For SSGCI competition we are using standard XML based formats that are often used by the research community and that are easy to read/write by majority of the commonly used programming languages.
Input for participant’s methods
The participants are provided the graphs in GraphML format (details here) and the ground-truth for the sample dataset is provided as an XML file for each query graph.
Output from participant’s methods
The competition is run in an “open mode” which means that we will not ask the participants for the code, executables nor the binaries or algorithms, but only the results. However, the participants are required to proved a detailed description of their method to the organizers, so that its authenticity could be ensured.
The participant’s method is required to provide the results for each query graph in the form of an XML file containing the following:
1. a list of result graphs for the query graph
2. the list of nodes in the result graph where the query graph is spotted
3. the corresponding node-to-node correspondences between the query and the result graph
Sample result files are provided with the “sample dataset”, so that the participants can adapt the output of their programs. The following two code snippets show the format of the result file.
Case: when a method can provide the node-to-node correspondences
<?xml version="1.0" encoding="UTF-8" ?>
<GroundTruthFile>
<NumOfResultGraphs>1</NumOfResultGraphs>
<GraphCollection>
<Graph>
<DBGraphFilename>DatabaseGraph_999.graphml</DBGraphFilename>
<QueryGraphFilename>QueryGraph_099.graphml</QueryGraphFilename>
<NodeCorrespondences>
<correspondence DBGraphNode="50" queryGraphNode="0"/>
<correspondence DBGraphNode="51" queryGraphNode="1"/>
<correspondence DBGraphNode="52" queryGraphNode="2"/>
<correspondence DBGraphNode="53" queryGraphNode="3"/>
<correspondence DBGraphNode="54" queryGraphNode="4"/>
<correspondence DBGraphNode="55" queryGraphNode="5"/>
</NodeCorrespondences>
</Graph>
</GraphCollection>
</GroundTruthFile>
Case: when a method can provide only the list of nodes in result graph
<?xml version="1.0" encoding="UTF-8" ?> <GroundTruthFile> <NumOfResultGraphs>3</NumOfResultGraphs> <GraphCollection> <Graph> <DBGraphFilename>DatabaseGraph_999.graphml</DBGraphFilename> <QueryGraphFilename>QueryGraph_099.graphml</QueryGraphFilename> <ListOfDBGraphNodes>24,25,26,27,28,29,30</ListOfDBGraphNodes> </Graph> <Graph> <DBGraphFilename>DatabaseGraph_999.graphml</DBGraphFilename> <QueryGraphFilename>QueryGraph_099.graphml</QueryGraphFilename> <ListOfDBGraphNodes>24,25,26,27,28,29,30</ListOfDBGraphNodes> </Graph> <Graph> <DBGraphFilename>DatabaseGraph_999.graphml</DBGraphFilename> <QueryGraphFilename>QueryGraph_099.graphml</QueryGraphFilename> <ListOfDBGraphNodes>24,25,26,27,28,29,30</ListOfDBGraphNodes> </Graph> </GraphCollection> </GroundTruthFile>
For generating the spotting results of your method, please follow the format of the files in the « Groundtruth » folder that has been provided with the sample dataset. For each query graph your method should generate one result file.
The result file for « xxx.graphml » should be named as « Result_xxx.xml ». For example, the result file for « QueryGraph_000.graphml » should be named as « Result_QueryGraph_000.xml »