Introduction
In May 2026, we announced a new strategic technology partnership with metaphacts: the integration of the SAP HANA Cloud Knowledge Graph Engine with metaphactory, an enterprise knowledge graph platform built on open W3C standards. Since that announcement, a key question from customers has been: how do I actually take my existing SAP HANA Cloud data and turn it into a semantic knowledge graph?
Drawing on a detailed walkthrough published by our partners at metaphactory, we'll show you how to model an ontology step by step on the metaphactory platform from an existing SAP HANA Cloud schema, with AI assistance throughout.
The SAP HANA Cloud Knowledge Graph Engine: Your Semantic Foundation
Before we walk through the modeling steps, it is worth briefly grounding the workflow in what the SAP HANA Cloud Knowledge Graph Engine provides.
The Knowledge Graph Engine is a dedicated RDF triple store built into SAP HANA Cloud. It stores graph data as subject-predicate-object triples, the foundational data model of the Semantic Web, and it queries that data using SPARQL — the W3C-standardized query language for RDF graphs. Key capabilities include:
- Native RDF triple store: store and manage knowledge graph data directly inside SAP HANA Cloud, without an external graph database
- Full SPARQL 1.1 support: query and update RDF data using the same SQL-like fluency your teams already use for relational data
- SQL–SPARQL interoperability: federate queries across the knowledge graph, relational, vector, and spatial engines in a single platform — no ETL required
- W3C standards compliance: ontologies expressed in OWL and constraints defined in SHACL are natively supported
- Fully transactional: the Knowledge Graph Engine participates in SAP HANA Cloud's unified transaction model, reading only committed data
- Multiple output formats: SPARQL query results can be returned as JSON, XML, CSV, or TSV, making it straightforward to integrate with downstream systems and AI applications
For SAP HANA Cloud users, this means your knowledge graph data lives right next to your transactional and analytical data. You don't manage a second system; you extend the one you already have.
For full technical details, see the SAP HANA Cloud Knowledge Graph Engine Guide.
Getting Started: Supplier Relationship Management
The walkthrough below is based on a real example from the metaphactory team: modeling a Supplier Relationship Management (SRM) Ontology on top of an SAP HANA Cloud database. The goal is to create a centralized resource that users from across the enterprise can query and retrieve specific supply chain information and data sources they need. This ontology will help people with and without technical expertise, and also AI, to locate data needed to answer supply chain questions. But like many large enterprises, the information they need is spread across disparate documents and files, all organized and formatted differently. Luckily, because data schemas on the SAP HANA Cloud account capture this valuable information; all of this data can be utilized to build an ontology.
Step 1: Connect metaphactory to SAP HANA Cloud & Import Datasource Schema
The first step is to create a datasource in metaphactory that points to your SAP HANA Cloud instance.
In the metaphactory UI, navigate to the Datasources page and create a new datasource. Give it a descriptive name — in our example, “SRM: Supplier Relationship Management” — and an optional description. Then provide the JDBC connection details for your SAP HANA Cloud instance:
- JDBC URL: jdbc:sap://<your-hana-host>:<port>
- Username and Password for your SAP HANA Cloud database user
Once the connection details are entered, use the Test Connection button to verify that metaphactory can reach your SAP HANA Cloud instance. Once you have made sure they are connected, you are ready to click “Select Schemas”.
Image: SAP HANA Cloud datasource creation and connectivity configuration in metaphactory.
With the connection established, the next step is to bring the schema into metaphactory. Click Select Schemas and choose the schema(s) you want to model — in our example, the KGE_USER schema.
Press “Create”, and wait for the diagram to load. You will see the main data schema represented as the node “SRM: Supplier Relationship Management: KGE_USER” with six other connected nodes, including the tables inside the data schema such as “PARTNER_CONTACTS”, “PARTNER_ADDRESSES” and “BUSINESS_PARTNERS”.
Image: Selecting the schema for the SRM Supplier Relationship Management datasource.
