A Personalised Ranking Framework with Multiple Sampling Criteria for Venue Recommendation
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-09899
- Type
- E - Conference contribution
- DOI
-
10.1145/3132847.3132985
- Title of conference / published proceedings
- 26th ACM International Conference on Information and Knowledge Management (CIKM 2017)
- First page
- 1469
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- -
- Year of publication
- 2017
- URL
-
http://eprints.gla.ac.uk/147491/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- 13
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: Enhances BPR, a popular recommendation model, to better suggest venues for users to visit. Addresses 4 identified limitations in the existing approaches. Shows how to combine multiple ways of sampling negative items to aid effective learning. RIGOUR: Gradients for each proposed model are mathematically derived. Experiments are conducted on 3 large-scale real-world datasets from location-based social networks, in comparison to 8 existing baselines. SIGNIFICANCE: Published in a top IR conference. Enhances performance over the best existing baseline by 37%. This work has significant potential impact for improving recommendations made by location-based social networks eg. TripAdvisor, Google Maps, Facebook, Foursquare.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -