Joint Optimization of Tree-based Index and Deep Model for Recommender Systems
Authors: Han Zhu, Daqing Chang, Ziru Xu, Pengye Zhang, Xiang Li, Jie He, Han Li, Jian Xu, Kun Gai
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental evaluations with two large-scale real-world datasets show that the proposed method improves recommendation accuracy significantly. Online A/B test results at a display advertising platform also demonstrate the effectiveness of the proposed method in production environments. |
| Researcher Affiliation | Collaboration | Han Zhu1, Daqing Chang1, Ziru Xu1,2 , Pengye Zhang1 1Alibaba Group 2School of Software, Tsinghua University Beijing, China {zhuhan.zh, daqing.cdq, ziru.xzr, pengye.zpy}@alibaba-inc.com XiAng Li, Jie He, Han Li, Jian Xu, Kun Gai Alibaba Group Beijing, China {yushi.lx, jay.hj, lihan.lh, xiyu.xj, jingshi.gk}@alibaba-inc.com |
| Pseudocode | Yes | Algorithm 1: Joint learning framework of the tree index and deep model |
| Open Source Code | No | The paper states: 'Following TDM’s open source work5, we implement all methods in Alibaba’s deep learning platform X-Deep Learning (XDL).' (Footnote 5 points to http://github.com/alibaba/x-deeplearning/tree/master/xdl-algorithm-solution/TDM). This indicates they used an existing open-source platform/TDM's work for implementation, but it does not explicitly state that the specific JTM methodology described in *this* paper is open-sourced or provide a link to its code. |
| Open Datasets | Yes | The offline experiments are conducted with two large-scale real-world datasets: 1) Amazon Books3[20, 9], a user-book review dataset made up of product reviews from Amazon. Here we use its largest subset Books; 2) User Behavior4[34], a subset of Taobao user behavior data. |
| Dataset Splits | Yes | Following TDM [34], we split users into training, validation and testing sets disjointly. Each useritem interaction in training set is a training sample, and the user’s behaviors before the interaction are the corresponding features. For each user in validation and testing set, we take the first half of behaviors along the time line as known features and the latter half as ground truth. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU models, CPU types, memory) used for running its experiments. It only mentions using 'Alibaba’s deep learning platform X-Deep Learning (XDL)' which is a software platform, not hardware specification. |
| Software Dependencies | No | The paper mentions implementing methods in 'Alibaba’s deep learning platform X-Deep Learning (XDL)' but does not provide specific version numbers for this platform or any other software dependencies crucial for reproducibility. |
| Experiment Setup | Yes | We deploy negative sampling for all methods except Item-CF and use the same negative sampling ratio. 100 negative items in Amazon Books and 200 in User Behavior are sampled for each training sample. |