[NLP] Detecting verbs

What's a simple algorithm (simple == no academia robodick sucking neural network bullshit) to pretty consistently find a verb in a given sentence? I wish to create a bot. I'm thinking of using a dictionary, I just don't know how to work out the details.

the best way of finding verbs is probably by having a search tree or hash set based on a dictinoary of verbs. the best way of finding a predicate would be automatically constructing a sentence diagram based on a given sentence, which is significantly more complex - but there are libraries for that (see https://www.abisource.com/projects/link-grammar/)

http://ai.neocities.org/EnParser.html

http://ai.neocities.org/EnVerbPhrase.html

http://ai.neocities.org/RuParser.html

http://ai.neocities.org/RuVerbPhrase.html

>>3
Mentifex, your codans will be used to shitpoast.

>>3
How do I integrate? Do you have an API?

>>5

Do you have an API?

There is no Application Programming Interface (API).

>>6
Unusable.
How do I integrate then? Is there anything that's not a textbox on a geocities web site?

>>7

geocities web site

Neocities

Finding a verb in a sentence means understanding all the rest of the sentence. Dictionaries alone won't do it for you. You have the AI-hard problem of

http://en.wikipedia.org/wiki/Natural_language_understanding

which has been solved by ghost.pl in Perl, MindForth in Forth, and

http://ai.neocities.org/FirstWorkingAGI.html -- not a "text box" -- in JavaScript for Internet Explorer.

>>8

Finding a verb in a sentence means understanding all the rest of the sentence

that's not true, MentifAnus. as I said in >>2, finding a predicate means understanding the structure of the sentence, which really boils down to parsing grammar, while finding a verb requires knowing which words are verbs, which is just a dictionary lookup. the latter solution will not understand neologisms but the former will still do.

now, finding a verb or predicate in an unknown language is something that only AI could do, but I doubt your're are chatbot will be able to do that

1. get a list of all verbs
2. tokenize/lemmatize each word (e.g., "runs" --> "run")
3. put each word in a search tree
4. test new verbs by tokenizing and searching
optionally autocorrect typos this way: http://norvig.com/spell-correct.html

This is 100x faster than any neural robodick algorithms and probably works better.

[NLP] Detecting dubs

>>10
Thankx my friend, I will use it!

>>11

[NLP] Detecting dubs

Now that's a new one, LOL. Can anybuddy beat it?

>>13
your're are an anus

[NLP] Detecting verbs
What's a simple algorithm (simple == no academia robodick sucking neural network bullshit) to pretty consistently find a verb in a given sentence? I wish to create a bot. I'm thinking of using a dictionary, I just don't know how to work out the details.

>>15
Verbs were indeed detected, good job,