View source for Item:OSW71ac62c3f919485f96483ad8dd302dbf


'''Connecting Conventional Software Engineering with Semantic Web Principles'''

{{Template:Viewer/Link|page=Item:OSWaeffcee25ccb5dd8b42a434dc644d62c|url=|label=Dr. Simon Stier}}''',''' {{Template:Viewer/Link|page=Item:OSWbc238d5a23ae58c1a201b5ab499731d6|url=|label=Andreas Räder}}<br>
Fraunhofer Institute for Silicate Research ISC, Würzburg, Germany<br>
<code>simon.stier@isc.fraunhofer.de</code> | ORCID: 0000-0003-0410-3616

==Abstract==
This half-day tutorial at the {{Template:Viewer/Link|page=Item:OSW3926b3fb037448cb8cbcd1ea9041445e|url=|label=}} introduces  '''Object-Oriented Linked Data (OO-LD)''', a practical framework that bridges conventional software engineering with Semantic Web principles by unifying JSON Schema and JSON-LD [https://doi.org/10.5281/zenodo.11401726 (DOI: 10.5281/zenodo.11401726)]. OO-LD enables developers to define syntactical data structures and semantics in a single source without abandoning familiar JSON-based tooling. Through hands-on exercises using '''interactive browser-based playgrounds''' (no installation required), participants learn to create semantic schemas, generate code with object-graph mappings, auto-generate UIs and APIs, import / export RDF, and integrate with LLMs. Real-world examples from materials science and industrial data spaces demonstrate practical applications. Target audience: software engineers, data scientists, and semantic web practitioners (no prior RDF/OWL knowledge required).

==Tutorial Objectives and Relevance==
'''Objectives:''' This tutorial addresses the persistent gap between mainstream software development and Semantic Web technologies. While RDF/OWL provide powerful semantic modeling, their adoption remains limited due to unfamiliarity and tooling gaps. OO-LD bridges this divide by leveraging JSON Schema (widely used in APIs like OpenAPI, OpenAI, general validation, and code generation, e.g. through pydantic dataclasses) and JSON-LD (W3C standard for linking data), creating a unified framework that is simultaneously a valid JSON Schema ''and'' a referenceable JSON-LD context.

'''Relevance to ESWC:''' (1) ''Introduces established semantic technologies'' (JSON-LD, RDF, SPARQL) to conventional developers through familiar JSON tooling; (2) ''demonstrates application in specific domains'' (materials science knowledge graphs, industrial dataspaces, battery production); (3) ''presents integration with emerging AI techniques'' (LLMs structured output, agentic workflows, semantic function calling); (4) ''shares best practices from real-world projects'' managing heterogeneous data in research (NFDI consortia, EU projects) and industry.

==Scope, Learning Goals, and Target Audience==
'''Scope:''' Introduction to OO-LD core principles (dual JSON Schema/JSON-LD interpretation); hands-on schema creation and composition; code generation (Python dataclasses with Object-Graph Mapping capabilities); UI/API generation (forms, OpenAPI specs); RDF import/export and SPARQL querying; LLM integration (structured output, function calling); interoperability with other schema languages (LinkML, AAS, SAMM).

'''Learning Goals:''' Participants will be able to: (1) create OO-LD schemas combining structural and semantic definitions; (2) generate and use Python code with object-graph mapping; (3) validate data using JSON Schema validators and query using SPARQL; (4) auto-generate forms with semantic autocomplete; (5) build FAIR-by-design data pipelines; (6) integrate OO-LD with LLM APIs; (7) assess when OO-LD is appropriate vs. pure RDF/OWL approaches; (8) map between OO-LD and other schema languages.

'''Target Audience:''' Software engineers and data scientists seeking practical semantic modeling approaches; semantic web researchers interested in mainstream developer adoption; industrial practitioners building interoperable data systems; materials scientists and domain experts implementing FAIR data practices. ''Prerequisites:'' basic JSON knowledge; familiarity with python dataclasses is helpful but not required. No prior RDF/OWL experience needed.

==Tutorial Outline (Half-Day: 3.5 hours)==

===Session 1: Foundations (60 min)===
*Introduction: The gap between software engineering and semantic web (10 min)
*JSON Schema and JSON-LD crash course (15 min)
*OO-LD core concepts: dual interpretation, composition, schema referencing (20 min)
*''Hands-on:'' Create first OO-LD schema in [https://oo-ld.github.io/playground-yaml/ interactive playground] (15 min)

'''Break (15 min)'''

===Session 2: Code and UI Generation (60 min)===
*Code generation: Python dataclasses with object-graph mapping (15 min)
*''Hands-on:'' Generate and use Python code from schemas ([https://oo-ld.github.io/playground-python-yaml/ Python playground]) (20 min)
*UI generation: Automatic form creation, range-based autocomplete, multilanguage support (10 min)
*''Hands-on:'' Build a semantic form for data entry (15 min)

'''Break (15 min)'''

===Session 3: Semantic Integration and Advanced Use Cases (60 min)===
*RDF export/import and SPARQL querying (15 min)
*''Hands-on:'' Convert documents to RDF, query with SPARQL (15 min)
*''Hands-on:'' LLM integration: Generate RDF from unstructured content (15 min)
*Real-world case studies: Materials science KGs, battery manufacturing dataspaces (10 min)
*Q&A and closing remarks (5 min)

