Emotion Detection Using Sentiment Analysis on Twitter Tweets

Online social networks have become an integral part of our everyday lives, where we share our ideas, knowledge, feelings, & news with the community, get feedback and comment on existing information. Researchers are trying to fetch opinion information to analyze and summarize the opinions express

Project Title

Project Area of Specialization

Artificial Intelligence

Project Summary

Everyday huge amounts of text data is generated from twitter. This study will analyze an individual's real-world psychological mood by tracking their online activity on Twitter. By capturing tweets of an individual from twitter and ground truth data our results would suggest that real-world moods can be tracked by deriving information from text called text mining [1]. The text would be analyzed and then the writer’s mood would be predicted.

To analyze the text and predict writer's mood we will use following three algorithms:

Support Vector machine
Naïve Bayse
K-Nearest Neighbour.

Project Objectives

The objective of this project is:

to build a models that’s capable of detecting different types of Emotion like Excited, Happy, Anger, Sad, and Scared

Project Implementation Method

PROJECT METHODOLOGY

The first phase of the project starts with the collection of data using twitter API or datasets.
Then data will be filtered, cleaned, case normalized, converted into first form, and converted into the required format.
Once the data set is cleaned and created the second part of the project will begin which is to classify sentences called as feature selection.
While classifying words we need to measure the weight of words to predict mood which is called Feature weighting, to see the impact of a word on the sentence by using techniques: term frequency and inverse document frequency.
To solve this problem we will use classification. For classification we will use these algorithms: Support Vector machine, Naïve Bayse, and K-Nearest Neighbour.
Above three classification algorithms will be used with the same data set to develop a better understanding of the problems related to text mining and to compare the results and accuracy of different models.

Benefits of the Project

Benefit of the project:

Can be used in Intelligence Bureau, Social and Economic trends[6].
Intentions of an individual can be identified.
Psychiatrists can use the data for early detection of the psychological disorders such as anxiety or depression.
Can be used in text messages to predict emoticon according to mood in the text.
Brands will be able to get an inside look at consumer behaviors and, ultimately, better serve their audiences with the products, services and experiences they offer.