Towards interpretable, data-derived distributional semantic representations for reasoning: A dataset of properties and concepts

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

This paper proposes a framework for investigating which types of semantic properties are represented by distributional data. The core of our framework consists of relations between concepts and properties. We provide hypotheses on which properties are reflected in distributional data or not based on the type of relation. We outline strategies for creating a dataset of positive and negative examples for various semantic properties, which cannot easily be separated on the basis of general similarity (e.g. fly: seagull, penguin). This way, a distributional model can only distinguish between positive and negative examples through evidence for a target property. Once completed, this dataset can be used to test our hypotheses and work towards data-derived interpretable representations.

Original languageEnglish
Title of host publicationProceedings of the tenth Global WordNet Conference
EditorsChristiane Fellbaum, Piek Vossen, Ewa Rudnicka, Marek Maziarz, Maciej Piasecki
Place of PublicationWroclaw
PublisherOficyna Wydawnicza Politechniki Wroclawskiej
Pages85-98
Number of pages14
ISBN (Electronic)9788374931083
ISBN (Print)9788374931083
Publication statusPublished - 2019
Event10th Global WordNet Conference, GWC 2019 - Wroclaw, Poland
Duration: 23 Jul 201927 Jul 2019

Conference

Conference10th Global WordNet Conference, GWC 2019
CountryPoland
CityWroclaw
Period23/07/1927/07/19

Fingerprint Dive into the research topics of 'Towards interpretable, data-derived distributional semantic representations for reasoning: A dataset of properties and concepts'. Together they form a unique fingerprint.

Cite this