioned below may be the second component within the technical spirit of the present invention.

The terminology used herein is for describing the embodiments and is not intended to limit the present invention. In this specification, the singular form also includes the plural form unless otherwise specified in the phrase. As used herein, “comprises” or “comprising” implies that the stated component or step does not exclude the presence or addition of one or more other components or steps.

Unless otherwise defined, all terms used in the present specification may be interpreted as meanings commonly understood by those skilled in the art to which the present invention pertains. In addition, terms defined in the commonly used dictionary are not ideally or excessively interpreted unless explicitly defined.

Hereinafter, a data search method and a data search apparatus 100 according to an embodiment of the present invention will be described in detail with reference to FIGS. 1 to 8.

1 is a flowchart illustrating a process of performing a data search method according to an embodiment of the present invention. Referring to FIG. 1, in a data search method according to an embodiment of the present invention, first, receiving at least one or more triple data corresponding to a graph including a node and an edge (S100) may be performed.

The graph may include a node and an edge connecting the nodes to each other. Edges included in the graph may have a direction. For example, the first node and the second node are connected to each other by an edge, and the edge may have a direction from the first node toward the second node. At this time, the first node may be referred to as a departure node, and the second node may be referred to as an arrival node. When the graph includes a plurality of edges, the graph may include at least three or more nodes.

Each data stored in the Resource Description Framework (RDF) storage may have a triple format including three information (Subject, Predicate, Object) corresponding to a subject, a verb, and an object. Accordingly, each data stored in the RDF storage may be referred to as triple data.

Each edge included in the graph may be represented by triple data. For example, the departure node may correspond to information corresponding to a subject among three pieces of information included in the treble data. Also, an edge may correspond to information corresponding to a verb among three pieces of information included in triple data. In addition, the arrival node may correspond to information corresponding to the target word among three pieces of information included in the triple data. Since one edge may be represented by one triple data, in the step of receiving triple data (S100), triple data having the same number of edges included in the graph may be received.

2 is a view showing an overall graph according to an embodiment of the present invention. Referring to FIG. 2, the entire graph may include four separate graphs of R1, R2, R3, and R4. In step S100 of receiving triple data, triple data corresponding to the entire graph may be received. In other words, triple data corresponding to each edge included in the entire graph may be received. Also, triple data having the same number of edges included in the entire graph may be received.

15 triple data corresponding to each edge included in the entire graph shown in FIG. 2 may be received. For example, four triples T1, T2, T3, and T4 corresponding to each edge included in the graph R1 may be received. For example, T1 = {v1, p3, v2}, T2 = {v5, p4, v2}, T3 = {v2, p2, v3}, T4 = {v3, p1, v4}.

Referring back to FIG. 1, next, an operation (S110) of generating an index for a subgraph of the graph using the triple data may be performed. 3 is a flowchart illustrating a process in which the step of generating an index for a subgraph (S110) is performed according to an embodiment of the present invention.

Referring to FIG. 3, in generating an index for a subgraph (S110), first, regenerating the graph using the triple data (S111) may be performed. For example, the entire graph as shown in FIG. 2 may be reconstructed using the triple data.

Next, a step (S112) of selecting any subgraph included in the graph may be performed. For example, a subgraph including only the edge p1 of the graph R1 may be selected from the entire graph illustrated in FIG. 2. The size of the subgraph may be the same as the number of edges included in the subgraph. Therefore, the size of the subgraph including only the edge p1 may be 1.

Referring back to FIG. 3, next, a step (S113) of calculating the number of times the selected subgraph is matched in the graph may be performed. For example, the number of times the edge p1 is matched in the entire graph illustrated in FIG. 2 may be calculated. Referring to FIG. 2, since the graphs R1, R2, and R3 each include one p1, the number of times the edge p1 is matched in the entire graph may be three.

Referring back to FIG. 3, next, when the calculated number of times is greater than or equal to the reference number, an operation (S114) of generating an index for the selected subgraph may be performed. By generating an index only when the calculated number of times is greater than or equal to the reference number, an index can be generated only for subgraphs having a relatively large number of times included in the entire graph. As described above, efficiency can be improved by allocating a limited index storage space to subgraphs having a relatively large number of times included in the entire graph.

For example, when the reference count is 3, an index for a subgraph including only edge p1 may be generated. 4 is a diagram showing an index generated by using the graph shown in FIG. 2 according to an embodiment of the present invention. For example, as an index for a subgraph including only the edge p1, a corresponding node list that is a list of nodes that can correspond to each node included in the subgraph among the nodes included in the graph may be generated. In other words, a list of nodes corresponding to nodes adjacent to each edge included in the subgraph among nodes included in the graph may be generated.

