Abstract:Sentiment analysis (SA) is a technique that lets people in different fields such as business, economy, research, government, and politics to know about people's opinions, which greatly affects the process of decision-making. SA techniques are classified into: lexicon-based techniques, machine learning techniques, and a hybrid between both approaches. Each approach has its limitations and drawbacks, the machine learning approach depends on manual feature extraction, lexicon-based approach relies on sentiment lexicons that are usually unscalable, unreliable, and manually annotated by human experts. Nowadays, word-embedding techniques have been commonly used in SA classification. Currently, Word2Vec and GloVe are some of the most accurate and usable word embedding techniques, which can transform words into meaningful semantic vectors. However, these techniques ignore sentiment information of texts and require a huge corpus of texts for training and generating accurate vectors, which are used as inputs of deep learning models. In this paper, we propose an enhanced ensemble classifier framework. Our framework is based on our previously published lexicon-based method, bag-of-words, and pre-trained word embedding, first the sentence is preprocessed by removing stop-words, POS tagging, stemming and lemmatization, shortening exaggerated word. Second, the processed sentence is passed to three modules, our previous lexicon-based method (Sum Votes), bag-of-words module and semantic module (Word2Vec and Glove) and produced feature vectors. Finally, the previous features vectors are fed into 11 different classifiers. The proposed framework is tested and evaluated over four datasets with five different lexicons, the experiment results show that our proposed model outperforms the previous lexicon based and the machine learning methods individually.

Sentiment-Aware Word Embedding for Emotion Classification

Exploring Spatio-Temporal Representations by Integrating Attention-based Bidirectional-LSTM-RNNs and FCNs for Speech Emotion Recognition

Deep Spectrum Feature Representations for Speech Emotion Recognition

Sentiment Embeddings with Applications to Sentiment Analysis

Learning Sentiment-Inherent Word Embedding for Word-Level and Sentence-Level Sentiment Analysis

Word Embedding Composition for Data Imbalances in Sentiment and Emotion Classification

A Hybrid Approach for Word Emotion Recognition

Sentiment Lexicon Enhanced Neural Sentiment Classification

Refined Global Word Embeddings Based on Sentiment Concept for Sentiment Analysis

Learning Sentiment-Specific Word Embedding for Twitter Sentiment Classification.

Encoding Sentiment Information into Word Vectors for Sentiment Analysis.

Attention-Based BiLSTM Network with Lexical Feature for Emotion Classification

WEMOTE - Word Embedding Based Minority Oversampling Technique for Imbalanced Emotion and Sentiment Classification

Disentangling Latent Emotions of Word Embeddings on Complex Emotional Narratives

Inferring Affective Meanings of Words from Word Embedding

Dimensional Sentiment Analysis of Traditional Chinese Words Using Pre-Trained Not-Quite-Right Sentiment Word Vectors and Supervised Ensemble Models

An Empirical Study on Sentiment Classification of Chinese Review using Word Embedding

Opinion Words Extraction and Sentiment Classification with Character Based Embedding

Improving Twitter Sentiment Classification via Multi-Level Sentiment-Enriched Word Embeddings

A Regression Approach to Valence-Arousal Ratings of Words from Word Embedding.

An Enhanced Sentiment Analysis Framework Based on Pre-Trained Word Embedding