Minutes from Dec 3 Telcon

Minutes from Dec 3 Property Graphs CG Telcon

[12:04] <Ashok> present:Zhe, Ivan, Gregg, Patrick, Michael, Ashok
[12:04] <zwu2> I can scribe
[12:04] <Ashok> scribenick:zwu2
[12:05] <zwu2> Topic: review meeting minutes
[12:06] <Ashok> http://lists.w3.org/Archives/Public/public-propertygraphs/2013Nov/0019.html
[12:06] <zwu2> any objection to approval?
[12:06] <zwu2> Ashok: meeting minutes from Nov 19 approved
[12:07] <zwu2> Ashok: a new member
[12:07] <@ivan> Norbert Martinez
[12:07] <zwu2> ... Norbert has nominated a colleague Rachel
[12:08] <zwu2> ... Rachel may join next week
[12:08] <zwu2> Ashok, today, let's discuss the use cases in the wiki
[12:08] <Ashok> http://www.w3.org/community/propertygraphs/wiki/Usecases
[12:09] <zwu2> ... have 9 bullets now
[12:09] <zwu2> ... each of them is a graph algorithm
[12:10] <zwu2> ... first one is clustering, two use cases there. 1) clustering from Zhe
[12:11] <zwu2> ... 2) second use case in the social area
[12:11] <zwu2> ... goal is to find experts
[12:11] <zwu2> ... a Boeing presentationhttp://www.w3.org/2013/socialweb/presentations/boyet.pdf
[12:12] <zwu2> ... second bullet Vertex ranking
[12:15] <zwu2> ... Embeddedness from Facebook, social networks
[12:15] <zwu2> ... they tell you your friends, who you are connected to
[12:16] <zwu2> ... and the people your friends are connected to
[12:16] <zwu2> ... Dispersion: predict whether a relationship is going to be long or short
[12:17] <zwu2> ... Page rank
[12:17] <zwu2> ... Shortest path. Patrick thought of this use case
[12:17] <zwu2> ... I don't quite understand it.
[12:17] <zwu2> ... would be great to elaborate on this a bit more
[12:18] <zwu2> Patrick: will do that
[12:18] <zwu2> Ashok:  Michael's usecases
[12:18] <zwu2> ... E-commerce. should not be a bullet here because it is not an algorithm
[12:18] <zwu2> ... put in there so that we can talk about Michael's use case
[12:19] <zwu2> Michael: a use case for property graphs
[12:19] <zwu2> Ashok: we should classify by algorithms
[12:20] <zwu2> Michael: I can find some examples and algorithms. For big companies, most of the algoirthms can fit in their usage scenarios
[12:20] <zwu2> Ashok: classifying by algorithms is clearer
[12:20] <zwu2> Michael: will do and send out this week
[12:21] <zwu2> ... want to extend the E-commerce usecase, right?
[12:21] <zwu2> Ashok: those are the use cases we have
[12:22] <zwu2> ... got this nice paper from R. Martinez
[12:22] <zwu2> ... turns out there is a Linked Data Benchmark Council
[12:22] <Ashok> Paper from Norbert Martinez http://ldbc.eu/sites/default/files/LDBC_D3.3.1.pdf
[12:23] <zwu2> ... please look at it
[12:23] <zwu2> ... talks about different graph models
[12:23] <zwu2> ... and different graphs (sparsely connected, strongly connected, clusters...)
[12:24] <zwu2> ... also talks about algorithms, languages that exist to query graphs,
[12:24] <zwu2> ... and a bunch of use cases
[12:24] <zwu2> ... very relevant to what we are doing here
[12:24] <zwu2> ... one thing is that all the graph languages are imperative
[12:25] <zwu2> ... no declarative graph language
[12:25] <zwu2> ... possibly has something that we can add to the wiki
[12:26] <zwu2> Michael: many implementations on graphs, languages, and they are targeting at specific use cases
[12:26] <zwu2> Ashok, yes, they are not general purpose
[12:29] <zwu2> ivan: looks like the paper does not even talk about RDF, SPARQL
