The research and its outcomes presented in this book, is about lexicon-based sentiment analysis. It uses single-, and multi-word concepts from the SenticNet sentiment lexicon as the source of sentiment information for the purpose of sentiment classification.In 6 chapters the book sheds light on the comparison of sentiment classification accuracy between single-word and multi-word concepts, for which a bespoke sentiment analysis system developed by the author was used.This book will be of interest to students, educators and researchers in the field of Sentic Computing.
1. Introduction1.1 Sentiment in Opinionated Text1.2 Background1.3 Research Problem2. Sentiment Analysis2.1 Sentiment Analysis Challenges2.2 Levels of Analysis2.3 Supervised vs. Unsupervised Sentiment Analysis2.4 Linguistics-based Sentiment Analysis2.5 Lexicon-based Sentiment Analysis2.6 Conclusion3. SenticNet3.1 The Common Sense Nature of SenticNet Knowledge3.2 A Seminal Approach to Concept-based Sentiment Analysis3.3 Producing SenticNet3.4 SenticNet Processes3.5 SenticNet Knowledge: Encoding3.6 SenticNet Access Methods3.7 SenticNet in Numbers3.7.1 Concept Types: Number of Words3.7.2 Analysis of Polarity Values: Single-Word vs. Multi-Word Concepts3.8 Conclusion4. Unsupervised Sentiment Classification4.1 Datasets4.2 Classification Design and Implementation4.2.1 Overview4.2.2 Sentiment Classification Process4.2.3 Polarity Value Thresholds4.2.4 Implementation4.3 Conclusion5. Evaluation5.1 Classification Performance5.1.1 Research Question5.1.2 Quali
tative Differences Between the Datasets5.1.3 SenticNet5.1.4 Sentiment Analysis System5.1.5 Sentiment Classification Design5.2 Limitations5.3 Conclusions6. Conclusion6.1 Future Work6.2 Final RemarksIndex
Lisainfo e-raamatute kohta