Emotion Detection From Micro-Blogs Using Novel Input Representation

Fahim Anzum

doi:10.1109/ACCESS.2023.3248506

2023-01-01

Fahim Anzum *

Marina L. Gavrilova

DOI：10.1109/ACCESS.2023.3248506

原文链接

原文求助

摘要

En 中文

Emotion is a natural intrinsic state of mind that drives human behavior, social interaction, and decision-making. Due to the rapid expansion in the current era of the Internet, online social media (OSM) platforms have become popular means of expressing opinions and communicating emotions. With the emergence of natural language processing (NLP) techniques powered by artificial intelligence (AI) algorithms, emotion detection (ED) from user-generated OSM data has become a prolific research domain. However, it is challenging to extract meaningful features for identifying discernible patterns from the short, informal, and unstructured texts that are common on micro-blogging platforms like Twitter. In this paper, we introduce a novel representation of features extracted from user-generated Twitter data that can capture users' emotional states. An advanced approach based on Genetic Algorithm (GA) is used to construct the input representation which is composed of stylistic, sentiment, and linguistic features extracted from tweets. A voting ensemble classifier with weights optimized by a GA is introduced to increase the accuracy of emotion detection using the novel feature representation. The proposed classifier is trained and tested on a benchmark Twitter emotion detection dataset where each sample is labeled with either of the six classes: sadness, joy, love, anger, fear, and surprise. The experimental results demonstrate that the proposed approach outperforms the state-of-the-art classical machine learning-based emotion detection techniques, achieving the highest level of precision (96.49%), recall (96.49%), F1-score (96.49%), and accuracy (96.49%).

Keyword:

Feature extraction

Emotion recognition

Social networking (online)

Linguistics

Genetic algorithms

Blogs

Predictive models

Affective computing

Machine learning

Natural language processing

emotion detection