TMQL: Minor issue: Plurals for paths

>> My problem with the current approach: Some think that Toma works
>> (it does not) and some think that TMAPI works (it does not).

> This sounds more harsh than it should. I meant: Translating TMDM
> literally into a query language does not work (IMO). It works, but it
> is not necessarily logical (since TMDM has no logical foundation) and
> results into a verbose language.

As a developer of one toma implementation, having gained some experience
in that process, I would go even a bit further:

Every attempt to create a semantic data query language that tries to
resemble the syntax or structure of sql is flawed by design.

Of course it works in some way or another, but it does not feel natural
or logical, simply for the reason that there is a different rationale
behind the data model they operate on:

 - sql operates on _relational_ data models
   data elements are put explicitly into relation using joins

 - a semantic query language operates on to _graph_ structures
   data elements are already implicitly in relation with each other

While a sql-based query structure for a semantic data model would allow
meaningless joins of data that does not relate to each other (and the
implementation has to take care it works in some way), it fails to
properly represent graph structures that are present in your model, and
would be of interest.

Simply spoken, who needs a query like that:

  select $p, $a where $t isa person and $a isa animal

while on the other hand, information that is already internally
represented in the model as a graph, that I want to query as a graph
(using path expression), and that I want to access as a graph, is
transformed into a sql-like resultset?

One, of course, can discuss a lot about the query syntax in general, but
the way MQL (http://mql.freebaseapps.com) tries to approach the whole
problem is at least very appealing to me and should be the way to go
forward, rather than trying to jam something into a structure that just
does not fit.

On a second thought, having a simple query language with some elegant
path expressions that resembles somehow sql would maybe not a revolution
in the semantic world, but would fit the envisioned markets and current
community of topic maps. The rest of the semantic world goes somewhere
else, but who cares?

just my 2c on this topic

Thomas