Future Internet 2014, 6, 518-541, doi:10.3390/fi6030518, August 2014
Abstract. The World Wide Web is the largest information repository available today. However, this information is very volatile and Web archiving is essential to preserve it for the future. Existing approaches to Web archiving are based on simple definitions of the scope of Web pages to crawl and are limited to basic interactions with Web servers. The aim of the ARCOMEM project is to overcome these limitations and to provide flexible, adaptive and intelligent content acquisition, relying on social media to create topical Web archives. In this article, we focus on ARCOMEM’s crawling architecture. We introduce the overall architecture and we describe its modules, such as the online analysis module, which computes a priority for the Web pages to be crawled, and the Application-Aware Helper which takes into account the type of Web sites and applications to extract structure from crawled content. We also describe a large-scale distributed crawler that has been developed, as well as the modifications we have implemented to adapt Heritrix, an open source crawler, to the needs of the project. Our experimental results from real crawls show that ARCOMEM’s crawling architecture is effective in acquiring focused information about a topic and leveraging the information from social media.