PHD Research proposal in Emotion classification

The advent of Information Communication Technology (ICT) especially, social networking sites enable the users to upload and share their opinions over the internet [1]. In consequence, the massive influx of natural language text in terms of reviews, forum discussions, microblogs, and blogs that have become available online. Several organizations and individuals seek to utilize this natural language information for better decision making due to the fact of human behaviors that heavily reflects their opinion. Therefore, emotion classification or opinion mining is necessary to recognize the emotion of the people from their unstructured natural language reviews [2]. The opinion mining also referred to sentiment analysis attempts to filer the haphazardly scattered emotion from the heterogeneous reviews.
In the context of opinion mining for an organization, there is no need to conduct opinion polls, surveys, and focus groups for capturing the public opinions since there are plenty of such data is publicly available. The opinion mining is an automated data mining process that geared towards revealing and making sense of emotions and moods of the people. The emotion classification imparts beneficial results in commercial applications like product feedback, product reviews, marketing analysis, news articles or political debate, and customer service [3].
The opinion mining continues to evolve and continue to improve across time. Despite it still leaves some challenges in recognition of peoples’ emotions or sentiment analysis [4]. The primary challenge is sarcasm detection in the data that reveals the negative emotion using only the positive text which induces the complexity in gauging the proper sentiment. Another critical bottleneck in the emotion classification is the domain dependency of emotion text, whereas the one feature set yield best results in one domain, on the other hand, it lacks to perform well in some other domain. Moreover, the inadequacy of labeled data poses an obstacle for the advancement of the field.
The diverse application gets to benefit from the emotional classification. However, it does not perform accurately while detecting the sentiments from the free form text of humans. The emotion classification system confronts the complexity while dealing with the scarce or unclear datasets. Furthermore, multi-lingual sentiment analysis, data sparsity, subject detection still leaves the challenges for recognizing the emotions that make the hindrance in the opinion mining task. Some of the research works lack in dictionaries lexicon and corpora. Also, the critical problem in emotion classification is that they lack to identify the spam opinion from the reviews. This fake opinion of the reviews along with the diverse writing style misleads the automated emotion classification system by providing the untruthful positive or negative opinions


  • [1] Wang, Wenbo, Lu Chen, Krishnaprasad Thirunarayan, and Amit P. Sheth, “Harnessing twitter” big data” for automatic emotion identification” In IEEE International Conference on Privacy, Security, Risk and Trust on Social Computing, pp.587-592, 2012.

  • [2] Kołakowska, Agata, Agnieszka Landowska, Mariusz Szwoch, Wioleta Szwoch, and Michal R. Wrobel, “Emotion recognition and its applications”, In Human-Computer Systems Interaction: Backgrounds and Applications, Vol.3, pp.51-62, 2014.

  • [3] Seerat, Bakhtawar, and Farouque Azam, “Opinion mining: Issues and challenges (a survey)”, International Journal of Computer Applications, Vol.49, No.9, 2012.

  • [4] Chaturvedi, Iti, Erik Cambria, Roy E. Welsch, and Francisco Herrera, “Distinguishing between facts and opinions for sentiment analysis: Survey and challenges” Information Fusion, Vol.44, pp.65-77, 2018.

Leave Comment

Your email address will not be published. Required fields are marked *

clear formSubmit