Overview of the TREC 2007 Enterprise Track

The collection consists of all the *.csiro.au (public) websites as they appeared in March 2007. The resulting data set consists of 370 715 documents, with total size 4.2 gigabytes. The web crawler visited the outward-facing pages of CSIRO in a fashion similar to the crawl used in CSIRO’s own search engine. In fact, the same crawler technology that CSIRO uses was used to gather the CSIRO documents (http://www.funnelback.com/). The corpus contains approximately 7.9 million hyperlinks, and 95% of pages have one or more outgoing links containing anchor text. One participant extracted email addresses of 3678 individuals, with 38% of documents containing at least one mailto field.