What is the difference between materialization and query re-writing semantic reasoning?

Next episode: click here

‍

Hello and welcome to another episode of the RDFox introductory series.

In this episode we're going to be looking at differences between reasoning solutions, focusing on materialisation versus query rewriting, and of course having a look at RDFox as a materialisation solution.

So what is materialisation? Well here when we reason, we actually add the result to the graph. So any rules that create insights, these triples are really added to our data store at the point that we add the rules or add new data that is impacted by those rules. Query rewriting on the other hand, simply rewrites the query at query time.

Now both of these methods have advantages, and the first is that query logic can be significantly reduced. So when you're writing queries, we can do that faster, it's easier to collaborate on them, and it can be easier to debug them as they're often much shorter and simpler.

But materialisation has some key benefits over query rewriting, the simplest being that of performance. So because materialisation actually adds the results to the graph, when it comes to query time, all the computation has already been done, so the query simply has to find a particular value and return that.

This can be orders of magnitude faster than computing that potentially complex property at query time, which as we've said, is what query rewriting does. If you'd like to see examples of these orders of magnitudes time savings then check out our other videos where we dive into rules in more detail.

Materialisation also offers a benefit called recursion when rules can be self referential. This is an incredibly powerful technique that allows us to encode things like transitive closure and complex property paths, where complex patterns are captured and we can traverse along the graph of any arbitrary number of steps.

So let's have a look at RDFox in comparison to all of this. So as we've said, RDFox is a materialisation-based solution, but it goes much further than many other materialisation solutions.

Often materialisation and query writing is simply ontological reasoning—that’s the reasoning that we've seen with OWL. But RDFox takes things even further still with Datalog that incorporates advanced functionalities such as negation aggregation and filtering. With this, we're able to do things like graph analytics, data analytics, local and global constraint validation, and all of all of the above, all at once.

It's these features that our clients have found incredibly valuable, enabling them to encode, automate and scale their domain expertise and business logic, ultimately empowering them to solve some of Enterprise's toughest challenges.

‍