[12:29] <zwu2> ... my first impression by scanning through the paper
[12:30] <zwu2> Ashok: Greg has mentioned that we can extend RDF
[12:30] <zwu2> Greg: I sent this to the SW interest group
[12:30] <zwu2> ... Andy followed up on that with a couple of research paper
[12:31] <zwu2> ... use subProperty + bNodes to allow attributes on links
[12:31] <zwu2> ... ivan said it may be a bit premature
[12:31] <zwu2> ... bNode properties in JSON LD
[12:32] <zwu2> Ashok, once we start a standards effort, then it becomes critical
[12:32] <zwu2> Greg: I think serialization format is important, we may repurpose JSON LD
[12:33] <zwu2> ... JSON LD does not promote RDF
[12:33] <zwu2> Ashok: it may be helpful politically
[12:34] <zwu2> Ashok: a question for ivan.
[12:34] <zwu2> ... we have been collecting use cases
[12:34] <zwu2> ... haven't really thought about direction
[12:34] <zwu2> ... if we want to argue that PG WG should be started, what kind of further arguments do we have to make
[12:35] <zwu2> ivan: two things: 1) it is still not clear to me based on the use cases, what kind of conclusions do you want to draw?
[12:35] <zwu2> ... what exactly do you want to put in the standard. In other words, what are the charters?
[12:35] <zwu2> ... It should be much clearer. 2) more social,
[12:36] <zwu2> ... a problem to me that in this CG we could not get many of the people from PG community
[12:36] <zwu2> ... Norbert is the only person as I know
[12:36] <zwu2> ... we are potentially getting into a social mine field,
[12:36] <zwu2> ... need consensus from major players
[12:37] <zwu2> ... at the end of the day, it is not my decision
[12:37] <zwu2> Ashok: we also had Danny Bickson who is one of the founders of GraphLab
[12:38] <zwu2> ... which is part of the graph community
[12:38] <zwu2> ... not sure if we can get others
[12:38] <zwu2> ivan: talked to many, unable to convince some folks, I am a bit stuck
[12:39] <zwu2> ... at the social web workshop
[12:39] <zwu2> ... a work at W3C that may lead to social web WG
[12:40] <zwu2> ... not talking about standardization across FB, GOOG, LinkedIn
[12:40] <zwu2> ... those players should be represented here or they should be aware of what is going on here
[12:41] <zwu2> Ashok: it occurred me just now. Suppose we put three bullets in charter
[12:41] <zwu2> ... 1) standardize PG
[12:41] <zwu2> ... 3) declarative query language
[12:41] <zwu2> ... 2) algorithms
[12:42] <zwu2> ... do you think these can be used to attract people to this community
[12:42] <zwu2> ivan: technically, I think more or less the same way. Maybe 4) Relate to Web protocols as we did with SPARQL
[12:45] <zwu2> Ashok, if you look at the URL Norbert sent us
[12:45] <zwu2> ... all of the significant graph players are relevant
[12:45] <zwu2> ... maybe we can speak with them
[12:46] <Ashok>http://ldbc.eu/results/deliverables
[12:48] <@ivan> NuvolaBas
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
[12:48] <zwu2> ivan: talked to Oracle,
[12:48] <@ivan> Aurelius/TinkerPop
[12:49] <@ivan> Neo Technology
[12:49] <@ivan> Objectivity
[12:50] <zwu2> ivan: If you get all these lined up plus the partners Harry collects, that it is a much better case
[12:51] <zwu2> ... the person to convince is Jeff
[12:51] <zwu2> Ashok, will contact Harry and Norbert for advice
[12:52] <zwu2> Topic: any other business?
[12:53] <zwu2> bye
[12:53] <mpetyx> bye

-- 
All the best, Ashok