How are email sentiments tracked?

Vtiger CRM associates an emoticon to each customer interaction that takes place via email. The system generates an emoticon, based on the email content.

Natural Language Processing (NLP) is used to detect email sentiments. Here is how:
1. Sentiment values are assigned to words first, and then to sentences.
2. The average sentiment value of all the sentences is the final sentiment of the email.

For example,
Sentence 1: I did not like the GP110 Wireless Bluetooth Headphone – Negative (Value=1)
Sentence 2: I would like to return it – Negative (Value=1)
Sentence 3: Also, I am disappointed with your customer service rep’s behavior – Negative (Value=1)
The average sentiment for the email is Negative.