Incorporating compactness to generate term-association view snippets for ontology search

A query-relevant snippet for ontology search is useful for deciding if an ontology fits users’ needs. In this paper, we illustrate a good snippet in a keyword-based ontology search engine should be with term-association view and compact, and propose an approach to generate it. To obtain term-association view snippets, a model of term association graph for ontology is proposed, and a concept of maximal r-radius subgraph is introduced to decompose the term association graph into connected subgraphs, which preserve close relations between terms. To achieve compactness, in a query-relevant maximal r-radius subgraph, a connected subgraph thereof with a small graph weight is extracted as a sub-snippet. Finally, a greedy method is used to select sub-snippets to form a snippet in consideration of query relevance and compactness without violating the length constraint. An empirical study on our implementation shows that our approach is feasible. An evaluation on effectiveness shows that the term-association view snippet is favored by users, and the compactness helps reading and judgment.

► Term association graph is proposed to model associations between terms. ► All maximal r-radius subgraphs are extracted for a TAG. ► A group Steiner algorithm is used to generate a compact sub-snippet. ► Greedy approach is used to assemble some sub-snippets in a structured snippet. ► Experiments show that it is feasible and useful for reading and judgment.