Considering the following two questions, as examples:
- How Can I SELECT Similar Rows in Two Different Tables in MySQL (Is It Possible?)
- Matching similar city names in SQL
The OPs ask for possible techniques to match similar strings in SQL. In both cases, the majority of the answers seem to suggest Levenshtein Distance or Soundex functions to tackle this problem.
My question is, should we tag these two questions with the
[soundex] tags, even though the OPs were probably not aware of these two methods?
I think there are some obvious advantages in doing that, because it makes it very easy for someone who is aware of the methods to search for the
[mysql]+[levenshtein-distance] tags, for example. It also makes it much easier to find duplicates when similar questions are asked.
However, there are a few reasons why I am not sure about this:
It may imply that the OP was already aware of these solutions, and this may alter the context of the question. For example, answers like these may look out of point, even though that answer was probably very helpful to the OP.
It could also make the context of the question appear to be focused on the newly tagged categories, while in reality, the OP would have been open to any possible solution. This could deter someone from suggesting an alternative, possibly better, solution.
What if there were 5+ suggested solutions to tackle the "similar strings" problem, apart from Levenshtein and Soundex? Where do you stop adding tags?
A more subtle example of this "retagging behaviour" happens in the
[polymorphic-associations] tag on Stack Overflow. There are currently 95 questions with this tag, and I think the vast majority of them were not tagged like this by the OPs. In fact, many questions have answers that say something like "What you're doing is called Polymorphic Associations" (example). This looks like a valid categorization method and has its benefits, but does it alter the context of the question, since the OPs knew nothing about polymorphic associations before asking the question?
Started a bounty to attract any other possible opinions. Should we accept @Ladybug's or @Lance's stance on this issue? Or can you think of some balanced compromise between the two views?