ADAPT_TCD: An Ontology-Based Context Aware Approach for Contextual Suggestion
Item Type:Conference Paper
Citation:Mostafa Bayomi, S?amus Lawless, ADAPT_TCD: An Ontology-Based Context Aware Approach for Contextual Suggestion, Twenty-Fifth Text REtrieval Conference (TREC 2016), Gaithersburg, Maryland, United States, November 15--18, Ellen M. Voorhees and Angela Ellis, National Institute for Standards and Technology (NIST), 2017
bayomi_lawless_trec2016_cs.pdf (PDF) 564.1Kb
In this paper we give an overview of the participation of the ADAPT Centre, Trinity College Dublin, Ireland, in both phases of the TREC 2016 Contextual Suggestion Track. We present our ontology-based approach that consists of three models that are based on an ontology that was extracted from the Foursquare category hierarchy. The three models are: User Model, Document Model and Rule Model. In the User Model we build two models, one for each phase of the task, based upon the attractions that were rated in the user’s profile. The Document Model enriches documents with extra metadata from Foursquare and categories (concepts) from the ontology are attached to each document. The Rule model is used to tune the score for candidate suggestions based on how the context of the trip aligns with the rules in the model. The results of our submitted runs, in both phases, demonstrate the effectiveness of the proposed methods.
Publisher:National Institute for Standards and Technology (NIST)
Type of material:Conference Paper
Availability:Full text available