View article

In this paper we describe a joint effort by the City University of New York (CUNY), University of Illinois at Urbana-Champaign (UIUC) and SRI International at participating in the mono-lingual entity linking (MLEL) and cross-lingual entity linking (CLEL) tasks for the NIST Text Analysis Conference (TAC) Knowledge Base Population (KBP2011) track. The MLEL system is based on a simple combination of two published systems by CUNY (Chen and Ji, 2011) and UIUC (Ratinov et al., 2011). Therefore, we mainly focus on describing our new CLEL system. In addition to a baseline system based on name translation, machine translation and MLEL, we propose two novel approaches. One is based on a cross-lingual name similarity matrix, iteratively updated based on monolingual co-occurrence, and the other uses topic modeling to enhance performance. Our best systems placed 4th in mono-lingual track and 2nd in cross-lingual track.