New Opportunities for Connected Data by Ian Robinson, Neo Technologies, Inc.
An in depth discussion of relational, NoSQL and graph database views of the world.
I must admit to being surprised when James Frazer’s Golden Bough came up in the presentation. It was used quite effectively as an illustration but I have learned to not expect humanities references or examples in CS presentations. This was a happy exception.
I agree with Ian that the relational world view remains extremely useful but also that it limits the data that can be represented and queried.
Complex relationships between entities simply don’t come up with relational databases because they aren’t easy (if possible) to represent.
I would take Ian’s point a step further and point out that logic, as in RDF and the Semantic Web, is a similar constraint.
Logic can be very useful in any number of areas, just like relational databases, but it only represents a very small slice of the world. A slice of the world that can be represented quite artificially without contradictions, omissions, inconsistencies, or any of the other issues that make logic systems fall over clutching their livers.
BTW, topic mappers need to take a look at timemark 34.26. The representation of the companies who employ workers and the “contagion” relationships. (You will have to watch the video to find out why I say “contagion.” It is worth the time.) Does that suggest to you that I could point topics to a common node based on their possession of some property, say a subject identifier? Such that when I traverse any of those topics I can go to the common node and produce a “merged” result if desired?
I say that because any topic could point to more than one common node, depending upon the world view of an author. That could be very interesting in terms of comparing how authors would merge topics.