Juicer: Scalable Extraction for Thread Meta-information of Web Forum

Abstract

In Web forum, thread meta-information contained in list-of-thread of board page provide fundamental data for the further forum mining. This paper describes a complete system named Juicer which was developed as a subsystem for an industrial application that involves forum mining. The task of Juicer is to extract thread meta-information from board pages of a great many of large scale online Web forums, which implies that scalable extraction is required with high accuracy and speed, and minimal user effort for maintenance. Among so many existed approaches about information extraction, we can not find any approach to fully satisfy the requirements, so we present simple scalable extraction approach behind Juicer to achieve the goal. Juicer is constituted by four modules: Template generation, Specifying labeling setting, Automatic extraction, Label assignment. Both experiments and practice show that Juicer successfully satisfied the requirements.

This work is partially supported by the National Grand Fundamental Research 973 Program of China under Grant 2004CB318109, and the National High Technology Development 863 Program of China under Grant 2007AA01Z147.