Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Intention Nets: Psychology-Inspired User Choice Behavior Modeling for Next-Basket Prediction
Authors: Shoujin Wang, Liang Hu, Yan Wang, Quan Z. Sheng, Mehmet Orgun, Longbing Cao6259-6266
AAAI 2020 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on real-world datasets show the superiority of our Psychology-inspired model Int Net over the state-of-the-art approaches. |
| Researcher Affiliation | Academia | 1Department of Computing, Macquarie University 2Advanced Analytics Institute, University of Technology Sydney 3University of Shanghai for Science and Technology EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Model parameter learning procedure |
| Open Source Code | No | No explicit statement or link providing access to the source code for the methodology described in the paper. |
| Open Datasets | Yes | Two real-world online transactional datasets are used for the experiments: (1) Tmall2 released by IJCAI-15 competition, which recorded the purchased baskets from each anonymous user on Tmall platform (The Chinese version of Amazon) in six months. Each basket is associated with a purchase date with no timestamp for each item inside transaction; and (2) Tafeng3 released on Kaggle, which contains the transactional data of a Chinese grocery store generated in four months, whose format is similar to Tmall. Both datasets are commonly used to test the performance of next-basket prediction (Yu et al. 2016; Guidotti et al. 2018). 2https://tianchi.aliyun.com/dataset/data Id=42 3https://www.kaggle.com/chiranjivdas09/ta-feng-grocerydataset |
| Dataset Splits | Yes | Second, we make three training-test splits on the sequence instance set by randomly selecting 20%, 30% and 40% of the instances whose target basket happens in the last 30 days respectively for test while others for training. The number of channels m is set to 3 by tuning on the validation set. |
| Hardware Specification | No | No specific hardware details (e.g., GPU/CPU models, memory) are provided for running the experiments. |
| Software Dependencies | No | The paper mentions 'Our model is implemented using Tensorflow 1.4.' but does not provide any other software dependencies with version numbers. |
| Experiment Setup | Yes | The initial learning rate is empirically set to 0.001 and the batch size is set to 50. In our model, the dimensions of item embeddings and intention states are empirically set to 100. The number of channels m is set to 3 by tuning on the validation set. |