Once the import is complete, you will see the schema successfully represented as RDF.
Image: Diagram of the loaded datasource schema in metaphactory on SAP HANA Cloud Knowledge Graph Engine.
Step 1 Demo:
Step 2: Build the Ontology
With the schema imported, navigate to the Ontologies section of metaphactory and create a new ontology. Set the title to “Supplier Relationship Management Ontology” and expand the additional metadata and settings and select the data source just created, “SRM: Supplier Relationship Management”. This binding is important: metaphactory's AI assistant metis will use this grounding to suggest an ontology structure that is directly aligned with your actual data.
Image: Creating the Supplier Relationship Management Ontology.
This is where the workflow starts to feel different from traditional ontology development.
metis is metaphactory's AI semantic modeling assistant, built directly into the metaphactory platform. metis combines knowledge from large language models and knowledge graphs to assist you with ontology building via its conversational interface. Following the LOT methodology, it will start by providing the initial context we have for the initial ontology state, and via back-and-forth conversation, metis will help to suggest the ontology structure best suited to your business case needs based on the imported schema — including classes, relations (object properties), and attributes (data properties).
For the SRM schema, metis reviews the five tables and generates a set of ontology elements. For example, it might propose:
- A Supplier class with attributes for name, identifier, and country of origin
- A Contract class with attributes for start date, end date, and value
- A Product class linked to suppliers via a suppliedBy relation
- A PurchaseOrder class linked to contracts via a coveredBy relation
metis can also propose competency questions, based on ontology metadata, and the datasource which is used in the context of this ontology. An example question generated in our example is “Which suppliers are currently approved for a given product, service, or category?”. If the ontology fails to answer the proposed questions, you can use the conversational interface to make relevant tweaks, or make manual mapping changes yourself.
You review the proposed elements in the metis panel, accept or adjust them, and instruct metis to encode the accepted elements into the ontology. The process is iterative and collaborative: you stay in control of the semantic design while metis does the heavy lifting of translating relational structures into ontology concepts.
Image: Using the metis conversational interface to build and refine an ontology.
Once you are satisfied with the proposed structure, metis encodes the accepted elements. You now have a formal ontology in OWL, ready for the next step.
Step 3: Manual Ontology Building and Refinement
Beyond AI-assisted ontology generation, users can also build and enhance ontologies directly using the Visual Ontology Editor. Following a human-in-the-loop approach is strongly recommended to ensure accuracy and quality throughout the modeling process.
The metaphactory interface is designed to simplify semantic modeling, allowing users with limited technical expertise to create and manage knowledge graphs efficiently. In this example, we can further enrich the ontology by reviewing the suggestions generated by Metis. These recommendations are available both in the Metis panel and directly on the diagram canvas, where they can be accepted, modified, or removed as needed.
For instance, we may decide to add the synonym “Business Partner” to the Supplier class. We can also enhance the class by adding a label attribute with a cardinality of 1.
Any modifications made in metaphactory are instantly synchronized with the connected SAP HANA Cloud Knowledge Graph Engine, ensuring that the underlying semantic model always remains up to date.
Image: Making manual ontology edits in metaphactory on SAP HANA Knowledge Graph Engine.
Step 3 Demo:
Step 4: Data Source Mappings in metaphactory
An ontology alone describes the shape of your knowledge. To populate it with data from SAP HANA Cloud, you need mappings — rules that connect relational tables and columns to ontology classes and properties.
Navigate to the datasource page, open the Datasource Schema, and create a new mapping set named “SRM mappings”. Select Map Data Schema from the interface, choose the Supplier Relationship Management Ontology from the dropdown, and then enable “Map and Generate AI Mappings”.
metaphactory's AI mapping engine analyzes the ontology you built and the relational schema you imported, and automatically generates the mappings between them. Once the generation is complete, you can expand the rows in the data table to explore the generated mappings.
Image: Mappings for SRM mappings in metaphactory on SAP HANA Cloud Knowledge Graph Engine.
