Emotex emotion classifier

As the utilization of web-based media and the intensity of network is expanding each day, increasingly more information is being set up by individuals communicating their perspectives and feelings. What's more, utilizing these information a great deal of expectations and investigation are being made

Project Title

Emotex emotion classifier

Project Area of Specialization

Artificial Intelligence

Project Summary

As the utilization of web-based media and the intensity of network is expanding each day, increasingly more information is being set up by individuals communicating their perspectives and feelings. What's more, utilizing these information a great deal of expectations and investigation are being made to upgrade the algorithms to comprehend individuals so as to build better advertising systems and choices. Besides, it can assume a significant part in relational human correspondence.    The idea of this project is to investigate the emotion analysis field of machine learning and furthermore explore natural language processing which is sought after. Through this venture we would have the option to decide human emotions and how they react on the world situations on social media platforms. These emotions can be exceptionally valuable for business to make systems as needs be and furthermore for the administration to examine the mind-set and reactions of mass public. 

Project Objectives

The fundamental goal of this project will be to make use of el-reg dataset and classify tweets into four emotion categories, namely, joy, fear, anger and sadness. This would be this used to classifying emotions of tweets scraped from twitter. The main aim of the project will be to predict the emotion of user entered hashtags/ topics through live streaming tweets. 

Project Implementation Method

Feature extraction:

TfidfVectorizer - Transforms text to feature vectors that can be used as input to estimator. Vocabulary_ Is a dictionary that converts each token (word) to feature index in the matrix, each unique token gets a feature index. ... In each vector the numbers (weights) represent features tf-idf score.

The TfidfVectorizer performs both term frequency and inverse document frequency.It will tokenize documents, learn the vocabulary and inverse document frequency weightings, and allow you to encode new documents.

Models:

First we tried linear svc, which is to fit to the data you provide, returning a "best fit" hyperplane that divides, or categorizes, your data. From there, after getting the hyperplane, you can then feed some features to your classifier to see what the "predicted" class is.

We also tested naïve bayes algorithms, like multinomial as our data consists of words which were converted into numbers through tf-idf, and multinomial is better to use when there are multiple features or labels. bernoulliNB gave a lesser accuracy as it is more suitable for true and false values

Finally, we found the best fit svm model by using GridSearchCV, which  is a library function that is a member of sklearn's model_selection package. It helps to loop through predefined hyperparameters and fit your estimator (model) on your training set. So, in the end, you can select the best parameters from the listed hyperparameters.

GUI:

For Gui, we have used Tkinter library of python. We have created simple functions to display predictions. We used textbox for string input and button for taking the input

Benefits of the Project

This research based project would be profoundly pertinent and would have the option to categorize the feeling in four distinct emotions (joy, sadness, fear and) utilizing different techniques. This research based project would be profoundly gainful in understanding the human state of mind and their responses to world affairs. Additionally, this type of analysis can help the marketing sector to observe the public reaction and plan strategies accordingly, and also to get a general overview to public reaction.   

Technical Details of Final Deliverable

We created a user friendly GUI which allows the user to input any text will then be classified into emotion labels like joy, anger, fear and sadness. Moreover, the user can also input hashtag and number of tweets to be extracted from twitter which can be classified into emotions.

The classification was done by the system. Firts the data set of tweets were cleaned using various tools of NLTK to remove unwanted text from the tweets like url, mentions etc. Then the sentences were converted into numeric values using tfidf vectorizer. These vectors were then trained using svm algorithm through which the machine learns the features to distinguish the label. After this we tested the model and created the gui using tkinter Library of python for implementation using labels, text boxes, buttons and tables.

Final Deliverable of the Project

Software System

Core Industry

Media

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Partnerships to achieve the Goal

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com