Article: The application of artificial intelligence in sentiment analysis and optimisation of mediation effect in legal disputes Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.1 pp.1 - 19 Abstract: Emotional expressions in legal disputes significantly influence mediation outcomes. To improve emotion recognition and prediction of mediation effects, this paper proposes an AI model integrating emotion analysis with mediation optimisation. Using Chinese BERT as the base encoder, the model employs a semantic-emotional dual-channel structure to capture factual and emotional information separately. An attention mechanism guided by an emotional dictionary enhances sensitivity to implicit emotional cues in legal texts. To address multi-round mediation, an emotion evolution module models the temporal dynamics of emotions. Experiments on a Chinese legal dispute dataset include emotion classification and mediation outcome prediction. Results show the model achieves an F1-score of 0.842 for sentiment classification, a 3.2% improvement over BERT, and 83% accuracy in mediation outcome prediction, with performance variation across dispute types under 2.1%, demonstrating strong stability and generalisation. This work supports intelligent judicial assistance systems and offers a novel approach to emotion modelling in legal texts. Inderscience Publishers - linking academia, business and industry through research

Title: The application of artificial intelligence in sentiment analysis and optimisation of mediation effect in legal disputes

Authors: Jing Zhang

Addresses: School of Chinese and Law, Yantai Institute of Technology, Yantai, 264000, China

Abstract: Emotional expressions in legal disputes significantly influence mediation outcomes. To improve emotion recognition and prediction of mediation effects, this paper proposes an AI model integrating emotion analysis with mediation optimisation. Using Chinese BERT as the base encoder, the model employs a semantic-emotional dual-channel structure to capture factual and emotional information separately. An attention mechanism guided by an emotional dictionary enhances sensitivity to implicit emotional cues in legal texts. To address multi-round mediation, an emotion evolution module models the temporal dynamics of emotions. Experiments on a Chinese legal dispute dataset include emotion classification and mediation outcome prediction. Results show the model achieves an F1-score of 0.842 for sentiment classification, a 3.2% improvement over BERT, and 83% accuracy in mediation outcome prediction, with performance variation across dispute types under 2.1%, demonstrating strong stability and generalisation. This work supports intelligent judicial assistance systems and offers a novel approach to emotion modelling in legal texts.

Keywords: artificial intelligence; legal disputes; sentiment analysis; mediation outcome prediction; attention mechanisms; emotion evolution modelling.

DOI: 10.1504/IJICT.2026.151312

International Journal of Information and Communication Technology, 2026 Vol.27 No.1, pp.1 - 19

Received: 28 Aug 2025
Accepted: 19 Sep 2025
Published online: 22 Jan 2026 *

Title: The application of artificial intelligence in sentiment analysis and optimisation of mediation effect in legal disputes

Keep up-to-date