CODEWITHSUNDEEP
nlpcpu-word
James
James

Reputation: 819

natural language processing word association

My project needs some natural language processing. I'm completely new to the field.

What I am trying to get is when user enter a character, I want to get a list of English characters that can follow that specific character in order to make a legitimate word.

What is the specific term in NLP for doing this? I tried googling for a while, but had no luck since I don't know the term. Any good tutorials to start with? Are there any good libraries in doing this specific task?

Thank you.

Upvotes: 1

Views: 1711

Answers (2)

alvas
alvas

Reputation: 121992

Welcome to the NLP community.

The term you're looking for is most probably query prediction or sentence prediction. For example when you type some characters in google, it starts to prediction certain word/phrases that you might want to search for. And behind the technology, they used both (a) language based heuristics and also (b) user-based search history to train their model. They call it Google Instant, see http://www.google.com/insidesearch/

If you're looking for sentence/word prediction, then it's more like when you use the phone there is a function that helps you type faster, technically, it's call autocomplete (https://en.wikipedia.org/wiki/Autocomplete), see https://en.wikipedia.org/wiki/Autocomplete. The modern day autocomplete makes use of NLP to predict syntactically/semantically words that follows what you type.

As @evan says, markov chain is a method of learning patterns from your training data and then when testing your system, it can guess the word accordingly to whatever heuristics/statistical prediction that you have implemented. see https://en.wikipedia.org/wiki/Markov_chain

Upvotes: 1

Evan Knowles
Evan Knowles

Reputation: 7501

Have a look at language models, as well as letter frequencies. Markov Chains are often used for this kind of problem as well.

Upvotes: 1

Related Questions