Natural language processing (NLP) has become an integral part of our lives with applications like Siri taking the lead. Natural language processing programs take input from users as sentences, and try to deduce what the person meant. For us, the ones who actually speak this language, this seems quite easy; however it is not so for computers, they have to cycle through each and every meaning of the words we used during the sentence, and try to logically link each word to the other, and finally give us back what we actually meant, and while doing this, they encounter many (and more) possibilities of what it could have meant, things which come naturally to us, so how are they supposed to know what we really meant? In this day and age, people have managed to advance these types of programs, with the most recent ones being Desti, Siri’s sister, however Desti has advanced further than Siri, which can only answer slightly above basic questions, and now can even help with questions pertaining to travel, which may include where the nearest museum is, or where one can find a restaurant where they don’t or do sell certain types of food, and so much more. Facebook has also used a natural language processing program to power its Graph Search, which basically takes input from the user, and then it thinks for itself what it means, checks with the user, and then produces results based on that, making the use of Facebook ever more easy and convenient for its users. However, it does suffer one limitation, this Graph Search is only limited to English, and they are trying to develop it for more languages. But what is it that makes NLP so difficult? And why is it so easy for us? Probably because NLP requires a good degree of AI in itself, which will be able to learn and based on that produce more accurate results, which we humans do naturally, but why is that? We humans have already accumulated a significant amount of experience just by living, and we have subconsciously deduced what some sentence means in most cases, however that is it, in most cases, because with all the different context a sentence might have, and all the different ones each and every human might use, even humans, let alone machines, will commit an error at a certain part of their lives.
Questions:
Is a perfect NLP possible?
Can a translating NLP be made?
What are the limitations of using NLP?
. http://gigaom.com/2013/04/29/facebook-relies-on-natural-language-processing-to-power-graph-search/