Mathematical Equation Structural Syntactical Similarity Patterns: A Tree Overlapping Algorithm and Its Evaluation

Abstract

In this paper we examine mathematical equations structural syntactical similarity patterns. The major focus of this contribution is an NLP tree overlapping algorithm modication adopted to the case of syntactical similarity of mathematical equations presented in MathML. We describe the software implementation and the tests arranged for the cases of both structural and subexpression based similarity. The paper also contains a discussion of algorithm evaluation problems conditioned by the lack of relevant syntac tical similarity centered equation corpora.