I am looking for UX interface patterns that can be used to dedupe records.

Suppose, for example, one has a database of customers. Perhaps Ronald T. Wiggins and Ronald Wiggins are both in the database and have the same address. Users would like to be able to merge those two records into one canonical record.

One pattern that facilitates record merging is provided by OpenRefine, the open source project by Google for data cleaning. We can see below various "clusters" where each cluster contains items that a user can merge into a single canonical record:

enter image description here

In my PhD research I was working on identifying visual copyright infringement, and rigged up a basic UI that let me identify whether or not images match and should be clustered:

enter image description here

Has anyone encountered interfaces that facilitate this kind of record merging? Any examples others can offer would be hugely helpful!

  • 1
    Are there multiple key-distance functions in your application as well? What parameters does the user need to input/manipulate to see the possible matches?
    – ikartik90
    Commented Oct 5, 2021 at 16:22
  • I think we can say that any two objects have one unified distance metric / similarity score. By default I think only matches k% similar or greater are displayed, and k is a range slider. But I'm just looking more for examples of this kind of interface in the wild so I can see what kinds of parameters would make sense to include...
    – duhaime
    Commented Oct 5, 2021 at 18:14
  • While this might already be accounted for in the match percentage, should two images that are of the same object but with different tones, say coloured vs. grayscale, be considered a match or not? Also, should the user be allowed to caliberate this, and other similar parameters such as hue, saturation etc., among the matching criteria?
    – ikartik90
    Commented Oct 5, 2021 at 18:44
  • 1
    Hmm, I'm afraid I wasn't clear enough--I am not building an interface for surfacing similar images. I'm just researching design pattens for data duplication more generally :)
    – duhaime
    Commented Oct 5, 2021 at 19:08
  • Hmm, if we already had a solution for identifying different nature of things, we would not drive a car, an algorithm would drive a car. Yes, there is already Tesla electromobiles but as I know they are still learning to recognise objects and not ready to drive everywhere.
    – Vitalii K.
    Commented Oct 8, 2021 at 18:11

1 Answer 1


Most commonly seen example I can think of is the Highlight Duplicate feature in Microsoft Excel where you can perform a number of operations on duplicate values depending on how you want to process the information.

enter image description here

enter image description here

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.