Quo Vadis, Math Information Retrieval

Warning

This publication doesn't include Faculty of Arts. It includes Faculty of Informatics. Official publication website can be found on muni.cz.

Authors	SOJKA Petr NOVOTNÝ Vít AYETIRAN Eniafe Festus LUPTÁK Dávid ŠTEFÁNIK Michal
Year of publication	2019
Type	Article in Proceedings
Conference	Proceedings of the Thirteenth Workshop on Recent Advances in Slavonic Natural Language Processing, RASLAN 2019
MU Faculty or unit	Faculty of Informatics
Citation
Web	Domovská stránka workshopu full paper
Keywords	math information retrieval; question answering; STEM; digital mathematical libraries; embeddings; MIaS; MIaSNG; DML
Description	With the exponential growth of information in the digital form, information retrieval and querying digital libraries is of paramount importance, and mathematical and technical STEM documents are not an exception. The key for precise searching is the adequate and unambiguous representation of documents, paragraphs, sentences and words, which we are going to evaluate. We are presenting a roadmap to tackle the problem of searching and question answering in the digital mathematical libraries, and discuss the pros and cons of promising approaches primarily for the key part, namely the document representation: several types of embeddings, topic mixtures and LSTM. The listed representation learning options will be evaluated at the next ARQMath evaluation lab of CLEF 2020 conference.
Related projects:	Critical Infrastructure Software Architectures, Computer Systems Security, Data Processing and Visualization Techniques, and Augmented Reality