Semantics, the language of business
Semantics is the logic of meaning and how referential relationships inform that. Concepts using natural language are the basis of the semantic model, an ontology which sets out this logic and ensures consistency of meaning.
Metadata links people and systems using both human concepts & the machine terminology of applications. Semantic tags are metadata applied to content. Entities are described by both human and machine-generated metadata, entities important to the user:
- words which reflect how the user might search for such content - exactly as we do on the web (the hierarchy of the ontology should reflect this rather than, for instance, a KPI which would be an attribute, not an entity);
- application metadata i.e. structured data such as table or field names, time stamps etc., which is taken directly from systems;
You may need more than one ontology - for different user audiences (e.g. in different countries, using different languages), In this scenario, entities in common (nodes) become mutual touchpoints, allowing attributes to be localised, keeping the ontology as lean as possible by excluding irrelevant items.
The meaning of content can be encoded in terms that people and machines understand - semantic tags. This means people can use the power of machines - speed, automation etc. to explore ideas or execute processes.
Developing an ontology, a native semantic model is valuable because it will:
- enable ontology alignment i.e. potential semantic integrity across domain or ecosystem;
- curate existing content (source, lineage and context);
- position content logically & consistently within a general information architecture context;
- identify specific content by tagging it with a unique identifier, without which analytics become challenging or even meaningless;
- identify logical relationships between concepts & entities, enriched by lineage, context and a unique identifier (URI);
- programmatically describe entities, their types & their characteristics which are then able to be represented in the graph and as a sub schemas;
- resolve multiple identifiers into a single concept, enabling logical errors to be detected before they enter a system.
This is what makes a knowledge graph the 'organising mind', capable of automated reasoning which humans can then exploit to derive new knowledge which would otherwise be beyond the range of human capabilities. That is the pursuit of knowledge.