Multi-Agent System for Semantic Extraction and Management of University Activities

This work is part of a novel agent-based framework designed to analyze and process semi-structured information from web pages in order to generate semantic annotations. The primary objective is to develop a design approach that enables the automated generation of RDF annotations by extracting semi-structured data from web content.The study focuses on a real-world use case: a web platform dedicated to the management, capitalization of academic activities, and traceability of students and staff at the Faculty of Legal, Economic and Social Sciences of Ain Sebaâ (FSJES), Hassan II University of Casablanca (UH2C). The proposed solution leverages XML, XHTML, XSLT, and RDF technologies, and is guided by an ontology formalized in RDFS within the framework of the European CoMMA project (Corporate Memory Management through Agents). The extraction process is driven by a society of cooperating agents as the core methodological approach.The experimental results were encouraging and aligned well with the research goals. Specifically, the extractors successfully annotated the target web platform using a template derived from a representative page. Furthermore, the integration of learning techniques enhanced the generation and updating of extraction templates. Notably, the adoption of relative XPath expressions for locating data elements contributed to improved robustness and adaptability of the extraction process.