Introducing Semantics shim Ontology Training shim Applications for Ontology shim About Hypercube Ltd. shim The Financial Industry Business Ontology (FIBO)

Ontology Training

Hypercube Ltd. offers in-person and on-line training courses on conceptual modeling and ontology engineering. We teach a range of techniques to define the formal semantics of business information. Learn how these kinds of models can be put to work in addressing common data modeling problems, reducing integration costs and revealing new insights from your data.

SPECIAL OFFER

For a limited period we are holding the introductory price of $190 for the Concept Ontology Engineering Tutorial until 26 March 2021.

Please go to Introduction to Concept Ontology Engineering (Special Offer) to enroll.

Once enrolled you may choose to attend the in person presentation of this course on Saturday 27 March and 3 April, or you may work through the materials and exercises at your own pace (recordings will be provided after the cut-off date). The course will remain available after these dates and new subscribers are welcome to enrol after 26 March at the higher rate.

ON-LINE TUTORIALS

Conceptual Ontology Engineering Tutorial

Enrol at the above link for 'learn as you go' the on line training course on conceptual ontology modeling. These is based on the real-time training sessions we ran in the fall of 2020, but pre-recorded so that you can learn at your own pace.

Cameo Concept Modeler Training

We are also offering training in the Cameo Concept Modeler (CCM) tool for creation of formal business semantic models. These are available either as real-time on-line training (by arrangement) or as pre-recorded 'learn as you go' modules (under development)

Other On-line Courses

Stay tuned for 'Learn as you Go' manifestations of the FIBO course and other courses in development.

NEW COURSE: GitHub for Numpties - coming soon. Need to use GitHub and not really a developer, or new to the thing? This will be a short, easy pre-recorded training session on the basics of GitHub and how to do branches and forks and things. You should be able to work through this material in about half a day. Mail me (or enrol in the Teachable School and send me a message there) and I'll get this set up.

Hypercube School at Teachable

For details of on-line courses currently available, please see the Hypercube School page on our Teachable site.

Bespoke Training

Bespoke and one on one training is also available - Please download our Prospectus on our 1-, 2- and 3-day courses on Conceptual Ontology Engineering here.

If you are a data architect, are responsible for development, risk or reporting in a data-intensive area (like finance) or if you are someone who wants to explore new opportunities in micro-finance, big data, Blockchain or data visualization, one of these courses is for you.

More details on the Training page.

Back to Top

Introducing Semantics

Data is the vital fluid of any organization, but just how fluid is your data? Does your data contribute to business knowledge? Most firms are familiar with the problem of data being located in different silos and framed in different formats - requiring costly technology initiatives whenever data needs to be used in more than one place, for example for regulatory compliance, risk management and reporting. The real costs however may go deeper, costs measured as failures in corporate governance or in lost business opportunities or unseen risks. Are these hidden costs affecting your bottom line or business flow? Can your business afford these risk exposures?

In the financial crisis of 2008 many large financial institutions were exposed to firms that failed through an intricate network of financial interconnections. There was no shortage of data and yet some firms took weeks to translate that data into actionable knowledge of their exposures. They did not lack data, but the data lacked business meaning when it was needed. Without meaning the data was not actionable as knowledge.

Ten years on, many of the same challenges remain, both in financial institutions and elsewhere.

The key is this: the ultimate management of IT assets does not happen in IT. The business as a whole needs a clear handle on the concepts that are reflected in the organization's data, and the business disciplines to ensure that IT delivers against business needs. There is a range of tools and techniques for this, collectively known as data management maturity. At Hypercube Ltd. we focus on one of those techniques: the creation and maintenance of common reference models for the meanings of things.

A reference model for meaning (or more accurately, for concepts) is known as an 'ontology'. Think of it like a kind of dictionary but written in pure logic rather than words. The development and maintenance of this kind of reference ontology starts with the business, and provides stakeholder oversight of data assets, for example in model driven development, integration, maintenance, reporting and other uses of these IT data assets.

There are other resources also called 'ontologies' that provide more targeted capabilities, and these sometimes use similar formalisms, languages and techniques. These provide valuable new IT application opportunities such as inference processing, enabling the user to gain new insights from existing data. Being another kind of application, these ontology applications are subject to the same disciplines as other IT assets. We are able to help you put in place the development and management processes for these kinds of resource based on our unique semantics-based methodology.

Back to Top

Applications for Ontology

Integration

We have a novel approach to solving integration problems, which we call an 'ontological' approach.

The key to integration across multiple sources of data is to have a single formal representation of the things in the world that that data deals with. This is the essence of an ontology. The kind of ontology we develop provides a single, computationally independent representation of business subject matter that can be processed by people and machines alike. This common point of reference is called a 'reference ontology' or 'concept ontology'.

The use of this kind of ontology enables a firm to progress beyond point to point mappings between data sources, to a hub and spoke mapping from each set of data elements to the formal, logic-based representation of the concepts ('meanings') of those terms. These mappings are not only more efficient but also provide an audit trail for the mapping activity, by referencing the intended meanings of each source data element.

