Interlinking Cross Language Metadata Using Heterogeneous Graphs and Wikipedia

Abstract

Cross-language metadata are essential in helping users overcome language barrier in information discovery and recommendation. The construction of cross-language vocabulary, however, is usually costly and intellectually laborious. This paper addresses these problems by proposing a Cross-Language Metadata Network (CLMN) approach, which uses Wikipedia as the intermediary for cross-language metadata linking. We conducted an experiment with key metadata in two digital libraries and in two different languages without using machine translation. The experiment result is encouraging and suggests that the CLMN approach has the potential not only to interlink metadata in different languages with reasonable rate of precision and quality but also to construct cross-language metadata vocabulary. Limitations and further research are also discussed.