Application of a Word-Alignment Algorithm to Bilingual Greek-Latin Documents

Abstract:- In this paper we address the problem of improving the access to bilingual Greek – Latin documents by incorporating their word level parallelization into the search process. More specifically we focus our experimentation on finding word correspondences between the Greek and Latin versions of the Gospel of Matthew by the utilization of the K – vec algorithm. We demonstrate the applicability of the algorithm in this broader class of language pairs by measuring its efficiency in identifying word correspondences that are valid mutual translations. Our implementation and evaluation of the K – vec algorithm for this specific language pair verifies that the 75 % of the top ranked candidate word correspondences produced by the algorithm are valid mutual translations. Key-Words:- parallel corpora, word alignment, statistical language processing, bilingual document processing, greek-latin text