Rainer Joswig wrote:
One thing that always makes me curious is, for what do people
need PERL for in the first place? I never really had the
need of PERL, regular expressions and the like (I have the
need for Persistent/Distributed CLOS, GUI Libs, and stuff
like that). Isn't the need for PERL a sign of the "wrong"
approach (manipulating text files instead of working with
objects)?
----
I took a course in statistical language procesing last term.
This was *not* a NLP course (in the AI sense), but rather had
to do with statistical analysis and mining of large corpora.
I initially thought of doing my work in Common Lisp (in
which I have more than beginner-level skill but certainly
not advanced-level skill) but found perl to be quite good
for it (mainly because of very good regular expression
manipulation facilities). The domain was all text, naturally,
so perl seemed a natural choice...actually everyone in
the class pretty much ended up using perl rather than
some other language. But that may have been a herd effect :-)
There was also the `network effect' -- a lot of stuff that
you use for this kind of work -- useful tools like part-of-
speech taggers, interfaces to the Wordnet semantic network,
etc, already exist in perl.
I am currently thinking of rewriting the stuff that I did,
in Common Lisp (mostly as a learning experience). However,
certain interfaces to existing software will have to be
in perl because of the reasons stated above. So, any tips
about integrating perl and lisp? How difficult is it to
integrate the two within the same application, and what
do I need to know/watch out for? (I plan to be using
Allegro Common Lisp running under Solaris).