Abstract:
HistoriSense is an AI-powered web application designed to transform the preservation and analysis of personal war testimonies. Traditional archival methods often fail to capture the nuanced emotional, geographical, and biographical details within these narratives, leaving valuable historical context and human perspective unexplored. This project addresses that gap by developing a sophisticated system that allows both individual contributors and institutional users, such as museums, to upload and deeply analyze testimony documents.
The application is built using a modern technology stack featuring a React-based frontend for an interactive user interface and a Python/Flask backend that handles core processing logic. Its analytical power comes from a dual Large Language Model (LLM) architecture, integrating OpenAI GPT-3.5 and Google GenerativeAI. This hybrid approach allows the system to perform specialized tasks: identifying emotional trajectories and key themes, extracting geographical locations, and retrieving biographical information about the writer and other individuals mentioned. The system supports distinct user roles, enabling individual users to analyze single testimonies while providing museum organizations with tools for batch processing and comparative analysis of entire collections.
Through an intuitive dashboard, HistoriSense presents the extracted insights via clear visualizations, including emotional spectra, geographical maps, and thematic summaries. This makes complex narrative analysis accessible to users regardless of technical expertise. The project demonstrates a practical and scalable solution for digitizing and understanding historical narratives, offering a powerful tool for historians, researchers, archivists, and families seeking to preserve and comprehend personal accounts of conflict. By automating the extraction of multidimensional insights, HistoriSense bridges the gap between raw personal narrative and structured historical understanding, contributing to both the preservation of cultural memory and the advancement of digital humanities methodologies.
