Linguistic Summaries as Explanation Mechanism for Classification Problems

Research output: Contribution to conferencePaperAcademic

133 Downloads (Pure)

Abstract

The amount and complexity of generated and collected data is rapidly growing. As a consequence, it is increasingly hard to understand the data and extract useful information. Transparency, interpretability and understandability contribute towards explainability of the data, which is crucial for the user for both efficient and effective usage of it and trust in these data-based decisions. In this paper, we investigate how linguistic summaries can serve as an explanation mechanism for classification results. Linguistic summaries are template-based, semi-natural language-like sentences that can verbalize these (classification) patterns. We develop linguistic summarizations for the classification
results of two publicly available data sets and perform an initial evaluation with a small group of potential users. The preliminary results look promising.
Original languageEnglish
Publication statusPublished - Nov 2022
EventThe 34th Benelux Conference on Artificial Intelligence and the 31th Belgian Dutch Conference on Machine Learning - Mechelen, Belgium
Duration: 7 Nov 20229 Nov 2022

Conference

ConferenceThe 34th Benelux Conference on Artificial Intelligence and the 31th Belgian Dutch Conference on Machine Learning
Abbreviated titleBNAIC 2022
Country/TerritoryBelgium
CityMechelen
Period7/11/229/11/22

Keywords

  • Explainable AI
  • Linguistic Summaries
  • Classification

Fingerprint

Dive into the research topics of 'Linguistic Summaries as Explanation Mechanism for Classification Problems'. Together they form a unique fingerprint.

Cite this