 So next we're going to hear from David Keeter. He's no stranger to INCF. He's part of it's part of the neuroimaging task force now You're leading you're one of the members of the NIDM Special Interest Group, so you're going to tell us a bit about what you guys have What you're up to and what you've done, right? Thank you. Um, so I'm the chair and my co-chair Camille and I are the chairs of the neuroimaging Data model special interest group So just to give you for those of you aren't familiar with NIDM just to give you like the Few sentence spiel on it. It's we use provenance just like we just heard about neuro shapes Basically, we build a series of specifications on top of semantic web technologies and the Prove family of specifications we have a Core of vocabulary and try to sort of push the semantic web Tools or terminologies ontologies in our domain of neuroimaging the idea here is that We would build these models for different stages of the scientific research process make data more broadly discoverable and reusable by having records of Precisely what was done to the data Rich descriptions of the experimental protocols and the study populations all the metadata and stuff that you can't sort of represent in the normal data formats that we use in neuroimaging This is a Sort of a working group that formed out of the INCF data sharing task force in about 2011. There were a few members at the NIDASH Working group who were interested in how metadata and analyses and provenance could be described in neuroimaging We're also sort of graph nerds And we like to work in the graph space and we decided, you know, we'll try Try out these semantic web technologies and see how well they work and how flexible they are And to this day we still meet almost every Monday morning anybody's free and welcome to join we have a Online Google hangout sort of meeting at the times you can see there and If you go to the links will be at the end But if you go to our Google Drive, you'll find our meeting minutes all the way back to 2012 and all kinds of information of what we've done along the way even any grants that have come out or been written About 90 am we try to encourage people to make those available all the code and everything else So try to just be open and transparent and it's just a group of people who are interested in sort of pushing these technologies So we I just put this slide up because I wanted to mention that we have a few posters 80 82 and 84 So kind of after this session will be out there will be some demos. You can see how some of these models look And some of the tools that are available So we had this scientific or special interest group meeting on a few days ago And so if you're interested in seeing the notes for that meeting again, everything's available So I made this little bit Lee bit Lee link 90 I'm sick 2018 so you can go there and you can see the notes and things that were captured during that meeting we had The participants you see there and so one of the discussions it was a three-hour meeting So we didn't really get a whole lot kind of done But we did have some discussions on common data elements. And so this this idea that so people say people are marking up their data or making their data available and One way or the other they use a concept. They're whatever they collected for say handedness They can maybe they collected handedness Edinburgh. Maybe they asked a participant, which you know hand you grab a broom with and Say you want to query across these data sets In a broad fashion, you know without knowing what each individual data set collected it'd be kind of hard to query across them So this discussion is about How do we push the terminologies and the ontologies to say, you know Maybe we need to define sort of a summary or a federated common data element that says, you know, simply binary handedness was collected and Maybe it was left or right and then you could use this to query across these data sets broadly Once you download the data sets then you find out that data set one really collected the Edinburgh handedness scale and Data set to ask these individuals what hand they grab the broom with and so forth So kind of what do we need to do to the terminologies that we have available and the ontologies that we have available in our space of neuroimaging to facilitate finding data and Subsetting it and using it essentially So then along those same lines. We had a discussion about a standardized query API. So You know, could we define a high-level query API? that folks who develop databases folks who develop semantic web applications Folks who work on data formats could sort of use to define queries for their data and so the idea is that You know be neat if I could issue a get project or what projects do you have query to a bunch of data resources that were available on the web and Expected that they had implemented this where appropriate and in some cases, you know Data from databases aren't broadly available and all you can get back as counts. So we had a discussion about What kind of queries that would work Among star communities so maybe we have we all implement a query that says just give me counts of what kind of how many subjects You might have in your database So that kind of discussion can we come up with a spec for a standard API and then all the different groups could go off and implement it And this would help, you know application developers and people trying to find data and so forth Worked with the Neuroshapes people Excuse me learning about shackle shapes descriptors and about how writing schema for your Particular data format Could be used by a tool called a pine For developing standardized query apis. So very appropriate. So we had some discussion surrounding that and then a subset of folks went off and worked on a little project We had other projects that were sort of submitted for You know things that people would like to do have done during our special interest group meeting But we didn't have time So one of the projects though that did make some progress is this this this project of being able to search through 90m documents or graphs with data lads search routines and So the Camille and others would Use 90m results documents, which you heard about yesterday And so in 90m results documents are documents that contain results from mass univariate neuroimaging sort of statistics And put them in this sort of RDF graph based models and the the goal was to you know How can we facilitate queries with data lad of these graphs? And so The top box which I guess it's not highlighted in gray but the top box is kind of what RDF looks like in particular serialization format called turtle And you can read that top line that says Neary ID one is a 90m design matrix. That's a URI you can go out and find out what 90m design matrix means And so forth. So then you read it neary ID one is at prov location Design matrix dot CSV and so forth. And so the idea is that you'd Prefer for tools like data lad and probably a lot of other tools a flatter sort of JSON structure That you can query over without using query tools like sparkle and so they did make some progress in that regard using indexing features of Json LD And Trying to sort of flatten out these graph structures. I mean these are graphs So, you know, you can't really flatten it to a tree without picking a node But so but you know get it as flat as we can Such that we can use some generic search tools within data lad to find data within these graphs And that's all I have We have a whole bunch of resources. Hopefully these slides would be made available But essentially if you want to see the sig notes 90m sig 2018 bitly If you want to go to our Google Drive where we have all this information 90m drive also a bitly link and then Get hub under incf-night-ash. There's a whole bunch of our repose That's it