In this talk we will describe the major results from the Mandarin-English Information (MEI) cross-language speech retrieval project at the Johns Hopkins Summer 2000 Workshop. Our goal was to build systems for searching Mandarin Chinese audio using text queries written in English in order to explore the potential benefits of coupling multiscale audio indexing with multiscale translation techniques. We will begin with an overview of our approach and the evaluation methodology that we adopted. We will then focus on four interesting results that we obtained: post-translation retokenization, named entity translation through cross-language phonetic mapping, comparison of structured queries with balanced translation, and combination of evidence from multiple speech recognition systems. We will conclude with a few words about the broader implications of what we have learned and the open questions that remain.
Dr. Gina-Anne Levow is a Postdoctoral Researcher at the University of Maryland Institute for Advanced Computer Studies. She holds a Ph.D. in Electrical Engineering and Computer Science from the Massachusetts Institute for Technology. Her research interests include information retrieval and discourse processing, with recent work focusing on cross-language information retrieval and translingual topic tracking. Additional information is available at http://www.umiacs.umd.edu/~gina.
Mr. Jianqiang Wang is a doctoral student in the College of Information Studies at the University of Maryland. His research interests are focused on cross-language information retrieval.
For the colloquium series schedule, see the UMD Computational Linguistics Colloquium Series web page at http://umiacs.umd.edu/~resnik/cl_colloquium/. If you are interested in meeting with the speaker, please contact Philip Resnik (resnik@umiacs.umd.edu).