Bi-directional Linkability From Wikipedia to Documents and Back Again: UMass at TREC 2012 Knowledge Base Acceleration Track
Abstract
This notebook details the participation of the University of Massachusetts Amherst in the Cumulative Citation Recommendation task (CCR) of the TREC 2012 Knowledge Base Acceleration Track. UMass' objective is to introduce a single model for Knowledge Base Entity Linking and Knowledge Base Acceleration stream filtering using bi-directional linkability between knowledge base (KB) entries and mentions of the entities in documents. Our system focuses on estimating linkability between documents and Knowledge Base entities which measures compatibility in two directions: (1) from a KB entity to documents and (2) from mentions of entities in documents to their KB entries. The entity to document direction, is modeled as a retrieval task where the goal is to identify the most relevant documents for an entity in the evaluation time range. However, if the entity is ambiguous, the retrieved documents may contain documents that are relevant to other entities with the same or similar name. To address this, we want to leverage information from the document to disambiguate the entity. We observe that this problem, from mention to KB entity, is very similar to the TAC Knowledge Base Population Entity Linking Task (Ji et al., 2011). The major goal of our participation is to explore how these two directions, from KB to documents and back can be combined.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2012
- Accession Number
- ADA581455
Entities
People
- Jeffrey Dalton
- Laura Dietz
Organizations
- University of Massachusetts Amherst