You probably won’t have heard of either SBS (Semantic Business Solutions) or VALID (Vastly Linked Data) but SBS VALID is a very interesting product. It is a tool that, in essence, provides a hybrid query/search capability.
Of course there are several such products on the market, which aim to combine query and search capability across both structured and unstructured data. However, VALID has a different underlying architecture from anything else that I have seen in this space and it is from this architecture that the product derives its strength.
First of all, VALID is based on a (distributed) in-memory query engine. That’s hardly original nowadays. What is original is that the engine is inspired by the tuple-space communication and coordination concept but I’m not going into that. What I will get into is that the tuple space has a graph engine built on top of it. So, what actually happens is that your data is represented as a graph within VALID, though this is actually hidden from users. What users see is a natural language (semantic) query that actually walks the graph to answer whatever it is you are querying. Note that it is important to understand that semantic query is different from semantic search because with the former you can do things like create temporary result sets, which you can’t do with semantic search.
Okay, I’m aware that this probably doesn’t give you a clear idea of how VALID works or what it does. Let me walk you through an example. The company showed me an example whereby your query starts with “Andy Warhol Events Berlin”: you simply type that in and get back a list of all recorded events that have taken place in Berlin that features art works or appearances by Andy Warhol. So far, so search. What is different is that you can also type in something like “last 15 months” and just get the results for that time period. SBS calls this last function a semantic operator and it is really what makes this is a query engine and not just a search capability. Semantic operators are defined in advance and the company is building up a library of pre-defined operators as it acquires more clients.
The other thing that needs to be defined in advance is the data that you are going to run queries against. VALID imports simple triple strings which can be easily generated from SQL- & NoSQL-databases. The unique thing is that the semantic queries work automatically (no query development is needed) and the semantics (the relationships between concepts and data) from the databases are retained (an ontology does not need to be developed). This way VALID significantly reduces the complexity inherent in semantic projects.
The truth is that you really need to see this and play with it to get a good idea of how it works and what it does—my few words do not do the product justice—fortunately, the company plans to provide a portal to Wikipedia, which should be available shortly. I recommend trying it out: I think you’ll be impressed.