Welcome to the Bahria University DSpace digital repository. DSpace is a digital service that collects, preserves, and distributes digital material. Repositories are important tools for preserving an organization's legacy; they facilitate digital preservation and scholarly communication.
dc.contributor.author | Dr Shehzad Khalid | |
dc.contributor.author | M.Taimoor Khan | |
dc.contributor.author | Mehr Durrani | |
dc.contributor.author | Kamran H. Khan | |
dc.contributor.author | Armughan ali | |
dc.date.accessioned | 2017-11-23T08:46:05Z | |
dc.date.available | 2017-11-23T08:46:05Z | |
dc.date.issued | 2015 | |
dc.identifier.issn | 1999-4974 | |
dc.identifier.uri | http://hdl.handle.net/123456789/5101 | |
dc.description.abstract | Topic models are successfully used for text analysis to identify product aspects and their associated sentiments. There are various extensions of topic models that focus on specific problems of Aspect-based Sentiment Analysis. The hybrid and semi-supervised models are used to improve on the model accuracy at the cost of some training or expert guidance. Knowledge based topic models are very popular in other research areas having recently applied to Natural Language Processing. The model uses the large volume of data to get intuition from, which is used to improve on the accuracy of the model. Automatic knowledge based models learn in human like manner, having a never ending learning mechanism. The models are evaluated through topic coherence where a better model produces more coherent topics. Performance has been an issue for topic models as the inference techniques require higher number of iterations to converge. With the newly introduced sub-domains of Sentiment analysis i.e. bias analysis, emotion analysis, influence analysis and information leakage etc. the topic models are expected to evolve. | en_US |
dc.language.iso | en | en_US |
dc.publisher | Bahria University Islamabad Campus | en_US |
dc.subject | Department of Computer Engineering CE | en_US |
dc.title | Aspect-Based Sentiment Analysis on a Large-Scale Data: Topic Models are the Preferred Solution | en_US |
dc.type | Article | en_US |