Diversified Interactive Recommendation with Implicit Feedback

Authors: Yong Liu, Yingtai Xiao, Qiong Wu, Chunyan Miao, Juyong Zhang, Binqiang Zhao, Haihong Tang4932-4939

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on real datasets are performed to demonstrate the effectiveness of the proposed method in balancing the recommendation accuracy and diversity.
Researcher Affiliation Collaboration 1Alibaba-NTU Singapore Joint Research Institute 2Joint NTU-UBC Research Centre of Excellence in Active Living in the Elderly (LILY) 3School of Computer Science and Engineering, Nanyang Technological University 4University of Science and Technology of China, 5Alibaba Group
Pseudocode Yes Algorithm 1 Thompson Sampling for DC2B; Algorithm 2 DPP Greedy Search S O( θ, Xt)
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository for their proposed method.
Open Datasets Yes Movielens-100K, Movielens-1M1, and Anime2. ... 1https://grouplens.org/datasets/movielens/ 2https://www.kaggle.com/Cooper Union/animerecommendations-database
Dataset Splits No A validation set is sampled from training data to choose hyper-parameters. (This mentions a validation set but does not provide specific details on its size or how it's sampled, making its split not fully reproducible.)
Hardware Specification No The paper does not provide any specific hardware details (e.g., CPU/GPU models, memory) used for conducting the experiments.
Software Dependencies No The paper mentions using BPRMF for embeddings but does not list any specific software dependencies with version numbers (e.g., Python, PyTorch versions).
Experiment Setup Yes Empirically, we set the dimensionality of the item embeddings to 10. ... N is set to 10, 30, and 50. ... For all methods, we empirically set the size of St to 10 in each trial. ... α is set to 0.9 for MMR. ϵ is set to 0.1 for ϵ-Greedy, and θ is set to 0.6 for DPPmap. In C2UCB, we set λ0 = 100, λ = 0.1, and σ = 1. In ECBandit, we set the parameter λ = 1. For DC2B, we empirically set α = 3, and λ = 1, on all datasets.