Resources on Property Graph vs. RDF Triple Stores

Ashleigh F. · 2020-10-06T16:40:31.027Z

Hi all, I am asking for resources or advice on the difference between property graph and RDF triple stores. I have a publication on this, but wanted to get a feel for how others think about this topic, similar to this article: https://towardsdatascience.com/representation-learning-on-rdf-and-lpg-knowledge-graphs-6a92f2660241

The Knowledge Graph Conference

Daniel C.
·
To throw a third option in here: I'd rather not use property graphs or RDF triples at all... leave the complexity, inflexibility and verbose mechanics at home for a simpler more expressive model and language. Grakn implements a hypergraph under the hood to abstract this complexity. This article doesn't get into the diffs btw RDF* but will better explain what I mean: https://towardsdatascience.com/comparing-grakn-to-semantic-web-technologies-part-1-3-3558c447214a
😀1
Ashleigh F.
·
Thanks Daniel C.. The audience I am looking at are not computer scientists so more COTS-like, or at least something with a basic GUI, is the main focus for now. I see Grakin has something sort of like a workspace/interface-like thing that could be similar to what Neo/Stardog-like tools have. Is that correct?
Daniel C.
·
Ashleigh F. correct, Grakn Workbase is Grakn's GUI or IDE. For the audience you are describing, I'll bet they fall in love with Graql our query language - notorious for helping non-technicians understand their data/ database
Ashleigh F.
·
Ok, that helps. What is your opinion on some saying database specific query languages may limit you? I know they are common on the property graph side, but isn't it difficult to switch DBs if you have everything calibrated to a custom query lang?
Efstratios K.
·
Hi Ashleigh F., I watched recently this 60 min webinar by TopQuadrant’s Irene Polikoff and I considered it a very intelligible presentation of the two approaches.
Martynas J.
·
Daniel C. maybe you should mention your bias and the fact that Grakn is a proprietary product with a proprietary query language, where as RDF is an ecosystem of open standards, a variety of compliant tools, both commercial and open-source as well open datasets
Martynas J.
·
so it's an apples to oranges comparison really
Daniel C.
·
Martynas J. you're right it is a bit apples and oranges when comparing the ecosystems (many technologies needing to be used together to do a job vs. 1 technology to do a job) however, I was providing an alternative approach to the same end. Grakn Core is an open source, free database for anyone to use along with its query language Graql - proprietary, yes, but available for anyone to use and consider alongside traditional RDF approaches.
Ashleigh F.
·
Thanks to both Martynas J. and Daniel C. for the great responses! I have high respect for anyone working in this space, both RDF and proprietary systems can have a rationale for selecting them depending on the use case. I also have a video discussing the differences (and why they actually can be used together), and did reference the TopQuadrant one, but this discussion has made me think maybe I should have one focused on the use cases for both as well.
🖖1
Ashleigh F.
·
Also, does anyone else feel like sometimes TQ is a bit too strict? Then again, they have high data gov. practices built into their system so maybe that is the cause. I wonder Juan S. you work in data gov/cataloging quite a bit (over-simplification 😋), do you have any suggestions on resources discussing how to retain good data gov while also being flexible enough to meet the biz needs?
Juan S.
·
IMO, the problem is that we start getting into the weeds of the tech and forget about the BIZ PROBLEM. They don't care if it's RDF/PG/Foo/Bar... provide the solution to the problem
👍4
❤️2
Juan S.
·
Our position is that you have to strive for agile data governance. Take a look at these series of posts: https://data.world/blog/principles-of-agile-data-governance/
Juan S.
·
We have several Catalog and Cocktails podcast episode on governance. This is a good one to listen: https://open.spotify.com/episode/7uTog5f0LHMEieVxSu9D2M?si=hCQvynDGSpqD28R_nOtw6g
Ashleigh F.
·
Couldn't agree more. Its all about the use case and what you are trying to achieve, too often we jump into the how...to quote Jurassic Park "scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should" :-p
Juan S.
·
Ahhhh so true! great use of that quote!
😁1

Resources on Property Graph vs. RDF Triple Stores

20 comments

Resources on Property Graph vs. RDF Triple Stores

20 comments