The use of concept-based ontologies goes beyond a bottom-up integration of data terms, by using a range of established ontological techniques for distinguishing between otherwise similarly named terms. By way of example, some firms struggle with integrating data about people and organizations, clients, prospects and other parties that they deal with. The ontological approach separates things by context and then aligns the data across different data resources against a single, simple model of the truth.

Risk, Compliance and Reporting

Ontology provides a representation of things in the business that are understood the same for machines and for people. One of the most important applications of the human-facing side of ontology is in business reporting. This may be internal management reports or external reporting mandated by laws and regulations.

Regulatory reporting is a massive but seemingly unavoidable overhead in IT spend, imposed by regulators. Each new regulation typically spawns a whole new IT project to provide the required reports from data in existing data stores.

However, any such overhead is really mainly imposed by the inability of IT to provide a single, unified view of the truth. Some regulations such as (in finance) the Basel regulation BCBS239, actively help the organization by pushing reporting entities to maintain a common taxonomy of business concepts and a common reporting data architecture, among other things. If you know how to implement this, it actively saves money on existing reporting.

The secret is this: with a common view of the meaning of concepts across the full range of existing data sources, reporting becomes a matter of tagging the various data elements with the intended meanings in the ontology and then deriving reports and management views with reference to this ontology.

This does not even require a big up-front spend. Some semantic technology evangelists make it sound like you must get rid of all your existing data structures and move everything to a fancy new thing called a 'triple store'. Nothing could be further from the truth. There are times when you might want to copy data into a triple store in order to derive new insights from that data using inference processing applications, particularly for risk management and compliance. For the rest, it is quite easy to interrogate existing data sources using semantic queries, via standard adaptors. All that is required is some up-front mapping of those data sources to their intended business meanings. The results of the semantic queries can be presented in report documents and in management dashboards in a wide range of easily digestible visualization formats.

These semantic queries use a standard semantic query language from the W3C, called SPARQL (pronounced 'sparkle'). Technologists can easily be taught how to use this querying language and there are visual tools to assist in the editing of these by business analysts and others. Meanwhile the data that is reported in remains in the systems of record where it is maintained under existing data lineage and data quality arrangements.

Inference Processing Applications

Ontologies enable a new class of application called inference processing. These enable an organisation to draw novel inferences from existing data using a process called 'reasoning'. These ontology-enabled applications reason over their data to throw up hidden insights or to automatically classify subject matter such as classes of financial transaction (swaps etc.). These applications are particularly valuable in risk management and compliance, as well as in marketing and customer relationship management (CRM).

To take advantage of these capabilities the organisation needs to move or copy the relevant data to a new kind of data structure called a 'triple store'. The ontology provides a kind of data scheme for the data in the triple store, linking data elements to their business meanings. The ontology is represented in a format called the Web Ontology Language (OWL), a W3C standard. Data is represented in the Resource Definition Framework (RDF), also known as 'Linked Data'.

The kind of ontology needed for this class of applications is subtly different to those described above for integration and reporting. These ontologies are optimised for efficiency in the reasoner and do not require or benefit from the conceptual abstractions used in reference (conceptual) ontologies.

The Financial Industry Business Ontology (FIBO) provides a set of fully optimised ontologies that can be used 'out of the box' for financial industry inference processing applications. This is available in a range of OWL formats from spec.edmcouncil.org/fibo.

Triple stores and RDF/OWL ontologies also enable semantic querying of the content of a triple store using the W3C standard querying language SPARQL. This enables the creation of reports and management dashboards from the triple store data. As noted above, these techniques may also be applied against a more conceptually grounded representation of the knowledge of the business as a whole using conceptual or reference ontologies.

Distributed Ledger Technology (Blockchain)

Blockchain is a term that now refers to a whole number of new enablers in an area called distributed ledgers. The key things to know here are that a distributed ledger is not a ledger, a smart contract is not a contract (and not necessarily very smart) and that the things called oracles are not from Oracle. But most importantly, these crypto-based ecosystems are not, or not necessarily, about cryptocurrency. Wherever there is a benefit in providing anonymous, non-repudiable records of things, there is something ripe for industry disruptions. Whether it's records of real estate or other assets, cyber assets, if you can think of a reason to provide records that can't be disputed later, you can probably identify a new business opportunity in this fast-changing space.

How semantics can help

The things we call Smart Contracts are not contracts but one of the many things they can do is what contracts do. Given a well-grounded semantic model, any contract-like thing can be defined in terms of combinations of formal commitments, one of the core building blocks of our reusable industry semantic model or ontology. If you can formally define the commitments and undertakings made between anyone, there is now a way to generate smart contracts from these.

More generally, at present there is a raft of different networks or blockchains (and non chain-like crypto things, like the Tangle from IOTA) and no unified standard for setting out the meanings of the concepts that the data in these things will represent. This is where our unified business semantic model (or ontology) fits in. With this ontology you can create applications, networks and ecosystems that can operate across one blockchain system or several, with common business meaning for interoperability.

