Proceedings of the 1st International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security, IACIDS 2023, 23-25 November 2023, Lavasa, Pune, India

Research Article

Aspect-Based Sentimental Analysis On Social Media Data Using Deep Learning Methods

Download60 downloads
  • @INPROCEEDINGS{10.4108/eai.23-11-2023.2343146,
        author={Samson Ebenezar Uthirapathy and Domnic  Sandanam},
        title={Aspect-Based Sentimental Analysis On Social Media Data Using Deep Learning Methods},
        proceedings={Proceedings of the 1st International Conference on Artificial Intelligence, Communication, IoT, Data Engineering and Security, IACIDS 2023, 23-25 November 2023, Lavasa, Pune, India},
        publisher={EAI},
        proceedings_a={IACIDS},
        year={2024},
        month={3},
        keywords={lstm rnn aspect based sentiment analysis machine learning},
        doi={10.4108/eai.23-11-2023.2343146}
    }
    
  • Samson Ebenezar Uthirapathy
    Domnic Sandanam
    Year: 2024
    Aspect-Based Sentimental Analysis On Social Media Data Using Deep Learning Methods
    IACIDS
    EAI
    DOI: 10.4108/eai.23-11-2023.2343146
Samson Ebenezar Uthirapathy1,*, Domnic Sandanam2
  • 1: Saveetha School of Engineering, Saveetha Institute of Medical And Technical Sciences, Chennai, India
  • 2: Department of CA, National Institute of Technology, Tiruchirappalli, India
*Contact email: u.samson@gmail.com

Abstract

Nowadays, reviews related to the products given by users and experts in social media are significant references to help product manufacturers make better decisions about prompting items and placing them in the real world. So analysing aspects such as product-related sentiments is an important task. The target of aspect-based opinion analysis is to determine the emotional valence of each aspect phrase included inside a given sentence. However, the earlier models make the error of identifying irrelevant contextual terms as cues for determining aspect sentiment. They also overlook syntactical limitations and long-range sentiment dependencies. Hence, we assumed a model using LSTM with GCN to distinguish engagers' review opinions under various aspects with three different labels: negative, neutral, and positive. The proposed model LSTM+GCN shows better performance than other models in terms of 4% with precision, 4% with recall, and 5% with accuracy.