Second March Post

Today, I continued learning chatbot, but mainly focusing on how to distinguish customer's feelings. When human readers approach a text, we use our understanding of the emotional intent of words to infer whether a section of text is positive or negative, or perhaps characterized by some other more nuanced emotion like surprise or disgust. It is important for the company to know what each customer is feeling when they ask questions on chatbot. It can also be applied to any industries such as movie companies, where they want to know feedbacks from audience by checking twitter or facebook and analyzing comments. If customers are using words that are considered to be negative, it is important for the company to know what is going on that makes them feel in certain way.

In order to analyze questions/comments, I learned how to code in R following steps on the website below.

https://www.tidytextmining.com/sentiment.html#sentiment-analysis-with-inner-join

There are a variety of methods and dictionaries that exist for evaluating the opinion or emotion in text.

1. AFINN lexicon: assigns words with a score that runs between -5 and 5, with negative scores indicating negative sentiment and positive scores indicating positive sentiment.

2. bing lexicon: categorizes words in a binary fashion into positive and negative categories.

3. nrc lexicon: categorizes words in a binary fashion (“yes”/“no”) into categories of positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust.

Here is the steps.

I used Jane Austen's books as examples. There is a list of each word appear in several books.

Based on the data I get, I can plot like this,

Here, you can see how each lexicon provides different result.

It is possible to count words that contain emotions.

This is another way to show which word appears the most in the book by varying the size of the words according to the frequency.

This analysis can be used for many different purposes. Although it is useful to know customers' emotions, it does not tell everything correctly. For example, words mean differently in different contexts. I think what to do to improve it is to give the computer many example sentences and make it understand various circumstances. I understand it is challenging to know all the usages or circumstances, if emotions are correctly determined by words, it would be amazing and useful to improve the company.