Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FarFetched: Entity-centric Reasoning and Claim Validation for the Greek Language based on Textually Represented Environments

Dimitris Papadopoulos, Katerina Metropoulou, Nikolaos Matsatsinis, Nikolaos Papadakis

TL;DR

FarFetched addresses automated claim validation for Greek by retrieving and combining evidence from diverse news sources through an entity-centric reasoning pipeline. It constructs a Neo4j graph of articles and WikiData entities, extracts evidence via shortest paths, and ranks candidates with a Greek-English STS model before applying a Greek NLI model to decide entailment, contradiction, or neutrality. The authors demonstrate end-to-end performance on a Greek FEVER subset and report strong STS and competitive NLI results, along with qualitative scenarios showing evidence-driven verdict shifts. The framework is modular and language-agnostic, enabling adaptation to other languages with appropriate models.

Abstract

Our collective attention span is shortened by the flood of online information. With \textit{FarFetched}, we address the need for automated claim validation based on the aggregated evidence derived from multiple online news sources. We introduce an entity-centric reasoning framework in which latent connections between events, actions, or statements are revealed via entity mentions and represented in a graph database. Using entity linking and semantic similarity, we offer a way for collecting and combining information from diverse sources in order to generate evidence relevant to the user's claim. Then, we leverage textual entailment recognition to quantitatively determine whether this assertion is credible, based on the created evidence. Our approach tries to fill the gap in automated claim validation for less-resourced languages and is showcased on the Greek language, complemented by the training of relevant semantic textual similarity (STS) and natural language inference (NLI) models that are evaluated on translated versions of common benchmarks.

FarFetched: Entity-centric Reasoning and Claim Validation for the Greek Language based on Textually Represented Environments

TL;DR

FarFetched addresses automated claim validation for Greek by retrieving and combining evidence from diverse news sources through an entity-centric reasoning pipeline. It constructs a Neo4j graph of articles and WikiData entities, extracts evidence via shortest paths, and ranks candidates with a Greek-English STS model before applying a Greek NLI model to decide entailment, contradiction, or neutrality. The authors demonstrate end-to-end performance on a Greek FEVER subset and report strong STS and competitive NLI results, along with qualitative scenarios showing evidence-driven verdict shifts. The framework is modular and language-agnostic, enabling adaptation to other languages with appropriate models.

Abstract

Our collective attention span is shortened by the flood of online information. With \textit{FarFetched}, we address the need for automated claim validation based on the aggregated evidence derived from multiple online news sources. We introduce an entity-centric reasoning framework in which latent connections between events, actions, or statements are revealed via entity mentions and represented in a graph database. Using entity linking and semantic similarity, we offer a way for collecting and combining information from diverse sources in order to generate evidence relevant to the user's claim. Then, we leverage textual entailment recognition to quantitatively determine whether this assertion is credible, based on the created evidence. Our approach tries to fill the gap in automated claim validation for less-resourced languages and is showcased on the Greek language, complemented by the training of relevant semantic textual similarity (STS) and natural language inference (NLI) models that are evaluated on translated versions of common benchmarks.
Paper Structure (20 sections, 5 figures, 9 tables, 1 algorithm)

This paper contains 20 sections, 5 figures, 9 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Problem Definition
  5. Our approach
  6. News Collection
  7. Graph Database Population
  8. Entity Linking
  9. Evidence Constructor
  10. Semantic Textual Similarity
  11. Natural Language Inference
  12. Experiments
  13. Setup
  14. Main results
  15. End-to-end performance
  16. ...and 5 more sections

Figures (5)

  • Figure 1: Claim validation example (translated from Greek) based on aggregated evidence using FarFetched.
  • Figure 2: The FarFetched modular framework.
  • Figure 3: Final structure of the graph database.
  • Figure 4: Shortest path example: The 3 entities (in brown) are connected with 2 article sections (in blue).
  • Figure 5: Shift in NLI verdict from Scenario 2a to Scenario 2b of Table \ref{['tab:result2']}.