Forums : Technical Issue Help

what is the difference between these tags:

• mathematics, math and maths
• script and scripting
• interpreter and interpreters
• science and scientific

should they not be merged?

Sam Steingold over 16 years ago

Yes, it would be awesome if the collective wisdom of Ohloh would come together and finally decide if we're going to go with chart, charts, charting, graph, or graphing.

What ends up happening instead is that everyone adds as many synonyms as they can to their project, in a sort of tag arms race.

Occasionally we go through and use some scripts to help prune the tags when things get out of control, but mostly we let people decide for themselves.

Given that our tag feature isn't really used for a whole lot besides searching (so far), it doesn't hurt much to allow a lot of tags (in fact, when it comes to searching, the extra tags might actually help).

I'm not sure what the best long-term solution is. Perhaps Ohloh will create a fixed list of tags/categories, and only those tags will be allowed. Perhaps Ohloh just needs additional UI hinting to encourage people not to use unusual or redundant tags.

Thoughts?

Robin Luckey over 16 years ago

the more the merrier is an illusion.
many tags obscure the view and confuse the searcher.

there should be a canonical name for each tag, with plenty of synonyms and a brief description.
e.g. interpreter - programs that run scripts.
interpreted - written in an interpreted language.

tag searches should match both tag names and synonyms - and, optionally, the description too.

adding a tag or a tag synonym should be semi-automatic: one requests a tag or a synonym by filling out a form, which then collects votes and comments, and if the vote passes and there are no conflicts with existing tags and synonyms, the tag is created.

Sam Steingold over 16 years ago

ohloh tags are an example of a Folksonomy which has deep problems, but is participatory in the same way that open source is.

What you are describing is a Controlled vocabulary which also have deep issues, but require specific in-depth training.

Controlled vocabularies are formal systems and can be shown to be incomplete in the Turing sense. Folksonomies are interpretative systems.

Stuart Yeates over 16 years ago

How much of an impact would it be to add total count of tags in the tag addition page? E.g. di - 5, dependencyinjection - 10, dependency-injection - 15, dependency_injection - 20 (numbers are made up of course). Or display tags with font-size proportional to usage, a-la tag cloud. Hopefully that would help people choose the most popular of the same tags, and the unpopular synonyms will just die off.

Also, I personally would prefer that people prefer non-abbreviated tags and use consistent character for word separation (dependency_injection vs di), but I'm sure others would disagree; maybe some sort of a policy from Ohloh could help.

Sergey Lipnevich over 14 years ago

Not sure what the forum protocols are on pinging old posts, but this is still a problem. There are often sets of tags that have exactly the same meaning, separated only by differing punctuation (e.g. static_anaysis, static+analysis).

You can have a folksonomy with synonyms, and a folksonomy with canonical synonyms aren't controlled vocabularies, as they still allow the creation and use of non-pre-defined terms. They can still be used as a controlled vocabulary, if there is a good reason to do that. Keeping the non-canonical synonyms allows a full folksonomy, and an API could allow the optional exclusion or inclusion of synonyms, or non-canonical terms. I you wanted to be really nifty, you could allow users to suggest synonyms, and use the number of suggestions per pair as a proxy for synonym strength, which would allow for more fine-grained calculation of things like project similarity.

BTW, there is a similar discussion at https://www.ohloh.net/forums/3491/topics/955

naught101 over 11 years ago