Why Cognitive Systems should combine Machine Learning with Semantic Technologies

Why Cognitive Systems should combine Machine Learning with Semantic Technologies

Will Artificial Intelligence make subject matter experts obsolete?

Imagine you want to build an application that helps to identify wine and cheese pairings. Who will perform best? Applications solely based on machine learning, those ones which are based on experts’ knowledge only, or a combination of both?

Most of the machine learning algorithms were developed to solve a well-known problem in AI, which is called the ‘Knowledge Acquisition Bottleneck’. It deals with the question how subject matter experts (SMEs) can be enabled to work together with data scientists on knowledge models in an efficient and sustainable way (See also: Taxonomies and Ontologies – The Yin and Yang of Knowledge Modelling).
Machine learning algorithms learn from data, and by that, successful implementations are obviously strongly related to data quality and the approaches taken to encode the semantics (meaning) of data. Semantic Knowledge Graphs help to increase data quality substantially. They also kick-start your machine learning project. A recently published article by Yanko Ivanov puts it in a nutshell: “A machine learning algorithm is a toddler who first needs to learn the basics of your language.”

In recent months, we have observed a trend on the market: Various organizations have implemented their first versions of applications based on machine learning. In a second iteration they are now looking for technologies and methodologies that help them to tackle these three issues:

  • Machine learning algorithms often do not get enough signals to ‘understand’ the correct meaning of the data, e.g. to dissolve ambiguities. Precision lags behind expectations.
  • Cognitive platforms often need sensitive data to learn from it, which should not be processed in the cloud.
  • Extensive experiences and knowledge of SMEs cannot be encoded and will be wasted when not included in the algorithms

AI is more than a technology

Facing the ‘Knowledge Acquisition Bottleneck’ also means that experts’ knowledge is recognized as an essential asset of any organization. This golden treasure shouldn’t be transferred into a cloud to be processed by some machines out of our control. Instead, implementing a good mix of various technologies and methodologies including SMEs efficiently is key. A good AI strategy is not just about better results being generated immediately, it’s a question of how we want to establish a working cooperation between people and machines.

A recently published IDC White Paper discusses socio-technical aspects of Artificial Intelligence and gets to the heart of the issue: “Embracing semantic technologies to deliver cognitive solutions can enable an organization to substantially reduce its dependency on developers and specialized IT professionals. Once the foundation of semantic data management is in place, the adoption of data-driven applications will be driven by domain experts and business users.”.

Why Cognitive Systems should combine Machine Learning with Semantic Technologies

A Standard to build Knowledge Graphs: 12 Facts about SKOS

A Standard to build Knowledge Graphs: 12 Facts about SKOS

These days, many organisations have begun to develop their own knowledge graphs. One reason might be to build a solid basis for various machine learning and cognitive computing efforts. For many of those, it remains still unclear where to start. SKOS offers a simple way to start and opens many doors to extend a knowledge graph over time.

1) Standardised: The Simple Knowledge Organization System (SKOS) is a standards-based ontology, which was published by the World Wide Web Consortium (W3C) in 2009.

2) Future-proof: SKOS is part of a larger set of open standards, which is also known as the Semantic Web. The usage of open standards for data and knowledge models eliminates proprietary vendor lock-in

3) Wide range of applications: SKOS is primarily used to build and to make controlled vocabularies like taxonomies, thesauri or business vocabularies available as a service. This builds the basis for a wide range of applications, starting from semantic search and text mining, ranging to data integration and data analytics. 

4) Graph-based: SKOS concepts can be related and linked to other concepts and instances of ontologies. By these means, SKOS constitutes the nucleus of a decentralised enterprise-wide knowledge graph. 

5) Taxonomy/ontology overlay: Without violating restrictions, any node in a knowledge graph can be part of the taxonomical (SKOS) and ontological structure (e.g. FIBO, FOAF, or schema.org) at the same time

6) Cost-efficient and incremental approach: Any SKOS-based taxonomy, thesaurus, or controlled vocabulary can be extended and enriched by additional ontologies step-by-step, thus various views on the same node can be created when needed. SKOS-based vocabularies can be used as a starting point for a cost-efficient development of more extensive semantic knowledge graphs. 

7) Actionable content: SKOS models, no matter if linked to more expressive ontologies or not, can be queried with SPARQL, or validated by the use of SHACL, a recently issued standard for describing and validating RDF graphs. By that means, knowledge models become actionable and can help to find answers in unstructured content, trigger alerts or to make better decisions. 

8) Things, not strings: With SKOS, solely term-based taxonomies get obsolete. Taxonomical knowledge becomes accessible, its semantics becomes explicit. Any business object represented in a knowledge graph receives a unique address and can then easily be integrated in a bot, service, or application. Terms and strings are no longer used to build ambiguous metadata, instead, a semantic layer on top of all content and data assets works like a multi-dimensional index. 

9) Widely adopted: Hundreds of SKOS vocabularies are available on the web. Large international bodies like the EU, UN, or The World Bank make use of SKOS to make their knowledge available to external and internal stakeholders as well. Many Fortune 500 companies have already adopted SKOS for internal use. 

10) Mandatory: SKOS, RDF and other standards can, for instance, be required in EU public procurement.

11) No black box: Organisations seeking for strategies to keep control over their cognitive applications and algorithms need to involve their own subject matter experts. SKOS is relatively easy to learn and can produce massive input to make machine learning tasks more precise. 

12) Easy to master

A Standard to build Knowledge Graphs: 12 Facts about SKOS