Detecting multiword phrases in mathematical text corpora

We present an approach for detecting multiword phrases in mathematical text corpora. The method used is based on characteristic features of mathematical terminology. It makes use of a software tool named Lingo which allows to identify words by means of previously defined dictionaries for specific word classes as adjectives, personal names or nouns. The detection of multiword groups is done algorithmically. Possible advantages of the method for indexing and information retrieval and conclusions for applying dictionary-based methods of automatic indexing instead of stemming procedures are discussed. Problems and goals We start by discussing an example. Given is the text of an abstract for a paper with mathematical content: "We study some rigidity properties for locally symmetrical Finsler manifolds. We obtain the local equivalent characterization for a Finsler manifold to be locally symmetric and prove that any locally symmetrical Finsler manifold with nonzero flag curvature must be Riemannian. We also generalize a rigidity result due to Akbar Zadeh." Looking for methods that will generate index terms automatically and that will have good representation and equally discrimination properties for retrieval purposes, the following question may be of interest: Which of the words are part of a multiword phrase representing a mathematical concept or a proper entity of mathematical terminology? Intellectual analysis can identify the following phrases: • rigidity properties • locally symmetrical Finsler manifold(s) • local equivalent characterization • nonzero flag curvature • rigidity result We have cited the respective longest sequences with a proper meaning. These sequences can contain shorter ones with normally a generic superordinated meaning. Next, we ask the following questions. Is it possible to identify sequences by applying automatic techniques? Is it possible to identify as much as possible sequences of words that can be seen as representations of mathematical concepts? Is it possible to avoid identification of almost all sequences 1 Abstract taken from the database Zentralblatt MATH (http://www.zentralblatt-math.org/zmath/) with permission of the editorial staff.