The Internet of Things

The Internet of Things is what it says. It takes the arrangements by which people have been able to interact with each other and share information, and extends this to things. These might be everyday things – is your toaster listening to you? It might be Smart Cities – taking the sensors and information that exists around a town and bringing it together to do new interesting things. Then there's factories, refineries, oil and gas - all those pressure and level sensors, parts per million measurements and so on. Instruments that fall into a relatively small number of arrangements as data but whose meaning (the implication of what the data says) represents a rich range of things in the world.

That's where ontology comes in. For the Internet of Things we start by using basic ontological techniques to distinguish between the measure and what is measured. This means that both of these matters can be surfaced in the realm of data: letting you report on the what and the how much, and the how much of what.

This can lead on into new opportunities for what you can do with sensor data, such as drawing inferences from the measurements that you have. This may be applied in smart factories for just-in-time and automation, or in safety critical systems like fire and gas or emergency shutdown (ESD). Smart production an smart monitoring widens the range of opportunities for what you can do with the same data; data you always had, but did not necessarily know what you knew.

And those Smart Cities? Things are only as smart as the brain behind them. And a brain only knows what it knows. Knowledge is information with meaning, and meaning requires a formal ontology that goes beyond mere data representation to a formal representation of the entire ecosystem. The things. Of which the Internet of Things is an internet.

Machine Learning, Natural Language Processing and Artificial Intelligence

This class of applications covers a range of uses from artificial intelligence to machine-assisted extraction of text from unstructured material. These are particularly useful in analysing legal and regulatory texts for conformance and other requirements.

Traditionally AI applications have been based either on statistics or semantics. The use of semantics is particularly relevant in natural language processing and text extraction, where similar words may have widely different meanings. Even to know that the word 'bank' refers to a financial institution rather than the edge of a river or the movement of an aircraft to one side, is not something that can be inferred simply from the presence of the four letters b a n k in some text. Natural language processing applications use techniques to analyse parts of speech in order to distinguish nouns from verbs ('a bank' versus 'to bank') but the rest requires a more complete semantic representation of the underlying concepts.

The use of ontology for this kind of application is clear, and there are several natural language processing applications on the market now that can consume formal ontologies in RDF or OWL formats. There are others that have their own bespoke formats, which can be derived from formal OWL-based ontologies. A more interesting question is what kind of ontology is best suited to this class of application. Unlike inference processing applications, the content that natural language processing deals with has no relation to any existing underlying data but refers exclusively to real things in the world. Clearly this requires an ontology of these real things (a reference ontology), not a data-focused application ontology. However, the output of these applications also needs to be placed into some formal data store.

We can help develop the kind of ontology that reflect the real things in the target problem domain, as well as provide guidance on how to generate usable data from this in a linked data or triple store environment. Our methodology distinguishes these different kinds of ontology and helps you implement the right kind of ontology for each class of application, and to integrate these with other applications and data sources.

Back to Top

About Hypercube Ltd.

Hypercube works in the financial sector to deliver business domain modelling, data architecture, mapping and systems integration services. Our focus is developing an understanding of the business meanings of data elements and messaging terms across the organisation.

We deliver this using the latest Semantic Web technology, including ontology, taxonomy modelling, business process and choreography. These models make up the formal requirements for development of message and database technology delivery.

How we deliver results

We have developed a business-readable format for capturing and displaying business meanings for data terms. The output is presented on spreadsheets and simple block diagrams, with no technical formatting or arcane graphical language. This provides an easily readable graphical summary for business domain experts to review and validate the content of data models. The end result is a formal "Requirements management" framework for data terms and relationships.

This is structured in line with the semantic web language for capturing business semantics (known as OWL) so that the content can be used in development and design of data structures and messaging.

Benefits to our clients

By understanding the meaning of data terms across the organisation clients can integrate systems, re-use data and communicate more effectively, providing the best return on the value of your data assets.

We believe our approach is the key to enterprise data management. It allows optimal use of data resources, effective and future-proof systems integration and transparent end-to-end communication across the entire supply chain.

Back to Top

The Financial Industry Business Ontology (FIBO)

In April 2008, the Enterprise Data Management Council (EDMC) appointed us to develop a global Financial Industry Business Ontology (FIBO). Mike Bennett was engaged full-time with the EDM Council until mid 2018 to complete this industry resource, and continues to participate in the FIBO development teams.

As of September 2017 this work is published directly by the EDM Council at the EDM Council FIBO Specification site.

Full background and details of the most current OMG FIBO specification can be found at the OMG Financial Standards Page.

Back to Top


The Semantic Shed

Hypercube Ltd is a member of The Shed Group, the consulting arm of the Semantic Shed community of practice.

Follow @MikeHypercube
The trademarks Unified Modeling LanguageTM and UMLTM used on this site are trademarks of the Object Management Group, Inc.
Prepared by: Mike Bennett
Hypercube Limited
Registered in England number 2461068
Registered Address; Suite 115, Devonshire House, Manor Way, Borehamwood, Hertfordshire WD6 1QQ
Site last updated 14 October 2018