Handling the data's temporal nature is not an easy task. Imagine you have a simple customers database which shows that Jane Doe is living in Sydney and works as a clerk. You manager is preparing to contact Jane, but he or she can only guess if Jane has not moved to another city or changed job, when this information was acquired, how long does she live in Sydney etc.
A typical operational database neglects time: it represents only the "current" state of the real-world objects, although it is not really current, but rather more or less outdated. It is unknown in this case, when the data was obtained and when the object actually was in this state.
A bit more advanced approach records all the object's properties changes into a history log. This allows changes tracking and guessing how outdated your information is, but nothing more.
What if we need to select all customers who were living in Sydney in 2015? Or the customers who had ever been working as a clerk? The most of the databases with changes history log won't give an answer.
We in DataVera believe that the temporal dimension of business data has yet to be unlocked. Our EKG Provider data virtualization platform offers an API which allows to retrieve easily the data state for each moment of time. Every data change written into the platform can be attributed with a timestamp reflecting the moment when the real object properties were changed (or will be changed). We believe this gives a new quality to the data, as now they can produce much more analytical information which could be monetized.
DataVera EKG Provider represents all the data according to ontological model. In the ontological world, the time is usually modelled explicitly. If some entity can have distinct states, each state is represented as a separate object called "temporal part". This means that if we want to track our customer's living places and job changes history, we have to create a separate model entity for each customer living in a certain city or having some job. This is semantically correct, but rather inconvenient and slow regarding the data processing algorithms. A simple question like "who were living in Sydney in 2015?" is becoming a complex SPARQL query not very efficient with millions of objects.
Our approach implements data temporality on the platform level. You can run a simple query like "who is living in Sydney?", augmenting it with a timestamp referring the moment of interest, and get a correct result reflecting the past state of the data. Forget hundreds of extra data objects for each object's states, forget properties reification just to reflect their temporal nature. Enjoy the simple data structure, which still represent a real-world "snapshot", combined with ability to scroll the time axis in any direction.
Every data piece, every property value in EKG Provider is annotated with the data source (a user, an external IT system etc.) as well as the time when the corresponding real-world event occurred. This increases data trust and value, opens the way to a more sophisticated decision making.
