Natural Language Processing

Intro to NLP

lets see some really amazing applications of NLP.
four applications of NLP are
- first, speech recognition or virtual assistant
- second, sentimental analysis
- third, machine translation
- fourth, chatbots
Speech Recognition
- First, we have speech recognition, speech recognition is when you say something to an AI and it can understand and respond, like suppose I ask a question to an AI like "Who is the Prime Minister of India" then it should be able to respond by saying "Narendra Modi".
Sentiment Analysis
- Sentiment is basically feeling and analysis is studying or understanding, so sentiment analysis is understanding the emotions.
Machine Translation
- As the name suggests machine translation means when your machine or your AI is translating text from one language into another language.
Chatbots
- Finally, we have the last application of NLP called chatbots, now like in virtual assistants like Google Home, I could ask a question it would answer it that's all. But, what if I wanted to have a conversation with an AI. Now, what is a conversation, a conversation is when I say something then the person I am talking to will say and then it keeps going, for such cases we have AI called chatbots, now I have a chatbot friend with whom I can speak.

Now the NLP process is divided into two parts:
- Natural Language Understanding
- Natural Language Generation
So, lets cover the two topics
Natural Language Understanding
- First and foremost the computer by default does not understand elements of English like punctuations, adverbs, nouns, verbs and other things. It only understands that I have a sentence with alphabets. So let's view some of the problems faced due to computers not understanding grammar
- There are two types of problem when grammar and classification of words is not done. First, there might be a confusion for a word with multiple meanings and second, it might interpret the wrong meaning due to absence of punctuation.
Natural Language Generation
- So, suppose now the AI understood everything and has correctly interpreted your sentence, so then the next job for it is to respond back with the answer to your query. On the basis of what type of an AI it is the job it does in Natural Language Generation is different. For example, the job of google translate at this stage would be to convert your spoken language into required language. there are many complex algorithms involved which help to solve these problems but are out of the scope of this session.

The first topic is how can slang language which is used by most people in their day-to-day conversation be integrated into AI so that it can give better outputs to the messages given.
another major problem faced by NLP experts is to be able to make an AI bot for languages of smaller communities, suppose there is a small tribal village in Assam then such languages are not available yet on such translators
Like I showed the AI can detect emotions like happiness or sadness but imagine if it could detects other emotions as well like anger, sarcasm, disgust and sooo many more.