Project
Robust GraphQL Query Evaluation Pipeline
Objective
GraphQL queries are increasingly sophisticated, enabling efficient data retrieval across diverse, nested data structures. However, validating the correctness of these complex queries—particularly those generated by large language models (LLMs)—is a significant challenge. Traditional formal verification approaches are difficult to apply, as GraphQL queries function as their own specification, without an underlying implementation for direct comparison. This project addresses the growing need to validate LLM-generated GraphQL queries against known ground truth queries. Such validation is essential for assessing the quality of LLM-generated queries and ensuring more reliable outcomes. Typically, GraphQL queries are validated by executing them on sample datasets and checking the accuracy of the returned results. However, this approach cannot guarantee that all possible errors are detected. To overcome these limitations, this project proposes the development of a systematic method for test data generation to model and uncover possible errors, thereby enhancing the reliability of the query evaluation pipeline and making it more robust.
Outcome
1. Prototype tool for automated GraphQL query validation 2. Research paper detailing the methodology for test data generation and query mutation techniques 3. Demo paper showcasing the practical implementation of the validation framework
Apply By Date 30 Sep 2024
Students 4 / 5
Duration 6 months
Mentor Manish Kesarwani
College
All College



Manish Kesarwani' Comment

GraphQL Engine: https://stepzen.com/
GraphQL Basics: https://spec.graphql.org/October2021/