AI-DRIVEN RECOMMENDER SYSTEM FOR WEB STACK SELECTION IN WEB DEVELOPMENT: A COMPREHENSIVE ANALYSIS AND IMPLEMENTATION

Abstract

The rapid evolution of web development technologies has created a complex ecosystem of frameworks, libraries, and tools, making technology stack selection a challenging task, particularly for students and early-career developers. This research aims to address this gap by developing an Al-driven recommender system that assists users in selecting optimal technology stacks for their web development projects. The system leverages a fine-tuned Large Language Model (LLM) and employs a Retrieval-Augmented Generation (RAG) architecture to analyze user-input project requirements and generate tailored recommendations. The methodology involved collecting a structured dataset from diverse sources including research papers, technical blogs, and expert interviews, covering frontend, backend, and database technologies. These were converted into vector embeddings and stored in a Supabase vector database. When a user submits a query, the system retrieves relevant records based on semantic similarity and generates a recommendation using OpenAI's GPT-4 model. A user interface built with Next.js facilitates seamless interaction. The system was evaluated through structured feedback from ten industry professionals, using four real-world project scenarios. Evaluation criteria included relevance, completeness, accuracy, practicality, and clarity. Results indicated strong performance, with high scores in relevance (4.6), clarity and usability (4.475), and accuracy (4.45). Areas such as completeness (4.4) and practicality (4.3) showed room for improvement, particularly in niche domains like IoT. This research demonstrates the potential of LLM-based recommender systems in educational and development contexts. By bridging the gap between academic understanding and industry standards, the system not only supports better decision-making for students but also highlights the broader applicability of AI in streamlining complex design choices in web development.