Researchers at the University of Texas at Austin (UT Austin) aim to improve information retrieval (IR) systems and enhance search engines by integrating artificial intelligence with annotation insights and information encoded in domain-specific resources.
One IR method combines input from multiple annotators to ascertain the best overall annotation for a given text.
UT Austin professor Matthew Lease's team was able to train a neural network to accurately predict named entities and extract pertinent information in unannotated texts.
The second technique proposes exploiting existing linguistic resources via weight sharing to augment natural language processing models for automatic text classification.
"If you could somehow reason about some words being related to other words a priori, then instead of having to have a parameter for each one of those words separately, you could tie together the parameters across multiple words and in that way need less data to learn the model," Lease says.
From Texas Advanced Computing Center
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found