|Think about Loose Coupling
Re^2: Finding Nearly Identical Setsby Limbic~Region (Chancellor)
|on Sep 28, 2016 at 17:33 UTC
I am going to answer some of your questions out of order (easiest to answer to hardest).
You already have a method of finding & extracting the number from the message; even if it is 8 or 10 digits.
What I'm getting at here is the insertions or deletions are easily detected, because the numbers are the wrong length. That only really leaves transpositions to detect?
Incorrect. While it is trivial to determine that a mistake has been made, it isn't necessarily trivial to determine what the mistake was (which set the number should have been).
You already have the set of known 9-digits numbers to compare against?
No. Think of it this way. The set represents what should be a unique identifier for the message sender. The first time I receive a message with a valid set (meets my aforementioned criterion) that doesn't appear to be similar to any other previously seen set - I will use that set to identify that sender. Not because it is correct but because it was first. A subsequent message from that sender may be objectively correct but for my purposes I don't know - I only know that it is nearly identical to what I have already seen.
If so, how are they currently stored?
I'm still figuring that out but because there will be parallel processing going on, I am assuming it will be in a database and finding good candidates will need to be 1 or more index based queries.Hopefully I have answered all of your questions satisfactorily enough to advance the problem but my cold medicine is really doing a number on my cognitive abilities.
Cheers - L~R