CDF: User Experience Issues

Thank you for your very well presented user experience issues, @ibrahim.alsyed. 

If you don’t mind I’d like to understand better your need to hide markups, annotations, scans in events from Domain SME, specifically:

1. Should your mentioned items be hidden only from your Domain SMEs or everybody that has access to Fusion? 
   1. In case if some of your coworkers need to see that data, how do expect this should be administered? 
2.  What are the definitions of these two types of events that you’ve concluded your paragraph with: “Data Engineering Events vs Real Plant Transaction in the faci”?
   1. Which of these events do you want to hide?
   2. How would you expect them to be hidden? 

This will be very beneficial for us to know as we’re planning the development of user preferences. In case if you’d like to answer your questions  over the call, please DM me.   

Many Thanks, 

On behalf of Cognite product team,
Gedi

On the topic of Data Sets, there will be substantial changes and improvements in how you organize data in CDF in 2023, which will also be reflect in the data exploration experience. For now, use of Data Sets requires some self-organization and naming conventions to make sense.

The same applies to 3D models - we are looking at ways to organize these better. But it seems you could also get some short-term gains with clearer naming conventions and maybe some cleanup?

On Events, the raw event list contains too much varied information. As you know, richer data modeling enables some of this to go away. In the meantime, you will need to use additional filters - selecting Events and filtering by type allows me to select workorders, notifications, etc.

@gedi 

Answers to your questions:

1. A Domain SME does not need to know when the data was created in CDF for example. We also need the ability to map all the data regardless of where and when it came into CDF. For examples, as an operations SME, I would like to see Alarms in XYZ Site. In order to understand this use case, we would have to go to Data Sets, have a good understanding of where these Alarms are coming from (i.e. which data source, and understand the naming convention) to get the results we want.  The other challenge we get into is that two different process units might have different Alarm Management Software (i.e. different data source). The SME should not care which source its coming form but rather just search Alarms and get consistent data without knowing naming convention. In Cognite Infield, Work orders, notifications, time series is clear, there is no hunting of data or figuring out which source its coming from as an example. I believe developer, data scientist and data engineer view should have this view but its needs to be layered with a DOMAIN SME View. We need to discuss this in a call.
2.  In CDF today,  we should hide transactions and events that are happening as an event that are not real plant events not visible to Domain SME. For example, Cognite Markups, Annotations are being created to support Remote or Infield. This is not something Domain SME cares for in CDF. Another example would be someone uploading 3D Scan in CDF and an event is created for that; for a Domain SME - 3D Scan is not an event, but real plant transactions or data are events:
   1. Work Orders/Notifications
   2. Sample Lab Events
   3. Alarms
   4. etc
3. I see Creation of SVG files to support Infield in CDF for Live interaction, and now that should be hidden too because Domain SME just ones to see one File not duplicates.

I hope this clarifies somewhat. I really do not want CDF to be a messy data lake but a usable product for people to find and search data in the most meaningful way. 

@terjelo i think we should setup a call on UI/UX in the new year to address and discuss challenges.

@terjelo, @ibrahim.alsyed  clean up is in progress and will help, as the names will follow the existing naming conventions. 

A few ideas:

One quick win could be to sort the dataset drop down.

On the 3D model, we could potentially gain from having a similar layout to the other resource types - e.g. Asset with filters.

Looking forward to improving the experience in 2023.

@ibrahim.alsyed those are excellent points. As more data comes into the project, it will be increasingly difficult for an SME to find something relevant using the UI.

It seems the project could benefit with additional metadata (e.g. labels) and naming convention, as others have pointed out. Having the capability of putting all together into the same place does not removes the need of data engineering and management.

In addition to the ideas above, perhaps applying some out-of-the-box unsupervised clustering of objects of the same type (in the absence of metadata) could help sort/group things.

My 2 cents.

To include hiding source files from domain experts  - those that are not mean to be viewed/explored via CDF e.g. point clouds 

cc: @Philippe Bettler  @rsiddha 

continues to be a challenge.

Hi Ibrahim,

All the points are excellent and part of our roadmap during this year. Thank you for sharing the details and context. 
A few solutions on the above items, that are coming soon, in the next releases (February and April) are: 

• Annotations will be removed and not visible anymore under Events. For scans we will need to find another solution. 
• Sorting based on frequency and alphabetically in the filtering menu, including for the dataset filter. 
• Organizing the 3D page

We are looking into a series of improvements of the filtering menu, would be great to detail them together and get more insights into what would be relevant to have here.

​​​​​​​Thank you,

Andreea. 

@Andreea Pastinaru whats the status of the items mentioned in the your post?

More issues:

Please see how text is overlapping  with the next column in two occasions

@ibrahim.alsyed: as for the point on improving the 3D model overview page, this is currently in progress and should be part of the June release.

Hi @ibrahim.alsyed,

Lars mentioned the timeline for the 3D model overview page, as for Point no 2 will be ready by the end of the month and for point no 1 will need some help from @Philippe Bettler with the timeline.  

Thank you,
Andreea.

@Andreea Pastinaru  what about my last post on overlapping text?

@Andreea Pastinaru @ibrahim.alsyed , as we roll out FDM objects, there will be more clarity on what the user is looking at, and it will become easier to navigate the data for #1.

The use case is interesting as the UX will need to make it easy to bring different but related objects constraints in parallel e.g. the users wants to see Alarms for a given site. 

@Andreea Pastinaru if we want to improve data exploration in CDF then it is important to acknowledge that data in CDF must have a domain expert UI. This means that there is no such thing as data sets, data source or any other terminology as meta data etc. 

Hi @ibrahim.alsyed ,

Ref to your last post on overlapping text, the team is working on it, should be very soon fixed. 

I totally agree and aligned with your last comment. The terminology will change as we roll out FDM objects, as Philippe mentioned as well. 

Thanks,
Andreea.

@ibrahim.alsyed The issue is fixed and we just released the newest features in Data Explorer. 

As for terminology and more improvements with focus on the Domain Expert UI,  will come by June. 

Hi! We wanted to come with a short update. We are currently working on a data exploration experience with the business/Subject matter expert user in focus, where our goal is to abstract away the “IT language”. We will be needing several iterations to nail this, and engaging in user testing across the customer base. An update on the status will be shared in October release. 

Hi there!
Providing an update on this request, as we have addressed or are in the process of addressing the points raised.

We are significantly simplifying the exploration experience with Industrial Data Search that focus on the Industrial/SME user and their needs.

What you can expect in Industrial Data Search; 

• Filters are not focused on the more “Data Engineer” user group, but rather the SME. We are not exposing all potential filters (Data Sets, IDs etc), and will also in the next couple of releases support customization within in each project to define what filters to show. 
• Event types over the next releases we will give the option of pre-filtering data for all users of a project, as part of supporting more customization. One such example, can be to define some filtering rules for the type of Evens you deem valuable for the Industrial/SME users to see, and have those filters auto-applied. 
• Location/Site/Facility is possible to configure, and acts as an efficient filter on the data a user has access to. For an Industrial/SME end user, this simplifies the data exploration, as they start their exploration on only the data belonging to the facility they care about. 
• Finding the right 3D model has been challenging. In Industrial Data Search, one can configure 3D scenes that combine 3D models (CAD, pointcloud, 360 images) and ties them to the correct facility. The user does not have to select which 3D model to open, this is configured for them.
  • Directly from the Landing page in Fusion.cognite.com, the user can click on “Browse 3D” and be taken straight into the digital twin. 

Best, 

Sofie Berge, Product Manager