Prof Thien Nguyen receives a Grant from the Army Research Office to investigate Text Structures for Information Extraction in Natural Language Processing. Events (protests, disease outbreaks, natural disasters) are prevalent in natural language texts from different languages and domains (news, books, social media, scientific papers). Extracting events from text is crucial to document understanding and intelligent systems such as question answering, text summarization, fake news detection, future event prediction, and decision making for resource allocation. Existing methods for Event Extraction (EE) have featured advanced deep learning architectures to effectively capture patterns of events in texts; however, the performance of those methods is not yet satisfactory, limiting their applications in different domains and languages.
The project, titled "Boosting and Extending Event Extraction using Interaction Structures from Texts", will present a new generation of models for Neural Event Extraction, proposing text structures (e.g., sentence or document structures) to boost the performance and extend the operation of current EE methods. Using graphs to directly model the dependency relations between objects of interest in text (i.e., words, entities, events), text structures facilitate the integration of various sources of information (i.e., syntax, semantics, background knowledge) to identify important context in text to aid the prediction tasks. Further, as text structures tend to be more invariant when we switch to texts of new information types, domains, and languages, text structure-based models for EE can effectively transfer the knowledge and achieve better performance in such new settings.
The project is funded by the Army Research Office (ARO) under Grant W911NF-21-1-0112 with an amount of (approximately) $353,000 for three years (April 2021 - March 2024). ARO serves as the Army’s principal agent for the planning, organization, selection, and management of extramural basic research in the physical sciences, engineering sciences, and informational sciences.