For example, since the starting nodes of the edges p1 included in the graphs R1, R2, and R3 are v3, v8, and v14, respectively, as shown in FIG. 4, the corresponding node list Vlist (which can correspond to the first node of the subgraph gp1) Vlist ( gp1,? v1) = {v3, v8, v14} can be generated. In addition, since the arrival nodes of the edge p1 included in the graphs R1, R2, and R3 are v4, v9, and v15, respectively, as shown in FIG. 4, the corresponding node list Vlist (gp1, which can correspond to the second node of the subgraph gp1) ? v2) = {v4, v9, v15} can be generated.

Referring back to FIG. 3, next, a step (S115) of repeating steps (S112) of selecting an arbitrary subgraph included in the graph to a step of generating an index for the selected subgraph (S114) is performed (S115). Can be. In other words, the process may be repeated for subgraphs included in the graph and different from the selected subgraph. Accordingly, for each of the subgraphs included in the entire graph, steps of selecting a subgraph (S112) to generating an index for the selected subgraph (S114) may be repeated.

The final completed index may be as shown in FIG. 4. Referring to FIG. 4, the finally completed index may be sorted according to the size of the subgraph. The maximum size of the subgraph to create an index may be determined in advance. For example, as illustrated in FIG. 4, the maximum size of a subgraph to generate an index may be predetermined to 3.

As described above, since the maximum size of a subgraph to generate an index is predetermined and an index is generated only for subgraphs having a relatively large number of times included in the entire graph, time and space required to create an index are saved. Can be.

Referring back to FIG. 3, next, a step (S120) of receiving query triple data corresponding to a query graph to be searched in the graph may be performed. 5 is a diagram illustrating a query graph according to an embodiment of the present invention. Accordingly, query triple data corresponding to each edge included in the query graph may be received. Also, the same number of query triple data may be received as the number of edges included in the query graph.

Referring back to FIG. 3, next, filtering the triple data using the query triple data and the generated index (S130) may be performed. 6 is a flowchart illustrating a process in which the step of filtering triple data (S130) is performed according to an embodiment of the present invention.

Referring to FIG. 6, in the step of filtering triple data (S130), a step of regenerating the query graph using the query triple data (S131) may be performed. For example, a query graph as shown in FIG. 5 may be regenerated using the query triple data.

Referring back to FIG. 6, next, a step (S132) of determining an order of performing a join operation on the triple data corresponding to each edge included in the query graph may be performed. 7 is a reference diagram for explaining a step (S132) of determining an order of performing a join operation on triple data according to an embodiment of the present invention.

Referring to FIG. 7, for example, first, a join operation may be performed on triple data corresponding to edge p1 and triple data corresponding to edge p2. Next, a join operation may be performed on the result of the join operation and triple data corresponding to edge p3. Next, a join operation may be performed on the result of the join operation and triple data corresponding to edge p4.

Referring back to FIG. 6, next, a step (S133) of filtering the triple data that is the target of each join operation may be performed using the generated index.

For example, triple data corresponding to edge p1 may be filtered first. Referring to FIG. 5, a third node (? V3) is located at a departure node of edge p1 and a fourth node (? V4) is located at an arrival node of edge p1. In addition, since a subgraph including edges p3, p4, and p2 is located at a point corresponding to the starting node of edge p1, the third node (? V3) is located at one end of the subgraph including edges p3, p4, and p2. do.

Referring to the index illustrated in FIG. 2, in order to calculate a corresponding node list that can correspond to the starting node of edge p1, a corresponding node list Vlist (gp1,? V1) for the first node (? V1) of subgraph gp1 ) And the corresponding node list Vlist (gp5,? V4) for the fourth node (? V4) of the subgraph gp5 may be referenced. Since the subgraph including the edges p3, p4, and p2 is located at the point corresponding to the starting node of the edge p1, the corresponding node list Vlist (gp5,? V4) for the fourth node (? V4) of the subgraph gp5 as described above Can be referenced.

The corresponding node list Vlist (gp1,? V1) includes v3, v8, v14, and the corresponding node list Vlist (gp5,? V4) contains v3, v18. In the query graph, the starting node of the edge p1 is the first node (? V1) of the subgraph gp1 and at the same time the fourth node (? V4) of the subgraph gp5, so the corresponding node list Vlist (gp1,? V1) and the corresponding node list Vlist Only nodes included in (gp5,? v4) at the same time may correspond to the starting node of edge p1. Therefore, only v3 can correspond to the starting node of edge p1.

