TopicMF: Simultaneously Exploiting Ratings and Reviews for Recommendation
Authors: Yang Bao, Hui Fang, Jie Zhang
AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on 22 real-world datasets show the superiority of our model over the state-of-the-art models, demonstrating its effectiveness for recommendation tasks. |
| Researcher Affiliation | Academia | Yang Bao Hui Fang Jie Zhang Nanyang Business School, Nanyang Technological University, Singapore School of Computer Engineering, Nanyang Technological University, Singapore {baoyang@ntu.edu.sg, hfang@e.ntu.edu.sg, zhangj@ntu.edu.sg} |
| Pseudocode | No | The paper describes the model's optimization steps and refers to existing algorithms but does not provide structured pseudocode or an algorithm block for its own method. |
| Open Source Code | No | The paper references third-party software libraries used for implementation (e.g., My Media Lite, liblbfgs, NMF code from CJ Lin) but does not state that the code for their proposed Topic MF model is open-source or publicly available. |
| Open Datasets | Yes | To make the fair evaluation of our model, we directly generate the 22 Amazon datasets from the datasets provided by (Mc Auley and Leskovec 2013). |
| Dataset Splits | No | The paper states: 'we randomly subdivide each dataset into the training and testing sets, where 80% of each dataset is used for training, and the rest is for testing', but does not explicitly mention a separate validation split or how hyperparameters were tuned. |
| Hardware Specification | No | The paper mentions 'due to our hardware limitation' but does not provide specific details about the CPU, GPU, or other hardware used for the experiments. |
| Software Dependencies | No | The paper mentions using 'My Media Lite Recommender System Library', 'L-BFGS' (with URL), and 'NMF' (with URL) but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | Yes | For all the methods, we set optimal parameters recommended in the literature, and set the number of latent factors (K) as 5. For our method, the balance parameter λ is set to 1, and λu = λv = λB = 0.001, while other parameters are fit using L-BFGS package and projected gradient technique. Firstly, we fix λ to its default value 1, and vary the number of latent factors to be 5, 10, 15, and 20, respectively. Next, we fix the parameter K to 5, and vary λ to be values of {0.1, 0.2, 0.5, 1, 2, 5, 10}. |