A formula retrieval method based on hierarchical structure clusters

Focusing on the issue that formula retrieval could not be realized with traditional full-text retrieval technology because of the spatial features of formulae, this paper proposed a formulae retrieval method with the full considerations of the hierarchical structures of formulae. The index was constructed according to the clusters of formulae with their baseline level structures. Based on the index, the math retrieval algorithm was designed with different query modes, such as precision matching and structural matching. The experimental results show the average query response times in accurate retrieval mode and structural mode are acceptable, which indicates the practicability of proposed method.