How do you handle "wikification" of words?

40 views
Skip to first unread message

Amirouche Boubekki

unread,
Feb 26, 2017, 9:07:48 AM2/26/17
to opencog, linasv...@gmail.com
Héllo,


I read that wikification is an important process of getting computers to understand what is written.

wikification aims to retrieving the concept vertex from a given piece text of text.

E.g:

apple is http://conceptnet.io/c/en/apple

Bob marley is http://conceptnet.io/c/en/bob_marley

How is this handled in opencog?


Best regards,


Amirouche aka. amz3

Amirouche Boubekki

unread,
Mar 6, 2017, 4:55:04 PM3/6/17
to opencog, linasv...@gmail.com
I made a small program that does entity link free text into conceptnet.io

The program is here: https://framagit.org/a-guile-mind/culturia.one/blob/master/src/conceptnetization-02.scm

But the interesting part is here: https://framagit.org/a-guile-mind/culturia.one/blob/master/src/nlp.scm

all-partition procedure does generate all slices of a given list (or list of words (or sentence)). It doesn't unk the text, that's a bug.

Kind of off-topic, I made a naive sat solver https://framagit.org/a-guile-mind/culturia.one/blob/guildhall/src/sat.scm



--
You received this message because you are subscribed to the Google Groups "opencog" group.
To unsubscribe from this group and stop receiving emails from it, send an email to opencog+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
Visit this group at https://groups.google.com/group/opencog.
To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/df1796d5-884e-4104-b7d4-e013f1d8bf96%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Amirouche Boubekki

unread,
Mar 6, 2017, 5:00:02 PM3/6/17
to opencog, linasv...@gmail.com
On Mon, Mar 6, 2017 at 10:55 PM Amirouche Boubekki <amirouche...@gmail.com> wrote:

all-partition procedure does generate all slices of a given list (or list of words (or sentence)). It doesn't unk the text, that's a bug.

I mean, you receive a '() when the word is not recognized instead of the its unique identifier. It doesn't try to recognize them harder using some spell checker algorithm or something. It's a direct equal? of what's in the concept/token field.

Now I need to relate that information with what Link grammar or RelEx parse. I am considering rewriting RelEx in Guile re-using its dictionary. Do you think it's possible?

Regards,

Amirouche ~ amz3

Amirouche Boubekki

unread,
Mar 6, 2017, 5:33:50 PM3/6/17
to opencog, linasv...@gmail.com
I did not finish reading the documentation @ http://wiki.opencog.org/w/RelEx_Dependency_Relationship_Extractor
Reply all
Reply to author
Forward
0 new messages