Abstract
Previous researchers have considered sentiment analysis as a document classification task, in which input documents are classified into predefined sentiment classes. Although there are sentences in a document that support important evidences for sentiment analysis and sentences that do not, they have treated the document as a bag of sentences. In other words, they have not considered the importance of each sentence in the document. To effectively determine polarity of a document, each sentence in the document should be dealt with different degrees of importance. To address this problem, we propose a document-level sentence classification model based on deep neural networks, in which the importance degrees of sentences in documents are automatically determined through gate mechanisms. To verify our new sentiment analysis model, we conducted experiments using the sentiment datasets in the four different domains such as movie reviews, hotel reviews, restaurant reviews, and music reviews. In the experiments, the proposed model outperformed previous state-of-the-art models that do not consider importance differences of sentences in a document. The experimental results show that the importance of sentences should be considered in a document-level sentiment classification task.
Highlights
Sentiment analysis is a natural language processing (NLP) task in which a given text is classified into predefined classes
We propose a deep neural networks (DNNs)-based document-level sentiment classification model to automatically reflect the sentence importance meaning on how much each sentence supports polarity of a whole document
We propose propose a DNN-based sentiment classification model in which sentences in a document differently contribute a DNN-based sentiment classification model in which sentences in a document differently contribute to document-level document-level classification classification according according to to their their importance
Summary
Sentiment analysis is a natural language processing (NLP) task in which a given text is classified into predefined classes (e.g., positive, neutral, and negative). The initial models on sentiment analysis use hand-made sentiment lexicons that contain sentiment words annotated with polarities [1,2,3] Based on discrete information, such as polarities and strengths of sentiment words, they classify sentences into sentiment classes with the strongest polarities [2,3,4]. These lexicon-based models are simple and efficient, they suffer from limitations. The manual construction of sentiment lexicons is a time-consuming and labor-intensive job To overcome these limitations, some models to automatically construct sentiment lexicons have been proposed [5,6]. A fixed polarity with strength should be assigned to each sentiment word it may have different polarities depending on application domains
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
