Enriching an authority file of scientific conferences with information extracted from the web

Abstract

Authority files maintain variant forms to refer to the same entity and they are very useful in digital libraries. However, collect data and keep an updated authority file is not a trivial task. This paper proposes an approach for the enrichment of a publication venue authority file by extracting information on conferences from their web pages. Collecting additional data is important to improve the effectiveness of data disambiguation tools and information retrieval, such as those that measure the quality of a scientific publication based on bibliometrics (e.g., Journal Impact Factor). Most applications use only basic citation metadata, such as author's names, work and publication venue titles. However, data external to the publication, contained in the publication venue web page, can be very useful in the disambiguation task. Our approach includes the steps for querying a web search engine, classifying documents obtained in the result sets and extracting information from the relevant pages. We evaluated two methods for classifying documents, one based on genre and content and one based on content only. The experiments show good results to trace a history of conference editions, with data such as URL, year of each edition and dates of changing in their names.