Publicação
Developing an automatic constructor of web interfaces for business meta-models
| Resumo: | The growing complexity of modern software systems has intensified the demand for user interfaces that can evolve alongside changes in the underlying domain without requiring manual intervention across multiple layers of the application stack. In domains such as energy management, where business logic and data models are subject to frequent revision, the cost of maintaining frontend applications tightly coupled to backend code becomes a critical bottleneck. This dissertation presents OntologyToUI, an ontology-driven module for automatic user interface generation that uses ontologies specified in OWL/RDF as the single source of truth. OntologyToUI integrates with OntologyToAPI, an existing Python-based framework of the SmartLEM platform that generates REST API routes from ontological domain descriptions, consuming its four modules: BusinessModel, Metadata, Communication, and ExternalCode, which together capture the backend structure of a domain in terms of operations, data fields, communication protocols, and external code references. To these, this work adds OntologyToUI-UserInterface, which defines nine UI element types: Label, Form, Table, Graph (with three chart variants: LinePlot, BarPlot, and CircularPlot), LiveTable, and Image. The module is delivered as two independent artifacts: a generic library that interprets these ontological specifications at runtime and generates user interfaces, and SmartLEM_UI, a declarative application layer that requires no source code, expressing its domain logic entirely through ontology files. At the application level, domain ontologies instantiate these abstractions for a specific system. View ontologies, written in Turtle syntax, compose UI elements by referencing domain models and serve as the sole input to the rendering engine. The rendering pipeline loads view ontologies into an in-memory RDF graph store, extracts UI element definitions through SPARQL queries and direct graph pattern matching, and maps each element to a corresponding React component, rendering them in the order specified by the ontology. The framework is validated through two case studies. The first applies it to the SmartLEM Local Electricity Market platform, where eleven view ontologies covering community management, member and asset operations, simulation, and data visualization are generated without modifying any application source file. The second integrates a real-time IoT dashboard in which smart home sensors publish timestamped readings to an MQTT broker and the browser renders incoming data in a self-updating table via WebSocket, with no REST API layer and no schema declaration required in the view ontology. Both case studies confirm that the framework is domain-agnostic and extensible: a new view or integration paradigm requires only ontology authoring, with no changes to the rendering engine. |
|---|---|
| Autores principais: | Costa, Daniel Theisges da |
| Assunto: | Ontology-Driven development OWL RDF Model-driven engineering User interface generation Domain-specific language Energy communities Local electricity markets IoT MQTT |
| Ano: | 2026 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Politécnica de Bragança |
| Idioma: | inglês |
| Origem: | Biblioteca Digital da UPB |
| Resumo: | The growing complexity of modern software systems has intensified the demand for user interfaces that can evolve alongside changes in the underlying domain without requiring manual intervention across multiple layers of the application stack. In domains such as energy management, where business logic and data models are subject to frequent revision, the cost of maintaining frontend applications tightly coupled to backend code becomes a critical bottleneck. This dissertation presents OntologyToUI, an ontology-driven module for automatic user interface generation that uses ontologies specified in OWL/RDF as the single source of truth. OntologyToUI integrates with OntologyToAPI, an existing Python-based framework of the SmartLEM platform that generates REST API routes from ontological domain descriptions, consuming its four modules: BusinessModel, Metadata, Communication, and ExternalCode, which together capture the backend structure of a domain in terms of operations, data fields, communication protocols, and external code references. To these, this work adds OntologyToUI-UserInterface, which defines nine UI element types: Label, Form, Table, Graph (with three chart variants: LinePlot, BarPlot, and CircularPlot), LiveTable, and Image. The module is delivered as two independent artifacts: a generic library that interprets these ontological specifications at runtime and generates user interfaces, and SmartLEM_UI, a declarative application layer that requires no source code, expressing its domain logic entirely through ontology files. At the application level, domain ontologies instantiate these abstractions for a specific system. View ontologies, written in Turtle syntax, compose UI elements by referencing domain models and serve as the sole input to the rendering engine. The rendering pipeline loads view ontologies into an in-memory RDF graph store, extracts UI element definitions through SPARQL queries and direct graph pattern matching, and maps each element to a corresponding React component, rendering them in the order specified by the ontology. The framework is validated through two case studies. The first applies it to the SmartLEM Local Electricity Market platform, where eleven view ontologies covering community management, member and asset operations, simulation, and data visualization are generated without modifying any application source file. The second integrates a real-time IoT dashboard in which smart home sensors publish timestamped readings to an MQTT broker and the browser renders incoming data in a self-updating table via WebSocket, with no REST API layer and no schema declaration required in the view ontology. Both case studies confirm that the framework is domain-agnostic and extensible: a new view or integration paradigm requires only ontology authoring, with no changes to the rendering engine. |
|---|