On Wednesday, December 12, 2018 at 5:54:41 PM UTC+1, Joe Armstrong wrote: ...
> What I did was to use Baysian inference to "learn" the relationship > between the words in the text and the supplied tags - so for each word in > the text I caculate the probability that the tiddler has tag <T> (forall > known tags <T>) - then in a second pass I tested the model and predicted > the tags from the text. This way I could correctly predict about 80% of the > tags from the text alone. The problem was that, to me, many of the tags > were meaningless and were used internally to organise the TW. > That's interesting. ... But I think this has some "evolutionary" causes. I think, it hasn't always been that way. Open classic.tiddlywiki.com ... You'll see wikipedia-like tag-box in every tiddler. ... The UI isn't "nice" with tags here. .. So a very common question in the group was: "How can I switch this box off?" I personally prefer something that's called "TagglyTagging" (... oh we love those weird names :) I think TagglyTagging was introduced with MPTW <http://mptw.tiddlyspot.com/#TagglyTagging> (Monkey Pirate TiddlyWiki). For me it was a completely new way to work with tiddler titles. ... TT is a set of plugins, that allows you to visualize the relation between different tiddlers ... It speeds up navigation between related tiddlers, in a very convenient way. ... The "sitemap" view is what we call TOC (Table of Content) in TW5 now. An other plugin, that imo influenced TW5 was: the fET-plugin (for each tiddler) <https://tiddlywiki.abego-software.de/#ForEachTiddlerMacro>. It allows *users *to iterate over the tiddler store and *create *many types of "*list-views*". This plugin was highly influential for the TW5 list-widget, and <<list-links ...>> macros that we have today. Both of those systems (mis)use tags to create internal structure, because the tagging mechanism *was and is* highly optimized. Both in the core-software and the UI. The core uses several caches to speed up tag and "backlink" lookups. ... We do have fields and filters, that are able to create invisible internal structure. But none of those possibilities offer the performance and "ease of use" from the UI perspective. In my opinion the TF*IDF were better than the assigned tags since they had > nothing > to do with the organisation, but more to do with the actual words in the > text. > For me it would be very interesting to have a mechanism, that would suggest "meaningful tags" by analysing the prose text(s). ... but it needs to work without a 3rd party server. It should be integrated into TW. ... Is this possible? have fun! mario -- You received this message because you are subscribed to the Google Groups "TiddlyWiki" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/tiddlywiki. To view this discussion on the web visit https://groups.google.com/d/msgid/tiddlywiki/90d0b5a8-88b2-4489-ac3d-ed724f08b01c%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
