Person attribute extraction from the textual parts of web pages

Abstract

We present a web mining system that clusters persons sharing the same name and also extracts bibliographical information about them. The input of our system is the result of web search engine queries in English or in Hungarian. For system evaluation in English, our system (RGAI) participated in the third Web People Search Task challenge [1]. The chief characteristics of our approach compared to the others are that we focus on the raw textual parts of the web pages instead of the structured parts, we group similar attribute classes together and we explicitly handle their interdependencies. The RGAI system achieved top results on the person attribute extraction subtask, and average results on the person clustering subtask. Following the shared task annotation principles, we also manually constructed a Hungarian person disambiguation corpus and adapted our system from English to Hungarian. We present experimental results on this as well.