Skip to main content
Parked

Entity matching features option on Interactive Diagrams screen

Related products:Contextualization
  • July 26, 2023
  • 4 replies
  • 64 views

Forum|alt.badge.img+3

When creating interactive engineering diagrams on the Cognite web UI, we have the option to select between standard and advanced model.

It would be beneficial to have the possibility to combine this with customized entity matching and be able to choose from available feature types (such as "simple", "bigram" etc.).

This option would be just as, if not more, useful as the advanced detection model.

4 replies

Forum|alt.badge.img

Hi @robert.rill

Thank you for this product idea. We are tracking and processing it separately, but expect this hub post to be updated occasionally as we learn more and decide whether to include this in our future product roadmap. The responsible PM for this area - @Elka Sierra - will keep you updated as the idea makes its way through our process.


Elka Sierra
MVP
Forum|alt.badge.img

Hi,

Thank you for your feedback. Currently, we are not focusing on this specific area of the product. However, it is included in our long-term roadmap, and we plan to address it in 2025. We will update you when we have more information.

Thanks and regards,
Elka


Que Tran
Practitioner
Forum|alt.badge.img
  • Practitioner
  • August 15, 2024

Hi @robert.rill,

Thank you for the suggestion.

In the Interactive Engineering Diagram, we don’t use entity matching to identify matches between tags and assets, so including Entity Matching features isn’t really an option. However, we are continuously working to improve our detection methods. It would be great if you could share any cases that aren’t working well, as your input could help us explore potential improvements.

Thanks and regards,
Que


Forum|alt.badge.img+3
  • Author
  • Committed
  • August 16, 2024

Hi @Que Tran ,

According to my understanding, entity matching relies on calculating similarities between strings (e.g. asset names and timeseries names). The same goes on with engineering diagrams: calculate similarities between text from the diagrams (detected by OCR) and e.g., asset names. I thought the same algorithm is used in the backend in these two cases.

Now, if for instance my asset names have leading zeros randomly (sometimes in the CDF asset names, sometimes on the engineering diagrams), the Interactive Engineering Diagrams feature from the web UI is not very helpful for me - i.e. I would still have to annotate manually the majority of the assets on the diagrams.

The entity matching API has the featureType parameter which helps with leading zeros, upper/lowercase differences etc. I was thinking that something similar would make the Interactive Engineering Diagrams also more robust.

 

Hope this helps,
Robert