downside of free-form Tags - Krioni

As the developer of TuneTags (here on VT), I am aware of some downsides to free-form tagging. The biggest one off the top of my head: synonyms. It is very easy to use two or three (or more!) different almost identical words as tags. Now, when you go to search for all items tagged "melancholy" you'll miss all the items tagged "sorrowful," "sad," "depressed," or whatever. If you don't a have a list somewhere of tags, you run into this problem. Even if you DO have a list, this problem pops up. Once the list is long enough, you can spend a lot of time checking the list to see whether you've already made a similar tag. There is a trade-off you have to make each time you tag between consistency (thus effectiveness of searching/viewing the "tag-cloud") and between quick tagging.

It seems to me that tagging has some issues. They could be rectified somewhat if the program helped you identify synonyms. I suppose that would be possible through some kind of automated thesaurus function that groups your tags together for you automatically. Then, when you search for "melancholy," the computer could show you related items. Coincidentally, this kind of semantic matching is something DEVONthink is supposed to excel at. The folder structure is merely the way to organize by general category. DEVONthink automatically analyzes the content of each record and indexes it. You can then ask DEVONthink for similar records based on the content. Seems to me this would accomplish a lot of what tagging does without ANY work on the user's part. Perhaps this kind of smart semantic analysis is better than either folders or tagging. Both of those methods require organizational work on the part of the user. DEVON's method lets the computer do most of the heavy-lifting.

Anyway, my point is that user-required organization involves work and has trade-offs, whether it is rigid folder hierarchy (even with replicants/aliases) or whether it is tagging (consistency versus speed). Find the balance that works best for you. I'll admit that tagging in DEVONthink would be nice. Ever thought of throwing your tags into the record's Comments field and then making Smart Folders? It would be possible to have a Folder that has a script attached which reads through every record's comments, creates Smart sub-folders within that folder for each tag. Hmm, maybe I'll do that - that could come in handy. :-) Even better would be something by the developer, of course.

Reply to This

Wednesday, February 28 2007 @ 01:30 PM PST

Replicants vs. Tags - epc

I agree with Krioni. One of the strengths do DT is the artificial intelligence. I chose to classify my data beforehand and I structured it in a way that made sense to me. But I could have dragged all of my documents into a DT database without any structure at all. I don’t think it’s ideal, but it’s possible and many users actually do that when they first start using DT.
You could add comments such as “recreation", "backpacking", "nevada", and "strenuous” and DT should be able to locate the relevant documents for you when you search for them (or use a Smart Folder).
I will try it as a learning experience with some of my data :-)

Here are two threads that deal with tagging ideas:
http://www.devon-technologies.com/phpBB2/viewtopic.php?t=2766&highlight=tagging

http://www.devon-technologies.com/phpBB2/viewtopic.php?t=4172&highlight=tag+comment

By the way, I did check out Yep some months ago. I think it was called Kip back then. Looked good. Part of the beauty of these programs is that we have many choices, and can find programs for our particular needs and preferences. Kip just wasn’t right for me.

In the end, the best program is the one that you use and that makes your work easier, whether it’s Kip or DT or Tinderbox or KIT :-)

Reply to This

Thursday, March 01 2007 @ 12:36 AM PST