Thematix uses a variety of tools in the modeling and construction of semantic systems. Some of these are proprietary; some are open source. They are all state-of-the-art, and in many cases obviate the need for ‘build versus buy’ decisions — where we can build on the accomplishments of others in this field, we buy.
The choice of tools will typically follow design decisions that we make with you. For instance, an intelligent query system, allowing for “concept” (rather than string) search might use data in databases that is converted statically or on-the-fly into triples. This allows us to use SPARQL queries, which can be expressed in specific terms (e.g. a particular make/model/year) or in more generic terms (light garden tractor) as defined by the ontology. A reasoner associated with the SPARQL query engine determines which specific machines satisfy the terms of the query.
A standard triple store integrated with a reasoner and SWRL rules engine could be used. Ontology construction and testing will commonly be done with tools such as Protégé, Visual Ontology Modeler and RDFGravity. Where transformation of existing data sources into triples for the intelligent query is required, tools such as D2RQ are commonly used in commercial systems.
Java code will be typically be needed to extend the capabilities of the SPARQL queries and SWRL rules. A standard Java IDE such as Eclipse can be used for this. For systems and software design, a standard UML design and analysis tool such as MagicDrawUML will be used.
Some of these tools are described in greater detail below.
Visual Ontology Modeler 2.0
VOM makes using semantic web technologies like Web Ontology Language (“OWL”)-DL accessible to modelers who find the RDF/XML text or Manchester syntax difficult to use. The diagrammatic notation shown below is based on UML’s profile capability and the modeler is delivered as a plug-in to a first class UML modeling tool.
One can easily examine complex relationship patterns by dragging OWL-DL elements onto a diagram, or by using a unique visual editor feature that allows elements related to an existing element to be added to the diagram. In this way, customized diagrams can be created for different points of view of the underlying ontology model.
Diagrammatic representations of OWL-DL simplify understanding and extension of complex ontologies. UML-based representations make semantic web technologies accessible to familiar with UML. They import OWL-DL in a variety of forms using the OWL-API. They export correct-by-construction OWL-DL for use in reasoners and other applications. Finally, the visual ontology browser helps manage complex import relationships between ontology packages.
New ontology elements can be created by dragging and dropping them from the visual editor palette onto the current diagram or by creating new elements in the model browser. Elements can be rearranged for visual clarity without losing connections as a sophisticated connection routing algorithm works to find the best path. In addition, visual connections can be routed manually when desired.
VOM 2.0 has import capabilities for a variety of OWL interchange formats by way of the OWL API. It can export OWL as RDF/XML, allowing it to be used in conjunction with other OWL tools such as the Protégé OWL editor from Stanford.
To inquire about obtaining VOM for your business, please see this page.
Thematix uses and markets RDFace — the RDFa Content Editor. Built by our friends at the University of Leipzig center for Agile Knowledge Engineering and Semantic Web, it is one of the more brilliant tools we have discovered. It is the best enabler for the taxonomy known as SCHEMA.ORG. It works as a simple plug-in to WordPress, providing functionality via the built-in TinyMCE toolbar. It is a simple “what-you-mean-is-what-you-get” tool, allowing the user to “paint” the page with RDFa. It serves to considerably simplify and empower our own RDFa efforts.
RDFaCE employs Sindice, Swoogle and Prefix.cc APIs for resource suggestion (providing appropriate URIs for subjects, properties and namespaces). It also uses — and combines the results of — multiple NLP APIs, further automating text annotation. Currently, RDFaCE supports 7 NLP APIs namely Alchemy, Extractiv, Open Calais, Ontos, Evri, Saplo, Lupedia and DBpedia spotlight.
If you are interested in using RDFaCE for your own site, please feel free to contact us.
Pellet, FaCT+, HermIT Reasoners
Reasoners are software systems that are able to infer logical consequences and deduce implicit knowledge from a set of axioms and asserted facts. Thus, given a set of concepts or classes (“dogs” and “mammals”) and roles or properties (“are members of” “have hair”), a computer can make inferences about objects or individuals (Heidi is a dog; Heidi has hair). The language used to describe concepts, axioms and properties are called “Description Logics” or “DL” for short.
Pellet is an a Java-based DL reasoner created offered in both an open-source and proprietary version by Clark & Parsia. FaCT++ is a DL reasoner supporting OWL DL and (partially) OWL 2, first developed by the University of Manchester. HermIT is an efficient OWL reasoner offered by the University of Oxford.
Stardog is a “fast, lightweight, commercial RDF database for mission-critical apps” that supports SPARQL query; HTTP connections and the SNARL protocol for remote access and control and OWL 2. It is offered in both free and for-pay enterprise versions by Clark & Parsia.
D2RQ enables access to conventional relational databases as RDF graphs, without having to replicate the database into an RDF store. It is an open source system offered by the Digital Enterprise Research Institute (DERI) and the Freie Universität of Berlin, among others.
The Eclipse Platform is a software development environment and integrated development environment supporting a wide variety of programming languages and plug-ins, and supporting Jena — a Java API for ontology management.
Protégé is an open source ontology modeler and knowledgebase framework created and maintained by the Stanford Center for Biomedical Informatics Research. It has a very large and active community of developers and corporate users.
Offered as open source software, the Unstructured Information Management Architecture (“UIMA”) system “analyzes large volumes of unstructured information in order to discover knowledge that is relevant to an end user” — for example, “ingesting plain text and identifying entities, such as persons, places, organizations; or relations, such as works-for or located-at.”
Production Rule Systems
Where fast execution and high-volume runtime environments require, Thematix converts rules expressed in ontologies into logic used buy production rule systems such as those offered by IBM, TIBCO and Fair Issac.