How to industry sign banking georgia executive summary template
okay now now it works no yeah weapons II when I got the first email from the EDMC about chords for presentation one of the topics was viable relationship to other industry standards and whether it's the area have worked in a lot and in the past years on semantics and industry standards so out of these I picked the Xperia L standard which is quite nice because there were several questions in the morning about XP al and I'm pretty sure that this presentation will in cyber a debauch fish that you will ever have about getting experience data into fiber and creating XP al out of survival and we go through the example of the US bank our report anybody familiar with it yes I've seen those who work in banking ah okay let's let's go so we know all five was the key to better data management or that can be difficult to get started with survival and to integrate industry standards now why are we here like this this guy here in the shadow he sits over there on the the fiber website there they had there's several reasons why to attend and the one that I care most about is to learn the best practices for deploying fiber I guess that's for many you of you here in the room that's the main reason that you travel to Atlanta and now with that we get for the finance side executive side we we get there's a use case of proof and trust in the data and for the technical folks of program management in architecture it will help us to manage these a daunting complexity about what we heard a lot today and now actually if you see this deck here like always here at the no it doesn't want anymore the least I don't see okay it is a bottom left corner you always see this little circle so he and F primarily addressed it fine and silence keep odd when you see a slide over secure when you see a slide with P it goes one level deeper details the key point was further explanation or the yellow slides here are architecture technical details so if you take the deck you can trim it down to an acceptor presentation of seven slides to 15 slides all these are for 40 pages now anybody has seen this picture nobody okay now this is the office of consolidated life insurance the New York office from 1960 and when a few years after I came to America by first big bezel project was at the bank in Charlotte North Carolina and it was exactly the same thing they cleaned out a whole floor and part of me there were a hundred other IBM consultants in there and as a hundred consultants from other vendors it turned it so again we had the hundreds of people all all in the floor and now what did they do hundreds of people mapping primarily between hundreds between numerous heterogeneous systems different languages different metadata and components and now this is a bright new word the future and how I see it it's actually a virtuous circle so where fibre integrates industry standards in turn supervisory regulations they are expressed in standards and then the regulations in turn they are a really good driver to build out your fiber program so it really becomes a self-enforcing beneficial cycle okay now how do we do that onboarding the industry standards compliance forms and reports into fiber it's really an easy two step process and we said it is source staging target and we've all done this and seen this 100 times in conventional ETA now the only thing that's different that's improved is that here this middle part everything here is Symantec ETL and with that we get the benefit of a common storage that every single image mapping every single stored in the front uniform same way everything is available okay and just one note about it her this the X BL and the core report ontology that I talked about it's it's open source it has the same license as the fiber so you're free to to download ZZ xpl files to download the FFIEC ontology and play with it okay so step one year from source to staging we extract the industry standard XP l and and the core report into ontology staging now see the core report the Federal Deposit Insurance Corporation they are the author of that report they are the auditor of the banks okay and now the FF IV sees it's an interagency body they work for FDIC the work for the FET and they are doing some of the IT stuff okay so they are the publisher of the core report and XBRL they are the ones who accept the bank filings they are the ones who disseminate the information to the public okay and all we do then is we want to go from ZZ xpl we reverse-engineer the schema we reverse-engineer the data into staging okay xpl extensive business reporting language it is a global standard to exchange business reports and now the XBRL ontology is the one-to-one representation of the XBRL schema okay so there's no praying work no movement really on it it's all want to worn a different representation of the same specification okay so what xbl does very nicely though therefore I like it a lot it separates the reporting items from presentation calculation edit checks definitions captions and so on and does this in the car with three files and that's all the core report X X link allows us to to link different points in XML documents and link base that extension to X link it puts a framework in a - to define rules and relations on these links so with that I can say I have links for presentation for calculation and edit checks so that's what link based us and then finally instance that is the basic reporting item with the meter data context units data types and formats so these three building blocks really make up the core of XBRL we know the benefit is we only have to load instance data into the ontology we do not have to care about the report structure and our input our data is develop even if the structure of the report changes okay so now how does it happen like here we are used to operate see is the ontology editor which is an import facility for xst and what it does is for every namespace in z-axis easy XML schema it it creates an ontology file so I get around the import I get X link link paste instance ontology files all files ok so how does the fire look like at the left hand side this is the from XML spy look at numeric item attributes which is a you attribute group within zzx sd4 for instance and now converted if I look at the otology class for it I see that here the precision and the decimals they have become they have become here class restrictions okay and and now very important what the input also does is this year this s XML take semantic XML take this is the original tag from my my XS T and this is how as in the tooling imports x PL instances and how it writes experiences it looks up it finds attack in the original it looks up whereas my class with attack and that is what writes the instance ok now so far I have the expiry framework