Abstract
The growing complexity of electricity pricing structures and the lack of standardized billing formats have made it increasingly difficult for consumers to understand their energy costs and compare tariffs effectively. This thesis presents an AI-driven framework designed to enhance transparency and explainability in the electricity market byautomatically extracting, processing, and visualizing data from electricity bills. The system leverages Microsoft Azure AI Document Intelligence to perform text recognition and entity extraction from heterogeneous invoice formats. Identified entities such as base fees, energy prices, network charges, and taxes are structured and stored in
a relational PostgreSQL database. A dedicated backend microservice, implemented in ASP.NET, exposes an API that enables external systems and clients to securely access and query the processed data. Building on this foundation, a web-based frontend developed with Angular provides users with an intuitive interface to explore, interpret, and compare their electricity costs. Through interactive visualizations and explanatory views, the application clarifies how individual cost components contribute to the total amount and highlights potential discrepancies between new-customer and existing-customer tariffs. Furthermore, a builtin tariff comparator allows users to benchmark their current contract against other available offers.
The combined architecture of intelligent document analysis, centralized data storage, and user-centered visualization demonstrates how explainable artificial intelligence can foster greater transparency and consumer empowerment in the energy sector. The evaluation results indicate that the proposed system achieves high data extraction accuracy
and provides a meaningful step toward fairer, more comprehensible energy billing.
| Date of Award | 2026 |
|---|---|
| Original language | English (American) |
| Awarding Institution |
|
| Supervisor | Marc Kurz (Supervisor) |
Studyprogram
- Energy Informatics
Cite this
- Standard