TREC-9 Experiments at Maryland: Interactive CLIR

Abstract

The University of Maryland team participated in the TREC-9 Cross-Language Information Retrieval Track, with the goal of exploring evaluation paradigms for interactive cross-language retrieval. Participants were asked to examine gloss translations of highly ranked documents and make relevance judgments, and those judgments were used to produce a new ranked list in which documents assessed as relevant were promoted and those assessed as nonrelevant were demoted. No improvement over fully automatic ranking was found, which suggests that additional work on user interface design and evaluation metrics is required.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2006
Accession Number
ADA456186

Entities

People

  • Clara I. Cabezas
  • Douglas W. Oard
  • Gina-anne Levow

Organizations

  • University of Maryland

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Computer Science
  • Computers
  • Information Operations
  • Information Retrieval
  • Instructions
  • Judgment
  • Language
  • Linguistics
  • Maryland
  • Test And Evaluation
  • Translations
  • Universities
  • User Interface
  • User Interface Engineering

Readers

  • Computational Linguistics
  • Psychological Intervention/Treatment for Stress, Anxiety, PTSD, and Related Emotional and Cognitive Health Symptoms.

Technology Areas

  • AI & ML
  • AI & ML - Information Retrieval
  • AI & ML - Machine Translation