Aspect-Based Sentiment Classification with Attentive Neural Turing Machines
Authors: Qianren Mao, Jianxin Li, Senzhang Wang, Yuanning Zhang, Hao Peng, Min He, Lihong Wang
IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate our model on Sem Eval2014 dataset which contains reviews of Laptop and Restaurant domains and Twitter review dataset. Experimental results verify that our model achieves state-of-the-art performance on aspect-based sentiment classification. Experimental results show that our model achieves substantial performance improvement over the two datasets. |
| Researcher Affiliation | Academia | Qianren Mao 1,2, Jianxin Li 1,2, Senzhang Wang 3, Yuanning Zhang 1,2, Hao Peng 1,2, Min He 4 and Lihong Wang 4 1Beijing Advanced Innovation Center for Big Data and Brain Computing, Beihang University, China 2State Key Laboratory of Software Development Environment, Beihang University, China 3Nanjing University of Aeronautics and Astronautics 4National Computer Network Emergency Response Technical Team/Coordination Center of China |
| Pseudocode | No | The paper describes the model architecture and mathematical formulations, but it does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit statement or link indicating that the source code for the described methodology is publicly available. It only references a GitHub link for BERT, which is an external component they used. |
| Open Datasets | Yes | We evaluate our model on Sem Eval2014 dataset which contains reviews of Laptop and Restaurant domains and Twitter review dataset. The first dataset comes from Sem Eval 2014 Task 4, and it contains two kinds of customers reviews from the Laptop and Restaurant domains. The second dataset is the Tweet collection [Dong et al., 2014], as table 1 shows. These are well-established benchmark datasets and a citation is provided for the Twitter dataset. |
| Dataset Splits | No | Table 1 provides statistics for 'Train' and 'Test' datasets (e.g., Laptop-Train, Laptop-Test, Restaurant-Train, Restaurant-Test, Twitter-Train, Twitter-Test), but no explicit 'validation' split sizes or percentages are mentioned for reproducing the experiment. |
| Hardware Specification | No | The paper mentions training models and running experiments but does not provide specific details on the hardware used (e.g., CPU, GPU models, memory specifications, or cloud computing instance types). |
| Software Dependencies | No | The paper mentions using components like 'BERTBase', 'BERTLarge', 'Glove42B', and 'Bi-GRU'. However, it does not specify any version numbers for these software components or other ancillary libraries, which are required for a reproducible description of software dependencies. |
| Experiment Setup | Yes | In our experiments, the dimension of the target and text word vectors are set to 300 in the case of considering the content of each memory slot being set by the word vector of Glove42B [Pennington et al., 2014]. While, these vectors are set to 768-dimension or 1024-dimension when we treat the corresponding sequence output representations from BERTBase or BERTLarge as our external memory respectively. We train our model with the L2-regularization weight of 0.001 and the initial learning rate of 0.01. We also set dropout of 0.5 to avoid over-fitting. |