TREC-10 Experiments at University of Maryland CLIR and Video

The University of Maryland Researchers participated in both the Arabic-English Cross Language Information Retrieval (CLIR) and Video tracks of TREC-10. In the CLIR track, our goal was to explore effective monolingual Arabic IR techniques and effective query translation from English to Arabic for cross language IR. For the monolingual part, the use of the different index terms including words, stems, roots, and character n-grams were explored. For the English-Arabic CLIR, the use of MT, wordlist based translation, and non-dictionary words transliteration was explored. In the video track, we participated in the shot boundary detection, and known item search with the primary goals being to evaluate existing technology for shot detection and a new approach to extending simple visual image queries to video sequences. We present a general overview of the approaches, summarize the results in discuss how the algorithms are being extended.