Exploiting Fine-Grained Skip Behaviors for Micro-Video Recommendation

Authors: Sanghyuck Lee, Sangkeun Park, Jaesung Lee

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments demonstrated that the proposed method outperformed three conventional methods across eight evaluation measures on two public datasets. ... As shown in Tables 2 and 3, the two experimental results demonstrate that the proposed method outperforms the comparison models across two different datasets.
Researcher Affiliation Academia Department of Artificial Intelligence, Chung-Ang University, Seoul, Republic of Korea EMAIL
Pseudocode No The paper describes the proposed method in narrative text and mathematical formulations (Equations 1-20), but it does not contain a clearly labeled pseudocode or algorithm block.
Open Source Code No The paper does not provide any statement regarding the availability of source code or links to a code repository.
Open Datasets Yes The MVA dataset (Shang et al. 2023) was collected from a mobile app platform. ... The Kuai Rand-Pure dataset (Gao et al. 2022b), collected from the Kuaishou app, one of the largest video-sharing platforms in China, provides an unbiased sequential recommendation dataset.
Dataset Splits Yes The entire set of interactions is randomly sampled at a 6:2:2 ratio for each user to generate the training, validation, and test sets.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies Yes All models were implemented using Py Torch 2.3.
Experiment Setup Yes Each model was trained with a batch size of 1024 using the Adam W optimizer (Loshchilov and Hutter 2019) with a momentum of 0.9 and a weight decay of 1e-4. The learning rate starts at 1e-3 and decays to 1e-6 following a cosine annealing schedule. ... Early stopping is applied if the recall@3 on the validation set does not improve for five consecutive epochs. The feature dimension d is set to 128, and λ is set to 0.5.