==Materials Provided==
*'''Interactive Playgrounds''' (browser-based, zero installation): [https://oo-ld.github.io/playground-yaml/ UI & RDF Generation], [https://oo-ld.github.io/playground-python-yaml/ Python Code Generation], [https://oo-ld.github.io/playground-awl/ Semantic Workflows]
*'''Advanced Notebooks''': [https://repolab.github.io/jupyterlite-playground/lab/index.html?fromURL=https://raw.githubusercontent.com/OO-LD/oold-python/refs/heads/main/examples/linked_data_editor.ipynb Linked Data Editor], [https://repolab.github.io/jupyterlite-playground/lab/index.html?fromURL=https://raw.githubusercontent.com/OO-LD/awl-python/refs/heads/main/examples/human_in_the_loop_async.ipynb Human-in-the-Loop Workflows]
*GitHub repositories with code samples: [https://github.com/OO-LD/schema OO-LD Schema], [https://github.com/OO-LD/oold-python oold-python]
*Example datasets from [https://github.com/OpenSemanticWorld-Packages OpenSemanticWorld Package Registry]
*[https://demo.open-semantic-lab.org/ OpenSemanticLab demo instance] for exploring full-stack implementation
*Slide decks and reference documentation

==Presenter Biography==
{{Template:Viewer/Link|page=Item:OSWaeffcee25ccb5dd8b42a434dc644d62c|url=|label=Dr. Simon Stier}} is Head of Digital Transformation and Senior Research Scientist at Fraunhofer ISC (Germany). He holds dual M.Sc. degrees (Computer Science, Functional Materials) and a Ph.D. (summa cum laude) in interdisciplinary materials engineering. As principal investigator of 10+ national and European research projects he leads development of FAIR data infrastructures and semantic technologies for materials science.

He is the lead developer of OO-LD and the OpenSemanticLab platform, combining 10+ years of software engineering experience (DevOps, full-stack development, cloud-native architectures) with semantic web expertise. His practical experience spans industrial automation, IoT sensor integration, and manufacturing data analytics. He serves as Co-Chair of the European Materials Modelling Council's Task Area on Digitalisation & Interoperability and as PI in multiple NFDI (German National Research Data Infrastructure) consortia.

With 10+ workshops/tutorials delivered, 30+ publications and talk on semantic technologies and materials informatics, and extensive experience teaching both researchers and industrial partners, Dr. Stier excels at bridging academic rigor with practical implementation. His teaching philosophy emphasizes "learn-by-building" with autonomous, reproducible workflows and hands-on coding sessions. He has supervised multiple theses and mentored early-career researchers in semantic data modeling and software development.

Technical Expertise: Semantic web technologies (RDF, OWL, JSON-LD, SPARQL); schema languages (JSON Schema, LinkML, SHACL); code generation and Object-Graph Mapping; Python / TypeScript / Java; Docker/Kubernetes; CI/CD pipelines; LLM integration; materials informatics; laboratory automation.

{{Template:Viewer/Link|page=Item:OSWbc238d5a23ae58c1a201b5ab499731d6|url=|label=Andreas Räder}} is a computer scientist and data engineer in Digital Transformation group at the Fraunhofer Institute for Silicate Research ISC, Germany. He holds an MSc in Computer Science (with a focus on cloud computing, databases, and big data) and a BEng in Electrical Engineering and Information Technology. Andreas has contributed as a process engineer in lithography and multi-layer roll-to-roll printing automation, is actively involved in major digitalization initiatives and is a core developer of OpenSemanticLab. His current work is centered on building production grade and scalable infrastructures, graph database interoperability, large-scale data processing, software architectures, material acceleration platforms (MAPs), UI development, as well as LLM pipelines design, and agentic AI workflow orchestration.

==Previous Editions and Dissemination==
'''This is the first formal tutorial on OO-LD at a major conference.''' Components have been presented in various formats:
*'''Community workshops:''' Accelerate 2024 (Vancouver, Canada), E-MRS Spring Meeting 2025 (Strasbourg), NFDI MatWerk and MaterialDigital working groups
*'''Industry training:''' Code labs in contract research projects; developer onboarding sessions for industrial partners
*'''Public resources:''' Interactive playgrounds; GitHub repositories ; published tutorials in OpenSemanticWorld

Tutorial materials will be developed specifically for ESWC, incorporating feedback from these informal sessions and expanding coverage based on ESWC's diverse audience. All materials will be made publicly available under open licenses (CC-BY for slides, MIT/Apache for code).

==Expected Attendance and Participant Attraction Strategy==
'''Expected Attendance:''' 20-35 participants. OO-LD appeals to multiple ESWC constituencies: software engineers seeking practical semantic modeling, researchers building knowledge graphs, industrial practitioners implementing FAIR data systems, and PhD students learning modern data integration approaches.

'''Attraction Strategy:'''
*'''Practical focus:''' Emphasis on production-ready tools and real-world applications rather than theoretical concepts
*'''Zero setup barrier:''' All exercises run in browser-based playgrounds - participants need only a laptop and internet connection
*'''Dual audience approach:''' Content structured for both semantic web newcomers (introduction to JSON-LD, RDF) and experienced practitioners (advanced topics, LLM integration)
*'''Concrete deliverables:''' Participants leave with working schemas, generated code, and deployable workflows
*'''Community engagement:''' Pre-tutorial teaser posts on social media; post-tutorial follow-up via GitHub discussions; integration with ESWC main conference themes
*'''Industry relevance:''' Case studies demonstrating ROI and practical impact in manufacturing, research data management, and dataspaces