[Top] [All Lists]

Re: [ontolog-forum] A flow chart with open-source NLP tools

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Mon, 29 Sep 2014 11:14:16 -0400
Message-id: <54297748.8020107@xxxxxxxxxxx>
Michael, Kingsley, and Dick,    (01)

In this morning's email, I found an announcement that illustrates
the difficulties of analyzing and untangling ambiguities in NLs.
To see the complexity of the issues, browse through the abstracts
about detecting "Evidentiality and Modality" in documents:    (02)

http://www.ucm.es/data/cont/docs/759-2014-09-16-EMEL14-BOOK_OF_ABSTRACTS.pdf    (03)

> What about neural networks and feature vectors / word embeddings?    (04)

I agree that there are many other tools that should be included.
A one-slide flow chart is useful.  But when you dig deeper, it
becomes so complex that a different format would be better.    (05)

>> 1. The kind of NLP that the URI points to happens to be a video about
>>    Neuro-Linguistic Programming, not Natural Language Processing.    (06)

> The HTTP URI <https://twitter.com/hashtag/NLP#this>, when placed in the
> address bar of any browser simply resolves to an HTML document that
> lists a collection of Tweets.    (07)

Yes, but acronyms are notoriously ambiguous.  Just type any
acronym to Wikipedia to find amusing examples.    (08)

I have found that outside AI and related areas, 'NLP' refers to
that bunch of quacks that claim to be doing neuro-linguistics.
They have a lot more money for TV ads.    (09)

> http://linkeddata.uriburner.com/c/9C3K2GJR -- a description of what
> <https://twitter.com/hashtag/NLP> actually identifies (or names)
> modulo owl:sameAs reasoning and inference    (010)

I agree that your URIBurner can be useful for untangling such tags.
But I have a request and a suggestion about that page:    (011)

Request:  In the following quotation, please replace "that NLP" with
"Neuro-linguistic programming" to make it clear what I'm criticizing.
And there's no need to mention my name, just quote Wikipedia.    (012)

> (014) As Wikipedia says about that NLP, 'The balance of scientific
> evidence reveals NLP to be a largely discredited pseudoscience' --
> John F. Sowa.    (013)

The two kinds of NLP are totally unrelated.  The following two entries
imply that there is some relationship:    (014)

> sameAs Natural language processing
> has related Neuro-linguistic programming    (015)

On LinkeIn, NLP usually means natural language processing.  But on
Facebook, it usually means quackery.  They are disjoint.    (016)

Suggestion:  The term 'computational linguistics' is a synonym
for 'natural language processing'.  You might add more to the
URIBurner page to say that.    (017)

> I am using log(N) as the measure of ambiguity for a concept,    (018)

For some purposes, that could be useful.  But no two dictionaries
list the same number of senses for any word.  For acronyms, there
is no relationship among the full forms.    (019)

> The typical ontology hierarchy is a lattice which includes all N
> definitions of a concept.  For instance, in John's recent email,
> NLP has two meanings, and a[NLP] := 1 bit.    (020)

I like lattices for precisely specified ontologies.  But it's more
confusing than helpful to mix the unanalyzed strings with the concepts
they represent.  See, for example, the many issues in that collection
of abstracts about Evidentiality and Modality.    (021)

John    (022)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (023)

<Prev in Thread] Current Thread [Next in Thread>