Text understanding and common sense learning from narrative texts

Document Type

Conference Proceeding

Publication Date

1-1-2020

Abstract

In this work, we develop a hybrid SVM / rule-based classification method for identification of scene-level contexts and subsequent extraction of temporal, spatial, and causal relations from within and between these contexts within narrative-style texts (such as novels and investigative news stories). We also develop methods for generalizing from narratives to larger contexts - aka, "common sense" type knowledge. Knowledge extraction results are compared against gold standard human annotation of a small dataset of 20 narrative stories across a mix of genre (news, fiction, non-fiction).

Publication Source (Journal or Book title)

Proceedings of the 2016 Industrial and Systems Engineering Research Conference, ISERC 2016

First Page

2098

Last Page

2103

This document is currently not available here.

Share

COinS