Overview of WEB Task at the Fourth NTCIR Workshop

This paper gives an overview of the WEB Task at the Fourth NTCIR Workshop (‘NTCIR-4 WEB’) conducted from 2003 to 2004. Through the NTCIR-4 WEB, we investigated the evaluation methods used to measure some tasks of Web information access, such as information retrieval, information classification, and information extraction. We used a 100-gigabyte document dataset that was mainly gathered from the ‘.jp’ domain. Some evaluation measures were applied to individual system results submitted by the participants, and the evaluated values from various viewpoints were compared.