ViHoRec: A Quality-Controlled Vietnamese Hotel Recommendation Dataset and Cold-Start Benchmark
2026-07-14 • Information Retrieval
Information RetrievalArtificial Intelligence
AI summaryⓘ
The authors created ViHoRec, a new Vietnamese hotel recommendation dataset to help improve and test recommendation systems. They combined data from three booking websites, carefully matched hotel names across platforms, and ensured the data is clean and respects user privacy. Their dataset highlights challenges with users who have very little history, showing that simple methods can sometimes work better than complex models. This makes ViHoRec useful for studying recommendation problems in situations with limited data.
recommender systemdatasetentity resolutionprivacy preservationcold-start problemuser interactionbenchmarkcollaborative filteringRecall@10UserKNN
Authors
Minh Hoang Nguyen
Abstract
Recommender-system research for Vietnamese remains limited by the absence of a public, well-documented hotel interaction resource. Building such a resource is challenging for three reasons: cross-platform hotel names must be reconciled before interactions are comparable; quality must be audited with reproducible metrics rather than ad hoc cleaning; and public release must preserve privacy while remaining benchmarkable under realistic cold-start conditions. We introduce ViHoRec, a quality-controlled Vietnamese hotel recommendation dataset of 18{,}267 interactions between 6{,}832 users and 560 hotels, crawled from Booking.com, Traveloka, and Ivivu. Our contributions are: (i) a reproducible construction pipeline with cross-platform entity resolution and quantitative quality control; (ii) a privacy-preserving release with HMAC pseudonyms; and (iii) a public cold-start benchmark with temporal leave-last-one-out split, data-centric ablations, and dependency-free baselines. On the public split, learned models degrade sharply for users with short histories (BPR-MF Recall@10: 0.065 vs. 0.120), while UserKNN remains strongest overall, establishing ViHoRec as a sparse, cold-start-dominated testbed for low-resource recommendation. All data are publicly available at https://github.com/MinhNguyenDS/ViHoRec.