Extension of symbol mechanism

[andreasbaumann]

In the following example:

WORD ^1 : /\b\p{L}+\b/;

KnownName = sequence_imm( firstName = "Hans" WORD, lastName = WORD );

Let's assume we have a list of known first names in a dictionary, so how can we make
sure we match exactly those as symbols in firstName?

The two options I can come up with, don't look too nice:

WORD ^1 : /\b\p{L}+\b/;

KnownName = sequence_imm( firstName = "Albert" WORD, lastName = WORD );
KnownName = sequence_imm( firstName = "Hans" WORD, lastName = WORD );
...
KnownName = sequence_imm( firstName = "Werner" WORD, lastName = WORD );

or even worse:

WORD ^1 : /\b\p{L}+\b/;
FIRST : /\b((Albert)|(Hans)|(Werner))\b/;

KnownName = sequence_imm( firstName = FIRST, lastName = WORD );

Allowing an API there and providing a dictionary for instance allows to filter for known
items.

I could also imagine an interface to a "SELECT first FROM known_customers" into a
database. Of course the question is, how far things should be done in the CLIs like
strusPatternMatcher and when you should use the APIs to do the filtering as a
subcomponent along to database and other processing.

[patrickfrey]

A database could be used in a post-filtering of candidates selected by pattern matching.
To query a database in the pattern matcher is not a real option.

The solution to the first problem could be a set of rules for the prename.

KnownFirstName = any( WORD "Albert",
WORD "Hans",
....
WORD "Werner",
WORD "Zora" );
KnownName = sequence_imm(
firstName = KnownFirstName,
lastName = WORD );