Help with database planning

[Rob Williscroft]

Juliano wrote in
news:0e64893a-af82-4004-bf3c-f397f2022846 at g22g2000prf.googlegroups.com
in comp.lang.python: 

[snip]

> So, for ONE *concept*, we have, usually, MANY *slots*, each *slot* has
> ONE *facet*, and each *facet* can have MORE THAN ONE *filler*.
> Besides, some *slots* and *fillers* are themselves *concepts*,
> creating a sort of recursive reference.

What I'm grokking from the data you show is that a "Concept" is
a table (or a python class), "slots" are attributes of the concept
and facets are the type of "slot", finally the "fillers" are 
(mostly) the data.

But also you seem to have "slots" that represent relationships
between "Concepts" and other tables, in this case the "slots"
should be tables.

For example with the slot type "INSTANCE-OF", what is 
"OTHER-NIGER-KORDOFANIAN-LANGUAGE" is it another concept or a 
list of concepts or a concept "OTHER-NIGER-KORDOFANIAN" and
a type "LANGUAGE", I suspect you have a table "Language"
and also an table "Region" in there too.

It may be time to start thinking in terms of what you are 
modeling and what the entities are (as apposed to trying to
convert the data you have), once you have that, work out 
how to load that data from your current file and check
that you can query your model correctly.

Maybe something like (using a made up ORM):

class Concept:
  name = TextField()
  definition = TextField()

  primary_key = PrimaryKey( name )

class IsASlot: 
  concept = ForeignKey( Concept )
  is_a = ForeignKey( Concept )

  primary_key = PrimaryKey( concept, is_a )

class LexeSlot:
  concept = ForeignKey( Concept )
  value = TextField()

  primary_key = PrimaryKey( concept, value )

class Region:
  region = TextField()  
  
  primary_key = PrimaryKey( region )

class Language:
  language = ForeignKey( Concept ) # or is it TextField() ?

  primary_key = PrimaryKey( language )

class LanguageOfSlot:
  language = ForeignKey( Language )
  region  = ForeignKey( Region )

  primary_key = PrimaryKey( language, region )


To reiterate, you should use your domain expertise to create
the model.

> 
> <begin table>
> line_no     concepts     slots     facets     fillers
> -----------------------------------------------------------------------
> ------- 00000     ABANDON     DEFINITION     VALUE     "to leave or
> desert something or someone"
> 00001     ABANDON     IS-A     VALUE     EXIT
> 00002     ABANDON     LEXE     MAP-LEX     "leave behind-V1"
> 00003     ABANDON     LEXE     MAP-LEX     abandon-V1

The "-V1" in the above looks worryingly like you have structure embedded
in the data field, if so you should extract is so its in its own field 
or table.

> (...)
> 97420     ZULU     DEFINITION     VALUE     "a language or dialect
> spoken in south africa and others"
> 97421     ZULU     INSTANCE-OF     VALUE    
>                                OTHER-NIGER-KORDOFANIAN-LANGUAGE 
> 97422     ZULU     LANGUAGE-OF     INV     LESOTHO
> 97423     ZULU     LANGUAGE-OF     INV     SOUTH-AFRICA 

Rob.