Discrete Social Recommendation
Authors: Chenghao Liu, Xin Wang, Tao Lu, Wenwu Zhu, Jianling Sun, Steven Hoi208-215
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on three real-world datasets demonstrate that DSR runs nearly 5 times faster and consumes only with 1/37 of its real-value competitor s memory usage at the cost of almost no loss in accuracy. In this section, we carry out extensive experiments on several real-world datasets and compare our proposed model with several state-of-the-art algorithms to demonstrate the advantages of the proposed DSR approach. |
| Researcher Affiliation | Collaboration | 1School of Information Systems, Singapore Management University, Singapore 2School of Computer Science and Technology, Zhejiang University, China 3Department of Computer Science and Technology, Tsinghua University, China 4Alibaba-Zhejiang University Joint Institute of Frontier Technologies, China |
| Pseudocode | Yes | Algorithm 1 depicts the details of the proposed DSR model. Algorithm 1 Discrete Social Recommendation Input: {Rij|i, j V}, {Sij|i, j N}: observed user-item ratings and user-user similarity in trust relationship. 1: Initialize B, D, F and X, Y, Z by Equation (4) 2: repeat 3: for i = 1 to n do 4: repeat 5: for k = 1 to r do 6: Update bik via the Equation (7). 7: end for 8: until converge 9: end for Update D and F according to Equation (10) and Equation (12) respectively. 10: Update X according to Equation (14). 11: Update of Y and Z in a similar way with X. 12: until converge 13: return B, D, F. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. No specific repository link or explicit code release statement is found. |
| Open Datasets | Yes | The evaluations are conducted on three public datasets from different real-world websites. Epinions: This dataset is collected in a 5-week crawl (November/December 2003) from a product review website 1. Users in Epinions are allowed to specify scores from 1 to 5 to rate items, and they can also establish relations with others. Film Trust: This dataset is extracted from the entire Film Trust website 2 in June, 2011. The dataset contains useruser friendships and user-movie ratings. Ciao DVD: This dataset is crawled from the entire category of DVDs on a UK DVD community website 3 in December, 2013. The dataset contains trust relationships among users as well as their ratings on DVDs. 1http://www.epinions.com/ 2http://trust.mindswap.org/Film Trust/ 3http:// dvd.ciao.co.uk |
| Dataset Splits | No | For the in-matrix recommendation task, we randomly sample 80% ratings as training and the rest 20% as testing for each user. For the out-matrix recommendation task, we randomly sample 80% users and put all ratings made by them into the training set, and then put the ratings made by the rest users into the test set. The paper describes training and testing splits but does not mention a separate validation set. |
| Hardware Specification | Yes | All the experiments are conducted on a computer equipped with an Intel(R) Core(TM) i5-7200U CPU @2.50GHZ, 16GB RAM and 64-bit Windows 10 operating system. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment. |
| Experiment Setup | Yes | The hyper-parameters α0 is tuned within the range of [10 3, . . . , 102], β1, β2 are tuned within the range of [10 5, . . . , 1], and β3 is tuned within the range of [10 4, . . . , 101]. |