Abstract
Our approach to the CLEF NewsREEL 2016 News RecommendationEvaluation Lab investigates the connection between imagesand users clicking behavior. Our goal is to gain a better understanding ofthe contribution of visual representations accompanying images (thumbnails)to the success of news recommendation algorithms as measured bystandard metrics. We experiment with visual information, namely FaceDetection and Saliency Map, extracted from the images that accompanynews items to see if they can be used to chose news items that have ahigher chance of being clicked by users. Initial results seems to suggestgreat CTR improvement in the Simulated Environment task, while somedecrease in performance has been found in the Living Lab task. Thelatter result must be further validated in the future.
Original language | English |
---|---|
Title of host publication | CLEF 2016 Working Notes |
Subtitle of host publication | Conference and Labs of the Evaluation Forum |
Editors | Krisztian Balog, Linda Cappellato, Nicola Ferro, Craig Macdonald |
Publisher | CEUR-WS |
Pages | 618-627 |
Number of pages | 10 |
Publication status | Published - 2016 |
Event | CLEF 2016 - Conference and Labs of the Evaluation Forum: Information Access Evaluation meets Multilinguality, and Interaction - University of Évora, Évora, Portugal Duration: 5 Sept 2016 → 8 Sept 2016 Conference number: 7th http://clef2016.clef-initiative.eu/ |
Publication series
Name | CEUR Workshop Proceedings |
---|---|
Publisher | CEUR |
Volume | 1609 |
ISSN (Electronic) | 1613-0073 |
Conference
Conference | CLEF 2016 - Conference and Labs of the Evaluation Forum |
---|---|
Country/Territory | Portugal |
City | Évora |
Period | 5/09/16 → 8/09/16 |
Internet address |
Keywords
- Recommender System
- News
- Image Analysis
- Face Detection
- Saliency Map
- Evaluation