Neural Network Methods for Targeted Sentiment Extraction
Targeted sentiment analysis, which has attracted increasing research attention, classifies the sentiment polarity towards each target entity mention in given text documents. Seminal methods extract manual discrete features from automatic syntactic parse trees in order to capture semantic information of the enclosing sentence with respect to a target entity mention. We claim that competitive accuracies can be achieved without using syntactic parsers, which can be highly inaccurate on noisy text such as tweets. This is achieved by applying pattern-based approach on distributed word representations with rich neural pooling functions over a simple and intuitive segmentation of tweets according to target entity mentions. Furthermore, we extend the idea by proposing a sentence-level neural model to address the limitation of pooling functions, which do not explicitly model tweet-level semantics. First, a bi-directional gated neural network is used to connect the words in a tweet so that pooling functions can be applied over the hidden layer instead of words for better representing the target and its contexts. Second, a three-way gated neural net- work structure is used to model the interaction between the target mention and its surrounding contexts. Experiments show that our proposed model gives significantly higher accuracies compared to the current best method for targeted sentiment analysis.
Duy Tin Vo is currently a PhD candidate at Singapore University of Technology and Design (SUTD), under the supervision of Prof Yue Zhang. Before joining SUTD, he worked as a lecturer at Cantho University. Duy Tin Vo received his undergraduate degree on Electronics and Telecommunication Engineering from Cantho University, Vietnam. His research interests include natural language processing, machine learning and artificial intelligence. He has been working on applying machine learning and deep learning techniques to sentiment analysis and text classification.