- Title
- Using analogical proportions for explanations
- Creator
- Lim, Suryani; Prade, Henri; Richard, Gilles
- Date
- 2022
- Type
- Text; Conference paper
- Identifier
- http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/190408
- Identifier
- vital:17634
- Identifier
-
https://doi.org/10.1007/978-3-031-18843-5_21
- Identifier
- ISBN:0302-9743 (ISSN); 9783031188428 (ISBN)
- Abstract
- In this article, we offer an introduction to the notion of analogical explanations. Because analogical reasoning is a widely used type of reasoning, we take the view that analogy-based explanations will be acceptable for humans. The cornerstone of the approach is the concept of analogical proportion (i.e., statements of the form “a is to b as c is to d”), comparing 2 pairs of items. Analogical proportions are not simply based on similarity but also involve differences between items. The approach applies to the explanation of the label of an item in a repository, whether the couple (item, label) belongs to a sample of a given population or the label is predicted via an algorithm. The output can be in terms of abductive/factual explanations (answering a “why?” question and providing examples having the same label) or contrastive/counter-factual (answering a “why not?” question and providing examples having a different label). For preliminary experiments, we build Boolean data sets where relevant attributes are known. Our results show that analogical proportion-based explanations can be effective. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
- Publisher
- Springer Science and Business Media Deutschland GmbH
- Relation
- 15th International Conference on Scalable Uncertainty Management, SUM 2022, Paris, France, 17-19 October 2022, Scalable Uncertainty Management: 15th International Conference, SUM 2022, Paris, France, October 17-19, 2022 Proceedings Vol. 13562 LNAI, p. 309-325
- Rights
- All metadata describing materials held in, or linked to, the repository is freely available under a CC0 licence
- Rights
- Copyright © 2022, The Author(s)
- Reviewed
- Hits: 89
- Visitors: 88
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|