Multi-agent restaurant review analyzer using Autogen

An AI-powered agent framework designed to analyze restaurant reviews and derive structured ratings for both food quality and customer service. Built using a multi-agent conversational pattern, the system simulates sequential chats to extract relevant data and compute insightful metrics — making it ideal for review aggregation, chatbot integrations, or personalized recommendation systems.

‍

How It Works

Restaurant Rating analyzer following sequential chats patterns

1. Entry point agent initiate the converstaion, e.g. "May I know how good is "restaurant_name"?
2. Data Fetch agent retreive the restaurant name from the string.Entry point agent then take the restaurant name as a prarmeter and call the registered fetch_restaurant_data method and return a list of all the reviews from the data.
3. The data is then used as a carryover to next chat and as something used by the entrypoint agent to initiate chat in the next converstation.
4. The review nalaysis agent returns all the keywords(awful, horrible, disgusting, bad, unpleasant, offensive, average, uninspiring, forgettable, good, enjoyable, satisfying, awesome, incredible, amazing).
5. The entrypoint agent then call the registerd score_reviews method.
6. Each review has 2 keywords, one as a 'food_score' and the second as customer_service_score" and the review are score based on the following method


Tech Stack

Python
Autogen
Chat-gpt mini
LLM
‍‍