Step 4 Demo:
Step 5: Setting up Conversational AI Using metis
In this step, we provide a brief overview of how to configure Conversational AI in metaphactory using the SAP HANA Knowledge Graph Engine.
To create a Search and Discovery Agent, select AI Services from the toolbar and create a template that includes the Conversational AI component. Once the template has been created, configure the agent by navigating to the Agents and Template & Applications Pages sections in the toolbar and associating the template with the desired agent configuration.
Image: Configuring conversational AI on the metaphactory on SAP HANA Cloud Knowledge Graph Engine.
Once you have configured your conversational AI, you can begin asking questions about the ontology and connected data sources. For example, if you ask “Describe the class Supplier”, metis will process the request and return a summary of the class, including its relevant properties, relationships, and associated metadata.
Image: Querying the data schema with the successfully configured conversational AI.
Step 5 Demo:
Step 6: Exploring Loaded Schema, Ontology and Mappings
Now that you have mapped your classes and relationships and refined any corner cases, you are ready to share your ontology with colleagues. They can explore and leverage the ontology through the Conversational AI agent configured in the previous step. The agent utilizes the ontology, the underlying data source, and the mappings you have created to enable users to query source data directly using natural language.
Let's explore some additional capabilities of the conversational AI agent. For example, a member of the compliance and regulatory team may need to identify suppliers that present regulatory or business risks. They could ask the search wizard: “Where do I find risk level and financial rating information for suppliers?“
The agent would analyze the request and guide the user to the relevant SAP HANA Cloud data schema, specifically the SRM (Supplier Relationship Management) schema imported as KGE_USER, as well as the entities and attributes that are relevant to the question. metis also highlights potential ambiguities and provides additional context that may help users better understand the results. In this example, the attributes RISK_LEVEL and FINANCIAL_RATING are mapped to an Assessment entity rather than directly to a mapped Supplier entity. As a result, metis explains the indirect relationship and helps users understand how supplier risk information is represented within the knowledge graph.
Image: Querying the Supplier Relationship Management ontology with metis.
Step 6 Demo:
Demo
For a full end-to-end demo of what we explained today, check out this tutorial by metaphactactory. This video starts with schema load, continues with ontology creation and datasource mappings and finishes with configuring and exploring the conversational AI.
Why This Matters for SAP HANA Cloud Users
The workflow described above represents a fundamental shift in how SAP HANA Cloud customers can approach knowledge graph adoption:
- No blank-slate problem: you do not start from an empty ontology editor. metis bootstraps the semantic model from the schema you already have.
- No data migration: mappings are virtual. Your data stays in SAP HANA Cloud's relational tables; the Knowledge Graph Engine surfaces it as RDF on demand.
- No specialist bottleneck: domain experts can co-build ontologies through metaphactory's visual interface, without needing to write OWL or SPARQL by hand.
- No additional infrastructure: the Knowledge Graph Engine is already part of your SAP HANA Cloud instance. You are extending what you have, not standing up a new system.
- AI-ready from the start: a governed semantic layer built on W3C standards is the ideal foundation for grounding GenAI and LLM applications in trusted enterprise data.
Get Started
If you are an SAP HANA Cloud customer and want to explore this integration, here is where to begin:
- Try out metaphactory's free trial: SAP Store
- SAP HANA Cloud Knowledge Graph Engine Guide: SAP Help Portal
- Original walkthrough by Ademar Crotti (metaphactactory): From relational to semantic: Ontology modeling on SAP HANA® Cloud
- Partnership announcement blog: SAP HANA Cloud Integration with metaphactory – For Knowledge Graph Modelling, Visualization & More
- Watch our recent TechEd 2026 strategy talk – ST108V | Build your custom knowledge layer with SAP HANA Cloud, jointly presented by SAP & metaphacts
Questions, comments, or experiences to share? Drop them in the comments below — we would love to hear how you are using knowledge graphs on SAP HANA Cloud.
