Leveraging AI to Enhance School Selection for Parents

Dec 8, 2023

Introduction

In the ever-evolving landscape of education, parents face the daunting task of selecting the best school for their children. To simplify this process and provide valuable insights, we embarked on a journey to develop a prototype that utilizes the power of Artificial Intelligence (AI). By leveraging Language Model AI (LLM) and the Retrieval-Augmented Generation (RAG) approach, this project aims to empower parents with data-driven recommendations for selecting the most suitable school for their child.

We have open-sourced the code to this project which you can find here.

Project Overview

The project focuses on harnessing education data from authoritative sources such as Ofsted and the Department for Education (DfE) to aid parents in making informed decisions about their child's education. By integrating AI technology into a chatbot interface, parents can effortlessly query school records and receive personalized recommendations based on their preferences and requirements.

Architecture and Technology Stack

The project utilizes the following architecture and technology stack:

  • Language Model AI (LLM): Powered by GPT-3.5, the LLM component handles natural language processing tasks, enabling seamless interaction between users and the chatbot.

  • Vector Database (FAISS): FAISS facilitates efficient similarity search operations, enabling the retrieval of relevant school records based on user queries.

  • User Interface (UI): Gradio, an open-source Python package, is employed to create an intuitive and customizable UI for interacting with the AI model.

  • Infrastructure as Code (IaC): Leveraging AWS CloudFormation templates or AWS CDK, the project ensures scalability and reliability in deploying and managing infrastructure components.

Functionality and Features

The prototype offers the following key features to assist parents in finding the best school for their child:

  1. Interactive Chatbot: The chatbot interface engages parents in conversational interactions, eliciting their preferences and requirements for school selection.

  2. Data Querying: Leveraging education data from authoritative sources, the chatbot queries records on school performance, ratings, and other relevant metrics.

  3. Personalized Recommendations: Based on user input and data analysis, the chatbot generates personalized recommendations for schools that align with the parent's preferences.

  4. AI-Driven Conversations: Through LLM-powered conversational AI, the chatbot provides natural and informative responses to user queries, enhancing the user experience.

  5. Continuous Learning: The chatbot is designed to adapt and improve over time, leveraging machine learning techniques to enhance recommendation accuracy and relevance.

Implementation and Code Snippets

The core functionality of the chatbot is implemented through Python code, utilizing the OpenAI API for LLM interactions and FAISS for vector search operations. Below are excerpts from the codebase illustrating key components:

  • LLM Message Handling: The send_llm_message function sends a message to the LLM model for chat completion, facilitating conversational interactions.

  • Data Handling: The convert_gradio_history_to_text function converts chat history into a text format for data processing and analysis.

  • Query Generation: Functions such as generate_doe_search_query generate search queries based on user input and chat history, enabling efficient data retrieval.

  • Response Generation: The generate_conversational_response function generates conversational responses based on chat history, enhancing the user experience.

Future Directions

As the project evolves, several avenues for enhancement and expansion are envisioned:

  • Integration of Additional Data Sources: Incorporating data from additional sources such as parental reviews and demographic insights to enrich school recommendations.

  • Enhanced User Experience: Continuously refining the chatbot interface and user interactions to provide a seamless and intuitive experience for parents.

  • Machine Learning Enhancements: Leveraging advanced machine learning techniques to improve recommendation accuracy, personalize interactions, and adapt to user feedback.


By harnessing the power of AI, we have developed a prototype that empowers parents in selecting the best school for their child. Through interactive chatbot interactions and data-driven recommendations, parents can make informed decisions about their child's education with confidence. As the project continues to evolve, it holds the promise of transforming the school selection process, ultimately contributing to improved educational outcomes for children.