Improving navigation in critique graphs
Loading...
Files
Accepted Version
Date
2016-11
Authors
Genc, Begum
O'Sullivan, Barry
Journal Title
Journal ISSN
Volume Title
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published Version
Abstract
Critique graphs were introduced as a device for analysing the behaviour of conversational recommender systems. A conversational recommender allows a user to critique a recommended product with statements such as "I'd like a similar product to this one, but cheaper". A critique graph is a directed multigraph in which the nodes represent products, and a directed edge between a pair of products represents how a user can move from one product to another by tweaking a particular product feature. It has been shown that critique graphs are not symmetric: if a user critiques a product pi and is presented with product pj, critiquing product pj in the opposite manner does not necessarily return product pi. Furthermore, it might not be possible to reach all products in a catalogue starting from a given product, or as a consequence of a particular critique some products become unreachable. This latter point is quite unsatisfactory since a user would assume that it is possible to explore the full catalogue by critiquing alone. A number of approaches to overcoming this problem have been proposed in the literature. In this paper we propose a novel approach that exploits the critique graph directly. Specifically, the unreachability is a consequence of a critique graph having more than one strongly connected component. We show how the critique graph can be modified in a minor way, thereby modifying the semantics of critiquing for a given catalogue, so that all products are always reachable.
Description
Keywords
Strongly connected components , Recommender systems , Critique graphs , Semantics , Recommender systems , Navigation , Data analysis , Computer science , Compounds , Standards
Citation
Genc, B. and O'Sullivan, B. (2016) 'Improving navigation in critique graphs', 2016 IEEE 28th International Conference on Tools with Artificial Intelligence (ICTAI), San Jose, CA, USA, 6-8 November. doi:10.1109/ICTAI.2016.0030
Link to publisher’s version
Copyright
© 2016, IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.