| | |
Summary: Predicting Extraction Performance using Context Language Models
Eugene Agichtein Silviu Cucerzan
Microsoft Research, Redmond, WA, USA
{eugeneag, silviu}@microsoft.com
ABSTRACT
Exploiting lexical and semantic relationships in text can
dramatically improve information retrieval accuracy. Most
notably, named entities and relations between entities are crucial
for effective question answering and other information retrieval
tasks. Unfortunately, the success in extracting these relationships
can vary for different domains and document collections.
Predicting extraction performance is an important step towards
integration of information extraction technology for high accuracy
information retrieval. In this paper, we present a general language
modeling method for quantifying the difficulty of information
extraction tasks. We demonstrate the viability of our approach by
predicting extraction performance of two real world tasks, Named
Entity Recognition and Relation Extraction.
Categories and Subject Descriptors
H.3.1 [INFORMATION STORAGE AND RETRIEVAL]:
|
