Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Lost in Recursion: Mining Rich Event Semantics in Knowledge Graphs

Florian Plötzky, Niklas Kiehne, Wolf-Tilo Balke

TL;DR

The paper tackles how to represent and compare complex real-world events when observation is mediated by narratives from different narrators. It introduces a recursive narrative model anchored in Event-Centric Knowledge Graphs (ECKGs) and a prompting-based mining algorithm that binds narrative events to KG subgraphs. The method enables multi-viewpoint narratives by modeling viewpoints and recursive event structures, with a proof-of-concept on the Iraq War showing binding to Wikidata for a substantial subset of events. The work highlights the potential for richer semantic grounding and cross-perspective analysis, while noting limitations in time extraction, binding quality, and scalability that require future research.

Abstract

Our world is shaped by events of various complexity. This includes both small-scale local events like local farmer markets and large complex events like political and military conflicts. The latter are typically not observed directly but through the lenses of intermediaries like newspapers or social media. In other words, we do not witness the unfolding of such events directly but are confronted with narratives surrounding them. Such narratives capture different aspects of a complex event and may also differ with respect to the narrator. Thus, they provide a rich semantics concerning real-world events. In this paper, we show how narratives concerning complex events can be constructed and utilized. We provide a formal representation of narratives based on recursive nodes to represent multiple levels of detail and discuss how narratives can be bound to event-centric knowledge graphs. Additionally, we provide an algorithm based on incremental prompting techniques that mines such narratives from texts to account for different perspectives on complex events. Finally, we show the effectiveness and future research directions in a proof of concept.

Lost in Recursion: Mining Rich Event Semantics in Knowledge Graphs

TL;DR

The paper tackles how to represent and compare complex real-world events when observation is mediated by narratives from different narrators. It introduces a recursive narrative model anchored in Event-Centric Knowledge Graphs (ECKGs) and a prompting-based mining algorithm that binds narrative events to KG subgraphs. The method enables multi-viewpoint narratives by modeling viewpoints and recursive event structures, with a proof-of-concept on the Iraq War showing binding to Wikidata for a substantial subset of events. The work highlights the potential for richer semantic grounding and cross-perspective analysis, while noting limitations in time extraction, binding quality, and scalability that require future research.

Abstract

Our world is shaped by events of various complexity. This includes both small-scale local events like local farmer markets and large complex events like political and military conflicts. The latter are typically not observed directly but through the lenses of intermediaries like newspapers or social media. In other words, we do not witness the unfolding of such events directly but are confronted with narratives surrounding them. Such narratives capture different aspects of a complex event and may also differ with respect to the narrator. Thus, they provide a rich semantics concerning real-world events. In this paper, we show how narratives concerning complex events can be constructed and utilized. We provide a formal representation of narratives based on recursive nodes to represent multiple levels of detail and discuss how narratives can be bound to event-centric knowledge graphs. Additionally, we provide an algorithm based on incremental prompting techniques that mines such narratives from texts to account for different perspectives on complex events. Finally, we show the effectiveness and future research directions in a proof of concept.
Paper Structure (16 sections, 4 figures, 1 table)

This paper contains 16 sections, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Events and Narratives
  3. Events
  4. Viewpoints
  5. Narratives
  6. Mining and Binding Narratives
  7. Narrative Relationships
  8. Binding Events in Narratives to ECKGs
  9. Towards Multi-Viewpoint Narrative Mining
  10. Proof of Concept
  11. Case Study and Document Collection
  12. Mining and Binding Narratives
  13. Narrative Comparison
  14. Limitations
  15. Related Work
  16. ...and 1 more sections

Figures (4)

  • Figure 1: A simplified narrative that depicts Vladimir Putin's address to the nation. The events "Redivision of the World" and "Iraq War" are recursive nodes in this example and reflect the inner structure of those events as exemplified in Putin's speech. All events also available in Wikidata are marked with their respective identifier.
  • Figure 2: Narratives and their bindings to ECKGs.
  • Figure 3: Sketch of our narrative mining algorithm.
  • Figure 4: An excerpt of a narrative mined by our algorithm.

Theorems & Definitions (5)

  • Definition 1: Events
  • Definition 2: Complex Events
  • Definition 3: Event-Centric Knowledge Graphs
  • Definition 4: Viewpoints
  • Definition 5: Narratives