Cross-lingual information retrieval (CLIR) is considered very important, since it results in more flexible information retrieval systems. This work presents an interesting approach for information retrieval in collections of Korean documents. Both monolingual and cross-lingual (CLIR) approaches are studied. For cross-lingual IR involving Korean and Chinese, English is used as the pivot or intermediate language.

Two key issues are investigated. The first is the effectiveness of IR for Korean document collections without extensive Korean text processing requirements. A similar approach has been followed for Chinese, with very good results. The second issue is the effectiveness of using the English language as an intermediate (pivot) for performing cross-lingual IR between two Asian languages (for example, Korean and Chinese). Since the English language is spoken extensively worldwide, there is a good chance that translation resources exist between language X and English, and between English and language Y. Therefore, the two Asian languages X and Y can be associated through the pivot language (English).

The experimental results provided support the following findings: bilingual CLIR from English to Korean has an accuracy of 71 to 79 percent compared to Korean to Korean; bilingual CLIR from Chinese to English has an accuracy of 76 to 84 percent compared to English to English; and using English as the pivot language, Chinese to Korean CLIR has an accuracy of 55 to 65 percent compared to the usage of Korean only.

Developers of Asian language search engines, as well as developers of special-purpose IR systems for Asian languages, will benefit from this research.