LexiBot Project Technical Report

Teachers often need to perform routine data-related tasks, such as retrieving student marks, calculating averages, or generating class summaries. However, many lack the programming ability required to interact directly with databases or manipulate structured data. Existing school management systems often require rigid form-based inputs or menu navigation, which can be time-consuming and unintuitive. Although large language models (LLMs) can interpret natural language queries, they are not tailored to specific classroom operations and may produce inconsistent or overly general outputs. Therefore, we developed LexiBot, a Domain-Specific Natural Language for Educational Tasks, which enables teachers to interact with the database using natural language. The system operates on the Object-Oriented Programming (OOP) paradigm to map instructions written in natural language into workable and accurate code, retrieving the data from the database to perform daily analysis tasks. Moreover, we integrated LLM to assist in natural language processing for words or phrases which are not predefined to resolve synonyms or ambiguous verbs.