Target-oriented Opinion Words Extraction with Target-fused Neural Sequence Labeling
- Zhifang Fan, Zhen Wu, Xinyu Dai, Shujian Huang, Jiajun Chen
- Computer ScienceNorth American Chapter of the Association for…
- 1 June 2019
This paper proposes a novel sequence labeling subtask for ABSA named TOWE (Target-oriented Opinion Words Extraction), which aims at extracting the corresponding opinion words for a given opinion target through a target-fused sequence labeling neural network model.
Latent Opinions Transfer Network for Target-Oriented Opinion Words Extraction
- Zhen Wu, Fei Zhao, Xinyu Dai, Shujian Huang, Jiajun Chen
- Computer ScienceAAAI Conference on Artificial Intelligence
- 7 January 2020
This paper designs an effective transformation method to obtain latent opinions, then integrates them into TOWE, which achieves better performance compared to other state-of-the-art methods and significantly outperforms the base model without transferring opinions knowledge.
Improving Review Representations with User Attention and Product Attention for Sentiment Classification
- Zhen Wu, Xinyu Dai, Cunyan Yin, Shujian Huang, Jiajun Chen
- Computer ScienceAAAI Conference on Artificial Intelligence
- 24 January 2018
This paper proposes a novel framework to encode user and product information by applying two individual hierarchical neural networks to generate two representations, with user attention or with product attention, and designs a combined strategy to make full use of the two representations for training and final prediction.
Margin Calibration for Long-Tailed Visual Recognition
- Yidong Wang, Bowen Zhang, Wenxin Hou, Zhen Wu, Jindong Wang, T. Shinozaki
- Computer ScienceArXiv
- 14 December 2021
This paper proposes MARC, a simple yet effective MARgin Calibration function to dynami- cally calibrate the biased margins for unbiased logits in long-tailed visual recognition and studies the relation- ship between the margins and logits (classification scores).
UniDrop: A Simple yet Effective Technique to Improve Transformer without Extra Cost
- Zhen Wu, Lijun Wu, Tie-Yan Liu
- Computer ScienceNorth American Chapter of the Association for…
- 10 April 2021
This paper proposes an approach named UniDrop to unites three different dropout techniques from fine-grain to coarse-grain, i.e., feature dropout, structure drop out, and data dropout into Transformer models, and demonstrates that these three dropouts play different roles from regularization perspectives.
Improving Aspect Identification with Reviews Segmentation
- Tianhao Ning, Zhen Wu, Xinyu Dai, Jiajun Huang, Shujian Huang, Jiajun Chen
- Computer ScienceNatural Language Processing and Chinese Computing
- 26 August 2018
A reviews-segmentation-based method to improve aspect identification by dividing a review into several segments according to the sentence structure, and then automatically transfer aspect labels from the original review to its derived segments.
SPRoBERTa: protein embedding learning with local fragment modeling
- Lijun Wu, Chengcan Yin, Tie-Yan Liu
- Computer ScienceBriefings Bioinform.
- 22 September 2022
This work presents an unsupervised protein tokenizer to learn protein representations with local fragment pattern, and a novel framework for deep pre-training model is introduced to learnprotein embeddings.
Exploiting Unlabeled Data for Target-Oriented Opinion Words Extraction
- Yidong Wang, Hao Wu, Yue Zhang
- Computer ScienceInternational Conference on Computational…
- 17 August 2022
This paper proposes a novel Multi-Grained Consistency Regularization (MGCR) method to make use of unlabeled data and design two filters specifically for TOWE to filter noisy data at different granularity.