@soandos I think your explanation works here: The edit was suggested BEFORE the auto-text was added. The suggested edit was updating a version before the auto-text. The resulting diff shows what changed, but in the suggested page, it is unaware of what happened between posts.
Like @Paul said, the auto-text should not be part of the editable post. The way it is now, it can be bad, especially when trying to find the original post which could have the answer you are looking for.
@soandos Yeah, sometimes I take a spin through some of the other tech SE sites to see if I can offer some more targetted help and they just seem riddled with low qualities and dupes. I see a question and go "well thats fairly easy to answer, surely it isn't new" and search and see the same question time and time again.
A bayesian algorithm can be used to categorise posts. For example, I use popfile, which watches the way that I file my emails in folders, and then begins to do it for me
We have a huge database of duplicates - I am wondering if we can use all that information as a predictor of some kind
But here we have a big database to seed it with - I think the problem would be that the number of unique questions outweighs the number of duplicates by a big margin
I think there are two "phases" to this - my feeling is that past duplicates are a predictor of future duplicates. Like we have "how do I reset admin password" is far more likely to come up as a duplicate in the future than something that has only ever been asked once before
So step 1 might be building a database of common dupes to identify them being duped again
Yeah - I'd say the common dupes are time-wasters - we know they are dupes, but we have to find a canon question to match them to. If the system did some of this work for us, then it wouldn't be nearly as frustrating.
@soandos Thats what I was getting at before - using the existing duplicates to find keywords or some other correlation to use as a predictor
But I am saying we can draw on what we already know. Like if you use the "reset admin password" as an example and find all the dupes for this type of question, then build up a set of keywords (with weighting), and then run those keywords against all the other questions to see how well they serve as a predictor
You give it all the dupes for a single question, then give it a bunch of non-dupes for a control and let it work out the probabilities, then give it everything and see if it correctly identifies the dupes
I have never written one but I understand they are relatively straightforward to code and use
@soandos Ultimately, any algorthim could work with both questions and answers, so if you got that far, there would be no reason not to run the answers through and see what came out
When a question gets closed as a possible duplicate, Community ♦ edits the post to insert the "duplicate link":
That's great! We should be directing users to the "original" question. The problem with this approach, however, is that this useful path between posts can be broken. Any user can...
@soandos Well, no. It's a text editor, not a log file viewer. If I didn't want to filter log files, I didn't need that feature, since search (and replace) suffices otherwise. But it's great how easy it is to add this feature, and well I can integrate it. Notepad++ cannot even sort/uniq lines that well integrated...
Stack Exchange sites have their own Twitter accounts, like http://twitter.com/StackUX. These accounts appear to just autotweet the latest interesting questions using the question's tags as hashtags. But there's a problem with this: Twitter is not an RSS feed that happens to have a 140 charac...
You can't parse [X]HTML with regex. Because HTML can't be parsed by regex. Regex is not a tool that can be used to correctly parse HTML. As I have answered in HTML-and-regex questions here so many times before, the use of regex will not allow you to consume HTML. Regular expressions are a tool th...
There are people that will tell you that the Earth is round (or perhaps that the Earth is an oblate spheroid, if they want to use strange words). They are lying.
There are people that will tell you that Regular Expressions shouldn't be recursive. They are limiting you. They need to subjugate yo...
When a question gets closed as a possible duplicate, Community ♦ edits the post to insert the "duplicate link":
That's great! We should be directing users to the "original" question. The problem with this approach, however, is that this useful path between posts can be broken. Any user can...
I would like to increase an exe size without altering it's behavior. It's possible to fill it with blank data?
Stack Exchange sites have their own Twitter accounts, like http://twitter.com/StackUX. These accounts appear to just autotweet the latest interesting questions using the question's tags as hashtags. But there's a problem with this: Twitter is not an RSS feed that happens to have a 140 charac...
