Cross-Domain Sentiment Classification Using a Sentiment Sensitive Thesaurus
read more
Citations
A survey on opinion mining and sentiment analysis
A review of affective computing
Sentiment Analysis of Twitter Data :A Survey of Techniques
A survey of sentiment analysis in social media
Sentiment Analysis of Twitter Data: A Survey of Techniques
References
Introduction to Modern Information Retrieval
Foundations of Statistical Natural Language Processing
Text Categorization with Suport Vector Machines: Learning with Many Relevant Features
Opinion Mining and Sentiment Analysis
Mining and summarizing customer reviews
Related Papers (5)
Frequently Asked Questions (10)
Q2. What are the future works mentioned in the paper "Cross-domain sentiment classification using a sentiment sensitive thesaurus" ?
In future, the authors plan to generalize the proposed method to solve other types of domain adaptation tasks.
Q3. How do the authors train a binary classifier to predict positive and negative sentiment in reviews?
Using the extended vectors d′ to represent reviews, the authors train a binary classifier from the source domain labeled reviews to predict positive and negative sentiment in reviews.
Q4. How can the authors create a thesaurus from a large set of reviews?
By using a sparse matrix format and approximate vector similarity computation techniques [21], the authors can efficiently create a thesaurus from a large set of reviews.
Q5. What is the main challenge of the cross-domain sentiment classification problem?
The authors model the cross-domain sentiment classification problem as one of feature expansion, where the authors append2 additional related features to feature vectors that represent source and target domain reviews in order to reduce the mis-match of features between the two domains.
Q6. Why do the authors believe that the lack of performance on books domain is a consequence of the proposed?
Because the proposed method relies upon the availability of unlabeled data for the construction of a sentiment sensitive thesaurus, the authors believe that the lack of performance on books domain is a consequence of this.
Q7. What is the potential for improving the performance of the proposed method?
given that it is much cheaper to obtain unlabeled data for a target domain than labeled data, there is strong potential for improving the performanceof the proposed method in this domain.
Q8. What is the effect of using multiple source domain in the proposed method?
To study the effect of using multiple source domain in the proposed method, the authors select the electronics domain as the target and train a sentiment classifier using all possible combinations of the three source domains books (B), kitchen appliances (K), and DVDs (D).
Q9. What is the way to analyze the features learned by the proposed method?
To analyze the features learned by the proposed method the authors train the proposed method using kitchen, DVDs, and electronics as source domains.
Q10. What is the proposed method for extending feature vectors in a binary classifier?
the proposed method is agnostic to the properties of the classifier and can be used to expand feature vectors for any binary classifier.