Blockchain-driven peer-to-peer system: Elevating trust in pharmaceutical manufacturer selection through bert-based sentiment analysis

Trustworthy manufacturer selection is essential for ensuring production quality control. Existing methodologies for selecting manufacturers often rely on participant sentiments or opinions, operating within a non-transparent framework. This paper proposes a comprehensive approach to trustworthy manufacturer selection by integrating a peer-to-peer system with a machine learning model. Specifically, it leverages blockchain technology to create a decentralized infrastructure that enhances transparency in the selection process. Furthermore, an intelligent model is introduced, incorporating a manufacturer evaluation module and the Bidirectional Encoder Representations from Transformers (BERT) language model to classify participant sentiments and opinions. To improve the efficiency of sentiment classification, under-sampling techniques are used to balance the datasets. The proposed method is applied within the pharmaceutical industry, where the increasing number of drug manufacturers has heightened the need for reliable manufacturer selection processes. The approach achieved a sentiment classification accuracy of 91% and an F1-score macro-average of 0.89. Comprehensive evaluations confirm the effectiveness of the proposed integrated model, offering a robust solution for trustworthy manufacturer selection.