now and taxonomy define specific sets of report ok so so then it extends the xvi based classes with subtypes for specific reporting items these are the sections tables calculations for very well-known of course is US GAAP XP al is worldwide the most common utilization of XP al I'm currently working on Soylent sees that the insurance industry where in Europe they utilize XP il for their supervisory regime and today we look at the car report no the FFIEC they publish they do a pretty good job for a government agency read Isaiah they publish the the schema here the taxonomy and say they they also publish instance file of Bank report so here's the deutscher from September last year so you can look up any bank any any schedule and you find a whole set of reports published by the FFIEC no the FFIEC ontology just like we had before is an ex tech older version of the taxonomy and also one-to-one representation of the FFIEC taxonomy file so we end up having here at the top the expiry instance FFIEC instance extends these basic reporting types we have FF IC concept and this is really like where the bulk is the FF IC has a micro Delta reference manual that defines like 4,000 different reporting items in the call report and other reports as well and these are mail mostly numeric items no asset values at sword codes or together over 4,000 items and then at the bottom I have the the instance file in this case it's JPMorgan Chase from September last year and that is the file that holds the actual data values so you have a whole hierarchy here of ontology files including each other or graphs including each other ok now the instance file here this is a eden example so here it has an item of CSS d it has a context and this is the actual value so if you open an XP l instance file from the FFIEC you will see like two thousand lines like this and it's actually it's it's quite nice I love I experienced over this because it's so simple to read to pass and to put into viable I know the import then generates instances of the concept so for every of the two thousand line and the Jaypee manches report I get instances here and they are all of the type here of the EM DRM item that it finds in the expel instant source file ok so we go from here and it imports it s instances at resources was a type of the FFIEC class okay and now then then again here very important like here we have Mir or the whole list of Z Z thousands of classes and then again the XML tag that preserves like the original x SD item okay and then I can query the two and I can compare your mind my download from the FFIEC this is the entity schedule from the choral report and I compare it here with with C query results on ontology staging and here just is a complete set of the FFIEC ontology file it's now over here for our data we only need this track here if we want to reproduce the core report out of the ontology was an Sparkman query then we need here the calculation edit check presentation definition and so on okay so this is how the query looks like you heard that already today it's very similar to stew sequel we specify like the items that we want to to select and the where clause is basically joining all these different triples and then the order by here's just by the Coward schedule and the line number and then I can compare like the top here you can download this like there's a presentation I think it's with distributed files up here is the FFIEC download here are the results of the queries that we saw in the in the previous page and we are in tentacle know so now what where are we now we got the data in ontology staging and we have validated note that the data is correct we reproduce nause is your original FFIEC report so next step step two is to transform experienced a gene and loaded into into five of classes so the way that works is that these gears here's the inference engine where that's the motor and it works through the mapping that we define and then it executes the load into target by using sparkle sparkle construct which is the same as in sequel insert okay so how do we do that so in the ontology editor we can pull in here our source in this case from old staging and FF IC concept here this is the one in the core report for the legal entity identifier target site FIFO already has a legal entity identifier excellent and then we just connect the two classes and we tell them the tool how to build the URI likewise we can take data property here this is the actual value of Z le I and we want this copied over to the unique identifier in five-o le I and and here like the way it works here we can freely define like what is the data here that we want we define the template and here I think this is a good convention at user use of fiber prefix le I to tell how the argument is made up and what comes out of it then is a full string of say you are ina starting with a fiber root ending this value of c le i and one note on on this I basically matter who will target up with the le I know so the le I propagates from the le ale I value to the stock cooperation to the day so everything identified by the le I which is a great thing that we is that we have said in in fiber really okay now yeah likewise I can met an the data property and for things that are bit out of the box unusual beyond the simple data movement we can always take a spark will rule so here what I want is for every every instance I create in fiber I want to know what is its source in old staging so I populate at an object property has source instance and this year is just the where class where I utilize the mapping context and it will write this attribute pointing to the origin of my record in fiber okay and and now we we run this engine and we will look at the fiber data mapping and the lineage so for the data we can seize it actually now my class tribal legal entity identifier it has a new resource a new instance and the source instance points here to my original staging record a spark will rule has populates the link to the cooperation so le I is the values the identifier code and it links in 502 the cooperation that it identifies and yeah and finally here the unique identifier it has a correct where you populated now in this year is a complete graph in fiber I don't know if it's if it's readable here also here we have our legal entity identifier which identifies a 500 stock corporation so stock corporation has issued capital and monetary amount there's a registered address in Ohio and it has in it is identity and this is the functional business entity and fiber another roles at JP monks place and which is a depository institution and also then it has another identifier which is the FDIC certificate number not so 628 is a number FDIC knows JP Mancha yes No and with that it's also here in the track its regulated by the FDIC and it's in the FDIC