Under review

NLP: how the matching works? is it possible to use somekind of wildcard ? or using something like contain:father

adez3d@gmail.com 3 years ago updated 3 years ago 4

Hi there. I've been doing some experimentation with NLP. 

I'm using Indonesian not english and still figure out.

In Indonesian, let's say when put something like "i love you" it doesn't seem to understand if user say : I love you and your family. 

Is it possible for NLP to process user input, let's say something contain work like "love", family, or any type of feeling like anger, sad, loss, sleep ?

I was developing chatbot using AIML (Artificial Intelligence Markup Language) version 1.0 for quite sometimes, which basically a simple matching pattern. 

So in AIML I can use wildcard, like

I love you * > that can parse sentence like: "I love you and the rest of your family."

or I love * > that can understand word like: "I love you, but I feel that I will loose you..etc..etc"

or event * love * that can parse sentence like: "do you know that my mother love you, but she has other things to consider"


1. Is it possible to have something like above? or at least find a way to set NLP understanding sentence that contain certain word and not exact match? like love, or any other word like father, mother, etc. 

2. Some user tend to type a very long sentence, or event a long paragraph/ this is where any NLP or Machine Learning seems to breakdown. How to make sure I can parse a sentence that contain certain word. 

Let say in case of Eliza the "psychology chatbot"...there are some word that parse certain word.

user: I feel neglected by my father

bot: tell me about your father (after parsing word "father")


Under review

Welcome to Sequel!

The NLP works in such a way that if you use the sentence "I love you", it will detect if user types a sentence which contains the NLP tag "I love you". For e.g. if user types " I love you a lot dear", then it will detect as "I love you" is present in user's input.

Hi Praveen. thanks for the reply

I already tested many times, 

Like you said: if user types " I love you a lot dear", then it will detect as "I love you" is present in user's input.

But if your said a very long paragraph, tend the bot won't recognize the sentence "i love you' in the middle of other sentence. and go into fallback mode.

Oh, and my case it's in "Indonesian" language,

so it might behave differently

We'll look into this further to optimize the NLP.