A novel framework based on neural network is proposed to predict sentiment polarity of aspects in a sentence. The proposed model finely tunes pre-trained word embedding in order to get more accurate embedding for aspect-based sentiment analysis task. Aspect attention based on its location and the whole sentence is calculated to generate importance of context words. The merits of our model are tested on three datasets: reviews of restaurants, laptops from SemEval2014, and review collected from twitter for testing robust of our model on language form. It is shown by experiments that the proposed model outperforms the state-of-the-art methods on these datasets.
Paper
Full text
An Aspect Level Sentiment Analysis Based on Embedding Tuning & Attention Neural Network
Semantic Scholar · Computer Science · 2018
Abstract
A novel framework based on neural network is proposed to predict sentiment polarity of aspects in a sentence. The proposed model finely tunes pre-trained word embedding in order to get more accurate embedding for aspect-based sentiment analysis task. Aspect attention based on its location and the whole sentence is calculated to generate importance of context words. The merits of our model are tested on three datasets: reviews of restaurants, laptops from SemEval2014, and review collected from twitter for testing robust of our model on language form. It is shown by experiments that the proposed model outperforms the state-of-the-art methods on these datasets.