HLTCOE Approaches to Knowledge Base Population at TAC 2009

Maryland Shared Open Access Repository (USMAI Consortium)Published 1 November 2009Open access

Paul McNamee, Mark Dredze, Adam Gerber, Nikesh Garera, Tim Finin, James Mayfield

Citations39

TL;DR

The HLTCOE participated in the entity linking and slot filling tasks at TAC 2009 and found that slot-filling based on sentence selection, application of weak patterns and exploitation of redundancy was ineffective in the slot filling task.

Abstract

The HLTCOE participated in the entity linking and slot filling tasks at TAC 2009. A machine learning-based approach to entity linking, operating over a wide range of feature types, yielded good performance on the entity linking task. Slot-filling based on sentence selection, application of weak patterns and exploitation of redundancy was ineffective in the slot filling task.

Keywords

Computer ScienceDecision Sciences

Optimizing search engines using clickthrough data

3,898 Citations2002Thorsten Joachims

The goal of this paper is to develop a method that utilizes clickthrough data for training, namely the query-log of the search engine in connection with the log of links the users clicked on in the presented ranking.

A Comparison of Personal Name Matching: Techniques and Practical Issues

333 Citations2006Peter Christen

The characteristics of personal names are discussed and potential sources of variations and errors are presented and a comprehensive number of commonly used, as well as some recently developed name matching techniques are overview.

Proceedings of the International AAAI Conference on Web and Social MediaWikipedia as an Ontology for Describing Documents

130 Citations2021Zareen Syed, Tim Finin +1 more

This work uses Wikipedia articles and the category and article link graphs to predict concepts common to a set of documents and describes several algorithms to aggregate and refine results, including the use of spreading activation to select the most appropriate terms.

View PDF

Latent-variable modeling of string transductions with finite-state methods

98 Citations2008Markus Dreyer, Jason Smith +1 more

A conditional loglinear model is presented for string-to-string transduction that employs overlapping features over latent alignment sequences, and which learns latent classes and latent string pair regions from incomplete training data, and it is demonstrated that latent variables can dramatically improve results, even when trained on small data sets.

View PDF

Structural, transitive and latent models for biographic fact extraction

29 Citations2009Nikesh Garera, David Yarowsky

Six novel approaches to biographic fact extraction that model structural, transitive and latent properties of biographical data are presented, achieving an average extraction accuracy of 80% over seven types of biographic attributes.

View PDF

A comparison of techniques for estimating IDF values to generate lexical signatures for the web

16 Citations2008Martin Klein, Michael L. Nelson

A comparison study of different methods to estimate IDF values of web pages finds a term overlap of 70 to 80% between the results of the two methods and the baseline, and a great agreement in rank correlation of TF-IDF ranked terms between the authors' methods.