Help with Sentiment Analysis

[uptightprogrammer]

Yo! I have some questions regarding Natural Language Processing, sentiment analysis to be precise. So if someone could help me out, it'd be great! 

 

I have done some reading and I want to implement an LSTM with pre-trained word embeddings (I also have plans to create my word embeddings, but lets cross that bridge when we come to it). I have a few questions and possible more depending on the answers. 

 

1. What is the general pipeline of the whole thing? It would be easier for me if you could take an example of a sentence and run through the whole thing. 

2. Are word embeddings features?

3. In any given sentence, you usually don't need to have all the words as most of them do not contribute to the sentiment such as stopwords and noise. So let's say there is a sentence. I remove the stopwords and anything else that I deem unnecessary for the project. And then I run the remaining words through the word embedding algorithm to get the word vectors. Then what? How does it represent the sequence or the sentence 'cause it's just vector for a word. 

 

For example, take the sentence: "The burger does not taste good". I could remove certain words and still retain the same sentiment like so: "Burger not good". 

Let's assume some arbitrary vectors for those three words: 

 

Burger: [0.45, -0.78, .., 1.2]

not:       [9.6, 4.0, .., 5.6]

good:    [3.5, 0.51, 0.8]

 

So those vectors represent the individual words. How do I make a sentence out of 'em? Just concatenate them? Sorry if I'm being naïve but I really have no clue. 

 

If someone could get back to me and let me know, I'd be very grateful.