Authors:
J. Angelin Jeba, S. Rubin Bose, Roja Boina

Addresses:
1Department of Electronics and Communication Engineering, Anna University, Chennai, Tamil Nadu, India. 2Department of Computer Science Engineering, SRM Institute of Science and Technology, Chennai, India. 3Department of Computer Science, University of New Haven, West Haven, North Carolina, United States of America. jebaangelin@gmail.com1, rubinbos@srmist.edu.in2, rboin1@unh.newhaven.edu3 

Abstract:

In recent years, the popularity of predictive models employing machine learning and natural language processing for identifying emotions from various text sources, including news articles, microblogs, and social media posts, has grown significantly. However, deploying such models in real-world sentiment and emotion applications faces challenges, particularly concerning poor out-of-domain generalizability. This difficulty likely arises due to the complexities of transferring multiple emotion identification models, which are influenced by domain-specific characteristics like themes, communicative aims, and annotation techniques. As an online broadcasting platform, microblogging has emerged as a prominent forum for expressing ideas and opinions, prompting researchers from various fields to explore emotion recognition (ER) from microblogs. Automatic emotion recognition from microblogs presents a formidable challenge in machine learning, especially when seeking improved results across different types of content. Emoticons have become increasingly common in microblog materials as they aid in conveying content meaning. This research proposes a method for emotion recognition from microblog data incorporating text and emoticons. Emoticons are considered distinctive ways for users to communicate their feelings, and their meanings can be further enriched by incorporating appropriate emotional phrases. A Multi-Source Information Learning Model is employed to classify emotions, which considers the sequence of emoticons appearing in the microblog data. Experimental findings demonstrate that the suggested emotion recognition algorithm outperforms other methods when tested on Twitter data.

Keywords: Artificial Intelligence; Transfer Learning Algorithm; Convolutional Neural Network (CNN); Recurrent Neural Network (RNN); Receiver Operating Characteristic (ROC) Curve.

Received on: 09/10/2022, Revised on: 27/11/2022, Accepted on: 08/01/2023, Published on: 09/02/2023

FMDB Transactions on Sustainable Computer Letters, 2023 Vol. 1 No. 1, Pages: 12-24

• Views : 757
• Downloads : 13

Download PDF