English-Chinese CLIR using a Simplified PIRCS System

English-Chinese CLIR using a Simplified PIRCS System K.L. Kwok, N. Dinstl and P. Deng Computer Science Department, Queens College, CUNY 65-30 Kissena Blvd. Flushing, N.Y. 11367 kwok@ir.cs.qc.edu ABSTRACT A GUI is presented with our PIRCS retrieval system for supporting English-Chinese cross language information retrieval. The query translation approach is employed using the LDC bilingual wordlist. Given an English query, different translation methods and their retrieval results can be demonstrated. 1. INTRODUCTION The purpose of cross language information retrieval (CLIR) is to allow a user to search, retrieve, and gain some content understanding of documents written in a language different from the one that the user is familiar with. This is to be accomplished automatically without expert linguist assistance. CLIR is of growing importance because it can literally open up a whole world of information for the user, especially with the ease and convenience of access and delivery of foreign documents provided by Internet logistics nowadays. Searching and retrieving Chinese documents via English is a major sub-problem within CLIR because many people in the world use these two languages. For example, one would expect trade between China and the U.S. (and other countries) to grow significantly in the near future because of the impending WTO membership for China. Monitoring trends and status information from Chinese sources may be an essential operation for organizations interested in these affairs. Chinese is a language completely different from English, and it is conceived to be difficult for foreigners to learn. This paper describes some of the methods that we employ to deal with this problem, and presents a demonstrable system to illustrate the workings of cross language document retrieval. In Section 2, techniques for the query translation approach to CLIR are discussed. Section 3 contains a description of our simplified PIRCS retrieval system that is the basis for monolingual retrieval. Section 4 describes the GUI supporting interactive query input, document output and other implementation issues, and Section 5 contains our conclusion and future work. 2. STRATEGY FOR CROSS LANGUAGE INFORMATION RETRIEVAL When faced with the situation of a language mismatch between the target documents and the query (information need statement) of a user, one could reduce them to a common representation language for retrieval purposes by automatically translating the query to the document language, by translating the documents to the query language, or by converting both to a third representation language [1]. By far the simplest and most common approach seems to be the first method, and probably as effective as the others, and we have also taken this route. The question is what tools to use for query translation. It is well known that machine translation is generally fuzzy and inaccurate [6]. This is particularly true when translation output are judged by humans, who tend to be unforgiving. However, translation for machine consumption (such as for information retrieval (IR)) may not be so bad because IR can operate with a bag of content terms without grammar, coherence or readability. What IR needs is that …

Content truncated to 500 words