Aspect-Specific Sentimental Word Embedding for Sentiment Analysis of Online Reviews

Abstract

Recently, Deep Convolutional Neural Networks (CNNs) have been widely applied to sentiment analysis of short texts. Naturally, word embedding techniques are used to learn continuous word representations for constructing sentence matrix as input to CNN. As for sentiment analysis of customer reviews, we argue that it is problematic to learn a single representation for a word while ignoring sentiment information and the discussed aspects. In this poster, we propose a novel word embedding model to learn sentimental word embedding given specific aspects by modeling both sentiment and syntactic context under the specific aspects. We apply our method as input to CNN for sentiment analysis in multiple domains. Experiments show that the CNN based on the proposed model can consistently achieve superior performance compared to CNN based on traditional word embedding method.

Sentiment Analysis of Persian Movie Reviews Using Deep Learning

“…The results demonstrated the effectiveness of RNN (achieving up to 95.33%) as compared to a state-of-the-art machine learning algorithm such as SVM (75.22%). Chen et al [ 47 ] proposed a novel framework to improve sentence-level sentiment analysis by employing Long-Short-Term Memory with a conditional random field layer (BiLSTM-CRF). The comparative simulation results with benchmark datasets showed that their proposed framework improved the overall accuracy of sentence-level sentiment analysis.…”

Learning Representations of Natural Language Texts with Generative Adversarial Networks at Document, Sentence, and Aspect Level

“…Finally, aspect-level analysis has been also explored. Du et al [41] modeled both sentiment and syntactic context under specific aspects to acquire better word embeddings, which were given as input to a CNN for sentiment classification of Amazon product reviews. Their results showed an improvement compared to traditional word-embedding methods.…”

“…There are many publicly available datasets that have been used extensively in academia and come from the SA subfield of NLP; the use of such datasets is fueled by an unprecedented flood of social network activity over the last decade and an interest in processing the social media enhanced with sentiment. Among them, the most popular are those datasets provided by the Stanford University, the SST1 and SST2, the Large Movie Review Dataset, the MPQA opinion corpus [37], a dataset of Amazon reviews [41], the ACL Anthology, and the 20 Newsgroups [33]. Below, we give a detailed description of the three benchmarks we used to perform document, sentence and aspect analysis.…”

Method of Feature Reduction in Short Text Classification Based on Feature Clustering

“…However, in the task of short text classification, the improvement of classification accuracy is limited when applying the word embedding model alone, primarily because a short text is concise and it contains many polysemous words, noise words, and internet words. The absence of effective features makes it difficult to extract enough semantic information [10].…”