Mining lexical hyponymy relations from large-scale concept set

Inner structures of Chinese lexical concepts have embedded some useful semantic relations. In this paper, we proposed a new statistical approach to mine lexical hyponymy relations from large-scale concept set, instead of analyzing inner structures. Firstly we designed common suffix tree to cluster the lexical concept set. Class concepts are then extracted by statistic-base rules we investigated in concept set. Finally, we export hyponymy relations from the common suffix tree. Experimental result showed us that this approach achieved a precision of 95.833% and a recall of 67.241% when the concept size achieved 800,000.