Bilingual PRESRI - Integration of Multiple Research Paper Databases

Collecting all the papers in a research field is a first step towards an exhaustive survey. A number of research paper databases are available for searching papers. However, searchers are compelled to repeat the same search operation for each database if there are multiple databases for a research field. To improve such inefficient searching, we have developed PRESRI, which can construct an exhaustive database by integrating multiple research paper databases. First, we collect Postscript and PDF files on the WWW, and construct a database ('WEB-DB') by extracting bibliographic information from the files. Second, we construct an exhaustive database by integrating WEB-DB with other databases. As a key technique for constructing an exhaustive database, we propose a method for extracting bibliographic information from Postscript and PDF files based on a SVM. To investigate the effectiveness of our method, we conducted an examination. We found that our method is useful for both Japanese and English. In this paper, we also focus on the presentation of search results, which is an important factor in constructing an efficient survey environment. We have developed a system that makes it possible to understand the relationships between papers intuitively based on citation information.