Hello Yuyao,

I don’t believe this to be a best practice, but if you are set on displaying data from multiple knowledge models / data models at once on a view, I believe the only way to do this would be in legacy views - each knowledge model’s KPIs would be displayed using separate embedded views, as you can only have one KM assigned to a view. So make a view for each KM, then create a master view with each of these embedded. The limitation here is that you couldn’t have any KPI’s that make use of data in another knowledge / data model, as there isn’t really a way for them to interact with eachother (each KM can only reference one DM)

As for best practices, in this scenario I think the best thing to do would to have all of the data within one data model, which would avoid the workaround / limitations above, as well as other limitations not mentioned.

Hello, in Views (latest version) every component can have an individual Knowledge Model so you can easily combine data from multiple KMs/DMs in one View. 

Please also see the documentation here https://docs.celonis.com/en/view-components.html#UUID-ae42e80c-6003-0fc2-c074-1281cd9c9900_section-idm4655203625681634244349333046

Happy building!

Thank you for your reply.

However, I am facing two issues:

1. When I switch the Data Model (DM) in the Knowledge Model (KM), the visuals built on data from other DMs are no longer functional.

2. Another use case I need to realize involves having one KPI that retrieves data from multiple DMs, which I cannot achieve by simply switching the DM/KM.

Thank you so much for your reply. One use case I am trying to implement involves having a KPI that retrieves data from multiple data models (DMs). Even when using the legacy view, I’m unable to achieve this. Could you confirm whether this is feasible with the current setup, or if there is another approach we should consider?

Hi Yuyao,
considering your requirement that you have one KPI that is overarching multiple datasources this sounds a lot that the data you have should not be split into multiple datamodels.

Obvisouly it is never a black and white decision but by rule of thumb I would cut the datamodels always so that “what belongs together for the given enduser goes together in the datamodel”. What I would definitely not consider in cutting the datamodel is which source system the data is coming from.

So this goes to emphasize what Andrew already wrote: It sounds like you should transform your data in Data Integration and make it one datamodel and one Knowledge model above it. Then you have solved all your requirements which you described. I see no benefit of having them seperately in your case. 

If (and thats a big if) you want to show closely related processes here and do really need the splitted Datamodel you would still have to have one single overarching Datamodel for the KPI you described. Essentially ending up with a rather complex structure of Datamodels for each Busienss domain + one which combines all, but thats a road I would only take if absolutely necessary.

Dear Manuel,

Thank you for your response.

To provide some context, we are working with the procurement process, and within the company, there are two types of systems being used. The table structures in these systems are completely different. Despite this, we still want to calculate key metrics such as the total value of purchase orders, etc. This brings some complexities to the data integration.

Hi ​@yuyao.wang13 ...just an idea or a hint. Have you tried to create a Multi Even Log for your DMs and then a KM based this MEL which then could also be used for creation of views. Greets, Bunyod