institution directory and very nice about fiber these values here already populated yes for for this use case no I didn't it was all there with it's why I took this example I I did the I did the same thing for the with the Securities Exchange Commission the investment advisor Act and dodd-frank and there was quite a bit to to extend and to it no but this this for example it's all for the target yeah not and the targets like where I did extended on mainly for investment funds and and hedge funds and that is also open source know so you you you guys can can download and look at it where does it extend the fiber and of course you know like we can work together I can help like understanding it no yes this is this is correct yeah this is correct and and I mean here we have niklas here from Deutsche Bank we had this this debate you know can you really call this an ontology if you reverse engineer database table now that's why I just call it old state Jana but it is on purpose that I didn't add any smarts to the staging area because and we get to that I should run out of time because I also want to generate the core report out of fiber therefore I need to be staging be identical to the FFIEC sauce okay and then here again is the query Z joints and traverses now in the fiber and here what you see in the query here this is all fiber no so we we have the data loaded into fiber we run a sparkle and this is what comes out of it okay and now this was the data ok yeah sure so the well and it the inference engine or reasoner that is like a process that comes with otology editors the protege has it to operate hazard triple stores have it and they the main thing what they do is to compute the inferences so top right not sort of the top rate inference engine also execute these spin rules and the mapping rules now said that's how it how it works not and but but either way at least for for populating the staging I mean this is so simple like these these simple one-line statements in the Experian instance and then the the generated rdf is very simple you can script it even though and that's that's the beauty with everything XP airbase now it's it's uniform mm identical lines basically no now let me move on V so we looked at the data and fiber now because it's all semantics et al the mapping also is in triples so again I can just do a simple select and here I see all my mappings so I don't have to go to a separate ETL tool know and make a two-month project out of it for people to assemble the data it's in the same RDF store where I have my my fiber we're at the staging area so the mapping is in triples now the lineage is in triples and here I I hav
my fiber depository institution an instance of it here is a JP mom case there's a source instant and I can look it up here and find what does the FFIEC concept the class the depth so this gives me my lineage from fiber back to the FDIC and again it's a it's a simple spark will select and this is the outcome of this Karina I go from fiber class instance staging instance class all the way here to the tech in the xpl file complete Edward okay now the reverse same way here I take PI bow I do a semantic ETL into staging and I export a modified call report so there is no reverse mapping we have to do the work in the in the opposite direction in this case he is a fiber registration address it populates three core report items here for the zipcode state and the city and now our test case is JP monk chase moves to headquarters to Vermont and for that yeah I just in in the target in in 504 this instance I just overrode the city name and so on not so and now running the inference engine looking at the result and I look at this our SSD 91/30 item which is a city and I see okay here it has has changed to to Montpellier and now next step interpret here I can invoke the export at a top rate out of my ontology here I named it here JPMorgan Chase in in Vermont export this ontology to XML and then when I open the XML I can see like that it's a valid xpl open it here was with XML spy and here I get it does a syntax check for me not to say here okay XP r JP manches and Vermont is valid and so on okay now very fast note for namespaces an import that's how you should set up the architecture and we have already we have xpl we have CF fi see that input X period here we have five oh now fin rec on comm XBRL this is the namespace for the ez ontology for the x PL ontology files and fin reg Anton has reverse-engineered X BL Doc Ock and it imports fiber so that would be an example where where you have extensions 2 to 5 o and now this is still this is still a core ontology here fiber and fin record for the financial industry then you have operational ontology that really work with with data so here in this case the bank ontology that has the FFIEC classes and also it has like serves as the data files not for instances reverse-engineered from xpl instances move to to file and all this is this the reference architecture I can extend it right now I'm working with with insurances basically the same sink here with Europa that says a European insurance regulator another example we mentioned that before these Securities and Exchange Commission and here found ontology hedge fund ontology they have in the same way reverse engineer the investment advisor act form and form PF for for private funds which was a lot more difficult because they are not expressed unfortunately in XP a and as we have to figure out like the structure of the report may experience it a lot easier now final one and this one is very dear to me now besides here data files data file standard I also want to import the regulatory requirements and so the FINRA contest reverse engineered into the ontology the United States Code and the Code of Federal Regulations now a single you're all familiar with Fair Code of Federal Regulations yes yeah no so and why do I do this because it breaks it down to the subparagraph to the note I can try this requirement to my fiber class to my data movement rule to my defined class so I have to end to end from requirement to my implementation that's why it's so important to have you will see and see if I in it and you the same with with your business requirements not just like load them into an ontology on a granular level and tie your design to it okay lessons learn your fiber has a good support for compliance in particular the entity schedule from the experience and the input and transformations are kind of kind of slow in top rate and and fiber should have more content for four accounts and balances there's really not not enough like not where we we know that okay super okay to recap we want to get from the 60s to the 21st century and we can do that with semantic compliance at doing this virtuous circle of fiber standards and regulations [Applause] [Music] [Applause]