Using Intelligent Agent to Identify Incorrect Matches

What Wildbook should this feature be in?

Whale Sharks

What would you like to see?

I was returning to a conversation that I had with Jason Holmberg and Chris Rohner a while back to do with whale sharks at Ningaloo Reef, and the high probability of false negative / false positive matches in a large dataset with multiple processors. Would it be possible to implement a ‘validate’ function flagging potential errors for manual review and confirmation?

How would this functionality help you?

It would be excellent to clean up some large databases for use in mark-recapture modelling.

Thanks for the consideration,


We in the African Carnivore Wildbook community wholeheartedly support this feature request!!!

This is a great idea Simon! I think it’s relevant to multiple localities/species :slight_smile:

1 Like

@simonjpierce & @gonzo_araujo - hi there, I believe this is similar to or the same as a feature request I submitted last summer: Ability to mark a match result as "not a match"

Do you agree? Or am I misunderstanding your post here?


Hey all,
Thanks for posting and commenting! It’s great to see different communities of researchers agreeing on a common need. Even better when it’s a common need we’re actively working towards meeting!

One of the major developments we’re working to achieve this year is the finalization and release of a curation tool called LCA. This tool is intended to review platform data holistically and review for issues of misidentification. Instead of prompting for singular problems, it works to correct from the ground up, working through merges and splits and recategorization of individuals to create the most accurate data set possible.

@ACWadmin1 : I consider this and the feature request you linked to be distinct because of how we are implementing LCA. The ability to mark “not a match” is something we want to allow users to do while reviewing potential match results. We want to use intelligent agent to look at everything and determine right-and-wrong, not just dismiss not-matches.