Also, the corresponding node list Vlist (gp2,? V2) for the second node (? V2) of the subgraph gp1 may be referenced to calculate a list of corresponding nodes that can correspond to the arrival node of the edge p1. The corresponding node list Vlist (gp2,? V2) includes v4, v9, and v15. Among them, since only the arrival node corresponding to v3 corresponding to the departure node of the edge p1 is v4 of the graph R1, among all the triple data for the edge p1, the triple data except triple data {v3, p1, v4} can be filtered. have.

When filtering the triple data for the edge p2 similarly to the above, among the triple data for the edge p2, the triple data other than the triple data {v2, p2, v3}, and {v17, p2, v18} may be filtered. Also, among all triple data for the edge p3, the triple data other than the triple data {v1, p3, v2}, and {v16, p3, v17} may be filtered. In addition, among all triple data for the edge p4, the triple data other than the triple data {v5, p4, v2}, and {v19, p4, v17} may be filtered.

Referring back to FIG. 3, next, an operation (S140) of performing the join operation using the filtered triple data may be performed. Referring to the above embodiment, the number of triple data targeted for the join operation may be 1 + 2 + 2 + 2 = 7.

If the join operation is performed without filtering the triple data as described above, the join operation may be performed using all the triple data for each edge. At this time, the number of triple data that is the target of the join operation may be 3 + 5 + 4 + 3 = 15.

In addition, it is assumed that triple data is filtered using a list of corresponding nodes that can correspond to a departure node and an arrival node, rather than filtering triple data using the index generated as described above. At this time, the number of triple data that is the target of the join operation may be 3 + 4 + 3 + 1 = 11.

Therefore, as in the embodiment of the present invention, when generating indexes for each of the subgraphs including different numbers of edges, and filtering triple data using the generated index, the object of the join operation is compared to other methods This may be the smallest number of triple data.

Referring back to FIG. 3, next, a step (S150) of retrieving the query triple data from the result of the join operation may be performed. Referring to the above embodiment, triple data corresponding to the graph R1 of FIG. 2 may be searched among the results of the join operation.

Referring back to FIG. 6, next, a step (S160) of outputting the searched result may be performed.

8 is a block diagram showing the configuration of a data retrieval apparatus 100 according to an embodiment of the present invention. Referring to FIG. 8, the data retrieval apparatus 100 according to an embodiment of the present invention includes an input unit 110, an index generation unit 120, a filtering unit 130, a search unit 140, and an output unit 150 ).

The input unit 110 may receive at least one or more triple data corresponding to a graph including a node and an edge. Also, the input unit 110 may receive query triple data corresponding to a query graph to be searched in the graph. The input unit 110 may be, for example, a communication device such as a network adapter or other input device. Details of the input unit 110 correspond to the above-described step of receiving the triple data (S100) and the step of receiving the query triple data (S120), so a detailed description thereof will be omitted.

The index generator 120 may generate an index for a subgraph of the graph using the triple data. The index generator 120 may be, for example, a computing device such as a central processing unit, a database, a server, or a terminal device. Details of the index generator 120 correspond to step S110 of generating the above-described index, so a detailed description thereof will be omitted.

The filtering unit 130 may filter the triple data using the query triple data and the generated index. The filtering unit 130 may be, for example, a computing device such as a central processing unit, a database, a server, or a terminal device. The details of the filtering unit 130 correspond to the step (S130) of filtering the triple data described above, so a detailed description thereof will be omitted.

The search unit 140 may search for the query triple data using the filtered triple data. The search unit 140 may be, for example, a computing device such as a central processing unit, a database, a server, or a terminal device. The details of the search unit 140 correspond to the step S140 of performing the above-described join operation and the step S150 of searching the query triple data, so a detailed description thereof will be omitted.

The output unit 150 may output the searched result. The output unit 150 may be, for example, a communication device such as a network adapter or other output device. The details of the output unit 150 correspond to the step (S160) of outputting the above-mentioned searched results, so a detailed description thereof will be omitted.

According to the embodiment of the present invention described above, it is possible to reduce the time required for retrieving RDF (Resource Description Framework) data. In addition, it is possible to reduce the execution time of a join operation performed when processing a query in graph form data. In addition, it is possible to reduce the amount of data that is the object of a join operation performed when processing a query in graph-type data. In addition, it is possible to improve the speed at which the join operation using graph data is performed. In addition, by using information on the structure of the graph including nodes and edges, the amount of data in the form of a graph that is a target of a join operation can be reduced.

Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art to which the present invention pertains can implement the present invention in other specific forms without changing its technical spirit or essential features. You will understand that there is. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

100: data retrieval device
110: input unit
120: index generation unit
130: filtering unit
140: search unit
150: output unit

Claims (20)

In the data search method of the data search device,
Receiving at least one or more triple data corresponding to a graph including a node and an edge;
Generating an index for a subgraph of the graph using the triple data;
Receiving query triple data corresponding to a query graph to be searched in the graph;
Filtering the triple data using the query triple data and the generated index;
Retrieving the query triple data using the filtered triple data; And
Outputting the searched result
Data search method comprising a. According to claim 1,
Generating an index for the subgraph of the graph using the triple data,
Regenerating the graph using the triple data;
Selecting any subgraph included in the graph;
Generating the index for the selected subgraph; And
Repeating the steps of selecting any subgraph included in the graph or generating an index for the selected subgraph.
Data search method comprising a. According to claim 2,
The step of generating an index for the selected subgraph,
Calculating a number of times the selected subgraph is matched in the graph; And
Generating the index for the selected subgraph when the calculated number of times is greater than or equal to a reference number of times
Data search method comprising a. According to claim 2,
The repeating step,
Repeating the step of selecting the subgraph or generating the index for subgraphs included in the graph and different from the selected subgraph.
Data search method comprising a. According to claim 1,
Generating an index for the subgraph of the graph using the triple data,
Generating a list including a node corresponding to each node included in the subgraph among the nodes included in the graph
Data search method comprising a. According to claim 1,
Generating an index for the subgraph of the graph using the triple data,
Generating a list including nodes corresponding to nodes adjacent to each edge included in the subgraph among the nodes included in the graph.
Data search method comprising a. According to claim 1,
The subgraph includes at least two or more edges among the edges included in the graph. According to claim 1,
Filtering the triple data using the query triple data and the generated index may include:
Regenerating the query graph using the query triple data;
Determining an order of performing a join operation on the triple data corresponding to each edge included in the query graph; And
Filtering the triple data that is the object of each join operation using the generated index
Data search method comprising a. The method of claim 8,
Filtering the triple data, which is the target of each join operation, using the generated index,
Generating at least one corresponding node list including nodes corresponding to nodes adjacent to each edge included in the query graph among the nodes included in the graph using the generated index;
Generating a common node list including nodes commonly included in the respective corresponding node lists; And
Filtering the remaining triple data excluding the triple data corresponding to each node included in the common node list among the triple data corresponding to each edge included in the query graph.
Data search method comprising a. The method of claim 8,
Searching for the query triple data using the filtered triple data may include:
Performing the join operation using the filtered triple data; And
Retrieving the query triple data from the result of the join operation
Data search method comprising a. An input unit receiving at least one triple data corresponding to a graph including a node and an edge, and receiving query triple data corresponding to a query graph to be searched in the graph;
An index generator configured to generate an index for a subgraph of the graph using the triple data;
A filtering unit filtering the triple data using the query triple data and the generated index;
A search unit that searches the query triple data using the filtered triple data; And
An output unit that outputs the searched results
Data search device comprising a. The method of claim 11,
The index generator regenerates the graph using the triple data, the index generator selects any subgraph included in the graph, and the index generator generates an index for the selected subgraph, and the index The generation unit repeats the steps of selecting an arbitrary subgraph included in the graph or generating the index for the selected subgraph. The method of claim 12,
The index generator calculates the number of times the selected subgraph matches in the graph, and the index generator generates the index for the selected subgraph when the calculated number of times is greater than or equal to a reference number. The method of claim 12,
The index generation unit, the data retrieval apparatus that repeats the steps of selecting the subgraph or generating the index for subgraphs different from the selected subgraph included in the graph. The method of claim 11,
The index is a data search apparatus including a list including nodes corresponding to each node included in the subgraph among the nodes included in the graph. The method of claim 11,
The index, the data search apparatus including a list including a node corresponding to a node adjacent to each edge included in the subgraph among the nodes included in the graph. The method of claim 11,
The subgraph is a data retrieval apparatus including at least two or more edges among the edges included in the graph. The method of claim 11,
The filtering unit regenerates the query graph using the query triple data, the filtering unit determines an order of performing a join operation on the triple data corresponding to each edge included in the query graph, and the filtering unit A data retrieval device that filters the triple data that is the target of each join operation using the generated index. The method of claim 18,
The filtering unit generates a list of at least one corresponding node including a node corresponding to a node adjacent to each edge included in the query graph among the nodes included in the graph, by using the generated index, and filtering The unit generates a common node list including nodes commonly included in each corresponding node list, and the filtering unit is included in the common node list among the triple data corresponding to each edge included in the query graph. A data retrieval apparatus for filtering the remaining triple data except for the triple data corresponding to each node. The method of claim 18,
The search unit performs the join operation using the filtered triple data, and the search unit retrieves the query triple data from the result of the join operation.

Images