Bouraoui, Zied and Schockaert, Steven ORCID: https://orcid.org/0000-0002-9256-2881 2018. Learning conceptual space representations of interrelated concepts. Presented at: 27th International Joint Conference on Artificial Intelligence and the 23rd European Conference on Artificial Intelligence (IJCAI-ECAI-18), Stockholm, Sweden, 13-19 July 2018. Published in: Lang, Jerome ed. Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence Organization, pp. 1760-1766. 10.24963/ijcai.2018/243 |
Preview |
PDF
- Accepted Post-Print Version
Download (224kB) | Preview |
Abstract
Several recently proposed methods aim to learn conceptual space representations from large text collections. These learned representations associate each object from a given domain of interest with a point in a high-dimensional Euclidean space, but they do not model the concepts from this domain, and can thus not directly be used for categorization and related cognitive tasks. A natural solution is to represent concepts as Gaussians, learned from the representations of their instances, but this can only be reliably done if sufficiently many instances are given, which is often not the case. In this paper, we introduce a Bayesian model which addresses this problem by constructing informative priors from background knowledge about how the concepts of interest are interrelated with each other. We show that this leads to substantially better predictions in a knowledge base completion task.
Item Type: | Conference or Workshop Item (Paper) |
---|---|
Status: | Published |
Schools: | Computer Science & Informatics |
Publisher: | International Joint Conferences on Artificial Intelligence Organization |
ISBN: | 9780999241127 |
Date of First Compliant Deposit: | 18 July 2018 |
Date of Acceptance: | 4 May 2018 |
Last Modified: | 23 Oct 2022 14:04 |
URI: | https://orca.cardiff.ac.uk/id/eprint/112685 |
Citation Data
Cited 9 times in Scopus. View in Scopus. Powered By Scopus® Data
Actions (repository staff only)
Edit Item |