Sentiment Analysis of ChatGPT for Health-related Recommendations

Primary supervisor

Pari Delir Haghighi

Co-supervisors

  • Lim Mei Kuan
  • Dr Yuxin Zhang (Deakin University) Dr Ali Hassani (SNAToolbox)

Traditional recommendation generation approaches mostly need to either collect and use expert rules or train learning models to generate recommendations. Generative AI tools show high potential to facilitate this process as their model has been trained on a large volume of scientific studies. They can generate content in a specific sentiment or mirror one’s lexicon by following instructions from the user. This feature contributes to improving the user experience, particularly when it is used for seeking health-related recommendations. However, there is little understanding of how the sentiment of prompts affect the sentiment of responses.

Student cohort

Single Semester
Double Semester

Aim/outline

This research aims to investigate the relationship between the sentiment polarity of ChatGPT prompts and responses. The findings will contribute to improving prompt engineering on generating responses that are tailored to user’s emotions, thereby enhancing the user experience by offering more personalised and empathetic responses.

URLs/references

D. Q. Wang, L. Y. Feng, J. G. Ye, J. G. Zou, and Y. F. Zheng, " Accelerating the integration of ChatGPT and other large‐scale AI models into biomedical research and healthcare, MedComm–Future Medicine, vol. 2, no. 2, p. e43, 2023.

Required knowledge

 

- Strong skills with NLP methods, particularly with sentiment analysis

- Familiarity with Generative AI and prompt engineering

- Familiarity with correlation tests

Preferred skills: pervious experience with sentiment analysis models like FLAIR, SPACY, NLTK