Three Heads are Better than One

Three Heads are Better than One Robert Frederking Center for Machine Translation Carnegie Mellon University Pittsburgh, PA 15213 ref@cs.cmu.edu Sergei Nirenburg Computing Research Laboratory New Mexico State University Las Cruces, NM 88003 sergei@crl.nmsu.edu Abstract Machine translation (MT) systems do not currently achieve optimal quality trans- lation on free text, whatever translation method they employ. Our hypothesis is that the quality of MT will improve if an MT environment uses output from a vari- ety of MT systems working on the same text. In the latest version of the Pan- gloss MT project, we collect the results of three translation engines -- typically, sub- sentential chunks -- in a chart data struc- ture. Since the individual MT systems op- erate completely independently, their re- sults may be incomplete, conflicting, or re- dundant. We use simple scoring heuristics to estimate the quality of each chunk, and find the highest-score sequence of chunks (the "best cover"). This paper describes in detail the combining method, present- ing the algorithm and illustrations of its progress on one of many actual translations it has produced. It uses dynamic program- ming to efficiently compare weighted aver- ages of sets of adjacent scored component translations. The current system operates primarily in a human-aided MT mode. The translation delivery system and its associ- ated post-editing aide are briefly described, as is an initial evaluation of the usefulness of this method. Individual M T engines will be reported separately and are not, there- fore, described in detail here. 95 1 INTRODUCTION Current MT systems, whatever translation method they employ, do not reach an optimal output on free text. In part, this is due to the inherent problems of a particular method -- for instance, the inabil- ity of statistics-based MT to take into account long- distance dependencies, the difficulty in achieving ex- tremely broad coverage in knowledge-based MT sys- tems, or the reliance of most transfer-oriented MT systems on similarities in syntactic structures of the source and the target languages. Our hypothesis is that if an MT environment can use the best results from a variety of MT systems working simultaneously on the same text, the overall quality will improve. Using this novel approach to MT in the latest version of the Pangloss MT project, we submit an input text to a battery of machine translation systems (engines), collect their (possibly, incomplete) results in a joint chart data structure and select the overall best translation using a set of simple heuristics. 2 INTEGRATING MULTI-ENGINE OUTPUT In our experiment we used three MT engines: * a knowledge-based MT (KBMT) system, the mainline Pangloss engine (Frederking et al., 1993b); • an example-based MT (EBMT) system (see (Nirenburg et al., 1993; Nirenburg et al., 1994b); the original idea is due to Nagao (Na- gao, 1984)); and • a lexical transfer system, fortified with mor- phological analysis and synthesis modules and relying on a number of databases -- a machine-readable dictionary (the Collins Span- ish/English), the lexicons used by the KBMT modules, a large …

Content truncated to 500 words