Usability problem in the widespread 'is a' approach

As introduced earlier, taxonomies as thesauri and classifications organize terms in a net similar to a tree. Hierarchically, the terms get ordered mainly by is a relationships, seldom by has a ones. Also there are relationships between terms that have "something" to do with each other, but cannot be ordered validly hierarchically. Like bird and bird cage. Some taxonomies force one parent term per node, i.e. a structure that easily can be identified as being mainly a tree, with exception of some cross reference like associative relationships. Other taxonomies don't force the one parent per term rule, so that such a one might easily look more like a net but like a tree. The prior ones are called "mono-hierarchical", while the latters' label is "poly-hierarchical".

What's the problem?

Taxonomies don't do anything more but relating the terms to other terms. The task to define the terms they leave to dictionaries.

This presupposition requires that the one who's using a taxonomy already is a kind of an expert in the field organized by the taxonomy. If you're looking for a term you don't know where to look up, you're lost.

Say, some part fell from your car, and since then it doesn't move anymore. You don't know the term for that part, and the very one is too heavy, so you cannot just take it to the garage. -- Because of its primary is a nature, a taxonomy is of no use here for you. Another chance is to go to the garage without the part and attempt to explain the nature of it to a worker there. (So far, my university information science teacher guided me.)

So, the problem of a taxonomy is that it doesn't support the most straight ahead approach to identify an item -- to select the most conspicuous properties of the item the taxonomy -- broader: knowledge storage, e.g. a garage worker's memory -- already knows about.<<

---

Updates: 20070624: Tagged the posting. Updated the posting style (layout) to my current style, such as using blockquotes when appropriate, more precise word picks, better grammar.

[Merged from (the now removed) ia: organizing notions:] If Google didn't get the idea it's just standing in front of understanding it. Using labels in GMail is a nice thing, but it results in many, many individual labels.

To apply them to mails will get quite inconveniently. Currently Google's using hot keys inside its gmail service, meaning that pressing a 'c' (for example) leads me to the "create new mail" screen. This intereferes a bit with the convenient use of a drop down box where you normally can use the first letter of an entry to navigate in the list.

To bypass that, someone could request for cascaded/stacked drop down lists, resulting in a kind of menu. There, for example, one could place all labels of interest within a bigger thread -- a mailing list for example. - But that's something Google cannot do since implementing such a concept would lead back to the monohierarchical folders concept. Therefore they have to find a way to keep such kind of menu polyhierarchically.

*snap* That's it. Then they'll get the idea.

The idea is, that a lot of things (but not all) can be identified by cascading labels -- or, further thought, features (that results in independence of language). But it has to be possible to order these features in a polyhierachical manner still. Then one can stack ideas. The name of a mailing list (or the "idea" of that mailing list) could be a feature of an individual mail sent by that list. Another feature of that mail could be the sender, cc recipients and so on.

Another case applies to the conversations, gmail is using instead of folders. One single mail can be part of different conversations - for example if one reader starts a new thread based on a mail that continues to be part of an earlier started conversation. Same problem as for the labels. -- I think, Google almost must get the idea.
 

Next obstacle: How to let gmail users access their gmail accounts by POP3 based mail readers? - One possible solution is to introduce a normally invisible field in the mail header, that contains gmail labels. Other freemail providers can adopt that principle -- then it may be used by a wide range of users. -- Maybe a possibility to fund on this to create a notion based search engine? -- I have thought about that some years, yet, but I've never had the time to really implement it. The few months between my last test and the start of research for my diploma thesis I used to just automatically generate an appropiate MySQL database. But after that I had to stop working on it, since making my diploma has a higher priority.

I'd like to hear from anyone who would be interested in implementing such an engine as an free software project with me together.<<

---

Updates: 20070624: Tagged the posting. Updated the posting style (layout) to my current style, such as using blockquotes when appropriate, more precise word picks, better grammar. Removed my workaround for backlinks blogger.com didn't support in earlier times. Now, backlings are there, therefore the bypass can be dropped.