LinkedDataHub is next-generation software by AtomGraph that fully exploits semantic technologies to help organizations overcome data silos and API hell
Why Knowledge Graphs?
The web and data technologies have changed our world dramatically, but the solutions are still far from perfect.
The are multiple problems with the current technological landscape:
- Data silos
- The relational model, still used in most IT systems, is inflexible as the schema is mandatory and hard to change. New kinds of NoSQL databases solve some of its problems but increase the number of incompatible systems. As a result, there is no overview over company information as the data is scattered across multiple data models and IT systems.
- API hell
- There are 10,000s of APIs and 100s of file formats. Application integration using APIs is inherently point-to-point, and as such it does not scale as the number of connectors grows exponentially with the growing number of APIs (in the worst case scenario). Additionally, most web applications are not API-driven, therefore developing a frontend web app as well as an API requires double effort.
- Vendor lock-in
- Most NoSQL databases lack standards such as query languages; domain models are baked into source code and cannot be easily reused between systems; most APIs define custom vocabularies and file formats. All of these factors create dependencies on vendor-specific solutions, which incur substantial switching costs for customers.
The solution is to use semantic technologies to build a uniform Knowledge Graph layer above various physical data sources.
- RDF graph
- RDF data model offers unprecedented flexibility by turning siloed data records into a global network of interconnected resources. It makes schema optional and offers no-cost merge operation as well as a standard query language (SPARQL). Since RDF resources use URIs as built-in global identifiers and have machine-readable resource descriptions, RDF Knowledge Graph can be automatically deployed as a web API.
- Data hub
- Instead of point-to-point integration, Knowledge Graph adopts the hub-and-spoke paradigm, where the RDF database serves as the hub with shared identifiers and vocabularies, while the spokes are RDF connectors to databases as well as transformers from various data formats to RDF. Using this approach, integration costs stay linear even with the growing number of data sources.
- Uniform API
- Linked Data Templates technology enables read-write access to RDF Knowledge Graphs based on a single generic API, which ensures API interoperability and keeps implementation costs fixed. Frontend application development does not require a separate effort, as it is simply a UI wrapper that renders the Knowledge Graph API.
This simple yet powerful architecture is non-disruptive and future-proof: it builds on existing data infrastructure and is based on W3C standards. It enables organizations to ask previously impossible questions, get new insights, control data quality across multiple sources, improve SEO and rich search results etc. It also keeps data integration costs linear over the long term.
LinkedDataHub is a Knowledge Graph management system and application platform. After data hub architecture is used to solve data integration, LinkedDataHub can be layered on top of it to provide a state of the art Knowledge Graph user experience as well as a uniform Knowledge Graph API for machine readable access. It comes in both open-source and cloud versions and offers the following features:
- Single sign-on
- LinkedDataHub uses WebID, a universal login and identity protocol for the Web, as the default authentication method. User agents need to sign up only once to authenticate with any application on the platform.
- 0% code
- LinkedDataHub applications are not physical webapps as J2EE ones; they do not need to be implemented in source code and then compiled/interpreted. Like everything within the platform, applications are data resources, created and configured by managing RDF data. A new application can be set up in a matter of minutes.
- 100% configuration
- The structure, appearance, and access rights of an application are completely controlled in the administration application by administration users.
- Applications can use any SPARQL 1.1 compatible datasource. You can switch datasources at any time and and remain in complete control of your data.
- However, LinkedDataHub also provides tighter integration with select triplestores, leading to optimized performance and easier configuration.
- Easy deployment
- Deploy locally or to the cloud using Docker-based architecture.
- Cloud access
- Applications live within the LinkedDataHub multi-tenant cloud environment; no infrastructure is required on the customer's part. If required however, a platform instance can be installed on premises.
- Data import
- LinkedDataHub supports file uploads and import of CSV data, by mapping it to RDF using SPARQL
CONSTRUCTqueries. Import of XML data is planned.
- Data quality control
- Consistency of the application's dataset is ensured by validating all incoming RDF data against SPARQL-based SPIN constraints (built-in as well as user-defined), which are a part of the application's sitemap.
- Access control
- All access to every document on LinkedDataHub is checked against access control rules (built-in as well as user-defined). Unauthorized access is denied.
- Uniform UX
- The user experience is consistent throughout the platform; all documents by default support the same actions.
- Administrative users can use the sitemap to extend an existing application structure, and upload new XSLT and CSS stylesheets to customize the user interface.
- Remote datasources
- An integrated browser component allows exploring remote Knowledge Graph resources without leaving the application. We are constantly improving the software to take advantage of the distributed nature of RDF data.
- HTTP API
- Our next-generation Linked Data Templates technology provides read-write HTTP API for every document, customizable using declarative templates.
- SEO support
- Customizable JSON-LD output allows a mapping of RDF resources to schema.org. That structured data is indexed by Google and other search engines and leads to improved rankings and rich search results.