Affect Units and Narrative Summarization.

Abstract

The analysis of narrative text involves various levels of description. On the lowest level are word meanings and syntactic structure within single sentences. On a higher level there are problems of generating inferences and integrating information into memory. At the highest level is the notion of a macro-structure or narrative plot. The identification of high-level narrative structures is central to the problem of narrative summarization. But the intuitive notion of a plot is useless for a process model of summarization unless we can specify the hierarchical representations that allow us to analyze input and produce plots as output. A representational strategy has been developed for high-level structural analysis in conjunction with the BORIS system (a narrative text understanding system). The structures produced effectively encode plot lines in terms of connected graph structures where graph nodes correspond to specific affect units. An affect unit is an abstraction of affective causalty which is recognized in a bottom-up manner at the time of understanding. Simple manipulations of these graph structures yield the conceptual basis for narrative summaries, but the actual process of summarization depends on certain connectivity properties in the graph structure.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 1980
Accession Number
ADA086735

Entities

People

  • Wendy G. Lehnert

Organizations

  • Yale University

Tags

Communities of Interest

  • C4I
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Automated Text Summarization
  • Base Lines
  • Cognitive Science
  • Computer Science
  • Construction
  • Electrical Engineering
  • Identification
  • Information Systems
  • Military Research
  • Natural Language Processing
  • New York
  • Pattern Recognition
  • Recognition
  • Standards
  • Structural Analysis

Readers

  • Computational Linguistics
  • Graph Algorithms and Convex Optimization.
  • Regression Analysis.

Technology Areas

  • AI & ML