This paper discusses the Answer Validation Task in Question Answering applied for Russian language. Due to poor language resources we are limited in selection of techniques for Question Answering. Dependency parse-based methods applied to factoid questions are in the primary focus. We notice that existing works use either pure syntactic dependency parsers or parsers which perform some extra shallow semantic analysis for English. The selection of either of the parsers is not justified in any of these works. We report experiments for Russian language in absence of WordNet on various combinations of rule-based parsers and graph matching algorithms, including our Parallel Graphs Traversal algorithm first published in ROMIP 2010 [23]. Performance is evaluated using a subset of ROMIP questions collection with ten-fold cross-validation. © 2013 Springer-Verlag.
CITATION STYLE
Solovyev, A. (2013). Dependency-based algorithms for answer validation task in Russian question answering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8105 LNAI, pp. 199–212). https://doi.org/10.1007/978-3-642-40722-2_20
Mendeley helps you to discover research relevant for your work.