SSR - Semantic Similarity Rating

Could we use LLMs to simulate of potential users to do customer research?

This is a test project for tinkering with the paper LLMs Reproduce Human Purchase Intent via Semantic Similarity Elicitation of Likert Ratings.

If you ask an AI directly to rate something on a scale of 1 to 5, it will give a boring answer like 3. By getting it to explain why it chose a rating, we can get a more nuanced response and convert it into a number. This approach takes the full text response and maps it to a probability distribution across all Likert points.

Quick Start

source venv/bin/activate
pip install -e .
python main.py

Citation

Maier, B. F., Aslak, U., Fiaschi, L., Pappas, K., Wiecki, T. (2025). Measuring Synthetic Consumer Purchase Intent Using Embeddings-Similarity Ratings.