Targeted Aspect-Based Sentiment Analysis via Embedding Commonsense Knowledge into an Attentive LSTM
Authors: Yukun Ma, Haiyun Peng, Erik Cambria
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct experiments on two publicly released datasets, which show that the combination of the proposed attention architecture and Sentic LSTM can outperform state-of-the-art methods in targeted aspect sentiment tasks. |
| Researcher Affiliation | Collaboration | Rolls-Royce@NTU Corporate Lab, Nanyang Technological University School of Computer Science and Engineering, Nanyang Technological University |
| Pseudocode | No | The paper provides mathematical equations for the LSTM and attention mechanisms but no structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide a direct link or explicit statement about the release of their own source code for the described methodology. It only provides links to external resources (SenticNet concept parser and Affective Space). |
| Open Datasets | Yes | We evaluate our method on two datasets: Senti Hood (Saeidi et al. 2016) and a subset of Semeval 2015 (Pontiki et al. 2015). |
| Dataset Splits | Yes | The whole dataset is split into train, test, and development set by the authors. Overall, the entire dataset contains 5,215 sentences, with 3,862 sentences containing a single target and 1,353 sentences containing multiple targets. It also shows that there are approximately two third of targets annotated with aspect-based sentiment polarity (train set: 2476 out of 2977; test set:1241 out of 1898; development set: 619 out of 955). Table 2: Senti Hood dataset (Train, Dev, Test columns with counts). To show the generalizability of our methods, we build a subset of the dataset used by Semeval-2015. In total, we have 1,197 targets left in the training set and 542 targets left in the testing set. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU, GPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions using a pre-trained skip-gram model and bi-directional LSTM, but does not provide specific version numbers for software dependencies (e.g., Python, TensorFlow/PyTorch versions, or other libraries). |
| Experiment Setup | Yes | To avoid overfitting, we add a dropout layer with dropout probability of 0.5 after the embedding layer. We stop the training process of our model after 10 epochs and select the model that achieves the best performance on the development set. The word embedding of the input layer is initialized by a pre-trained skip-gram model (Mikolov et al. 2013) with 150 hidden units on a combination of Yelp3 and Amazon review dataset (He and Mc Auley 2016) and 50 hidden units for the bi-directional LSTM. |