Dr.Martinovic Presents at Information Retrieval Workshops

Dr. Miroslav Martinovic of the Computer Science Department presented the results of his research done with TCNJ undergraduate students at two workshops. At the "8th International Conference on Speech, Text and Dialogue" September 2005, Carlsbad, Czech Republic, the paper entitled "AARLISS - an Algorithm for Anaphora Resolution in Long-distance Inter-Sentential Scenarios" was presented. This work, co-authored by students Anthony Curley and John Gaskins, presents a novel approach for boosting the performance of pro-nominal anaphora resolution algorithms when a search for antecedents has to span over a multi-sentential text passage. The approach is based on the identi-fication of sentences which are "most semantically related" to the sentence with anaphora.

At the "Workshop on the Computational Modeling and Lexical Acquisition", July 2005, Split, Croatia, Dr.Martinovic presented two papers. The first, entitled "SteLemMin - a Generic Minimal Stem Algorithm for Word Conflation and Lemmatization", was co-authored by student Louis Rufano. This paper introduces an algorithm for transforming any sequential word conflation or lemmatization algorithm into an algorithm whose final product is guaranteed to be a minimal stem. The SteLemMin method is based on an equivalence relation which partitions the sets of conflated word forms into mutually exclusive and exhaustive equivalence classes. This method clearly contributes to reducing the number of terms in an Information Retrieval system that would use it, but also does it in a space-wise most efficient manner. Dr.Martinovic's algorithm demonstrates no dependency on a particular natural language and can be tailored to very different morphological systems. This was illustrated by examples from English and German. The second presentation given by Dr.Martinovic was entitled "Open Domain Question Answering Systems - From Design to Development and Implementation - Tutorial".