Yiang Qian

This project compares and optimises BERT and LSTM-pretrained multi-domain sentiment analysis algorithms. We will integrate data sets from multiple sectors to compare the performance of machine learning algorithms, deep learning models, and pre-trained models on sentiment analysis tasks and determine the best multi-domain sentiment analysis model.
For sentiment analysis training and testing, combine datasets from diverse areas. Like SVM, naive bayes, and deep learning sentiment analysis baseline studies (such as CNN, etc.).
Compare BERT and LSTM pre-training models for sentiment analysis on the integrated data set. Optimize the pre-trained model's generalisation and performance in diverse sectors. The experimental data were evaluated, and the advantages and disadvantages of several multi-domain sentiment analysis models and methods were evaluated. Bert was used to achieve unprecedented accuracy in Twitter sentiment analysis, and suggestions were put forward for future research.

To produce a piece of analysis along with an AI demonstration on possible improvements to Diabetic healthcare and/or management, and write a research report on AI capabilities when implemented on diabetic dataset.