Frequent-Pattern Based Facet Extraction from Graph Data

Graph is a general data model which is applicable to complex data structures. Hence, there are lots of data which can be expressed using graph data models. For example, Web, social networking services, bibliographic database, and Linked Open Data. Since such graph data can be heterogeneous, users are imposed a huge burden on searching over the graph data to find desired sub graphs. Faceted search is a promising approach to reduce the burden of searching graph data. Applying faceted search for graph data requires to determine objects (target sub graphs) and facets. To achieve this, in this paper, we propose a framework for faceted search over graph data. The framework is organized into two phases, namely, extraction phase and search phase. The main objective of this paper is to develop the extraction phase which has two main tasks, one is to extract target sub graphs, and the other is to extract facets. This paper applies frequent sub graph mining techniques to extract target sub graphs and facets. The proposed framework is experimentally evaluated using publicly available graph datasets, namely, citation network data and review network data, which show the proposed framework works as expected.