Twitter sentiment Analysis using SVM — Flutter App

Introduction:

After a long period of time, I finally have the opportunity to write about our final year project, which focused on Twitter sentiment analysis using Support Vector Machines (SVM) and the development of a powerful Flutter application. It was an exciting journey, especially when Twitter APIs were free and easily accessible for data mining. In this article, I will share the key aspects of our project, the insights we gained from analyzing sentiments on Twitter, and how we utilized Flutter to create an intuitive and user-friendly interface for our application.

Setting up the Twitter API:

The first step involved setting up the Twitter API in our code. We generated the necessary keys from the Twitter developer portal, implemented authentication in our code, and began mining data from Twitter. This initial setup was crucial to access real-time tweets and performing sentiment analysis.

Data Preprocessing and Sentiment Analysis:

To ensure accurate sentiment analysis, we performed data preprocessing tasks such as removing stop words, eliminating?@tags?and usernames, and cleaning the tweets. We utilized the TextBlob library, which provided the necessary tools to analyze sentiments. Unlike previous approaches that focused on binary classification (positive or negative), we aimed to enhance the results by employing multiclassification techniques.

Evaluation and Algorithm Selection:

We evaluated different algorithms to determine their accuracy in sentiment analysis. After thorough testing, we discovered that SVM outperformed other methods such as Naive Bayes and Random Forest. Hence, we chose SVM as our preferred algorithm for sentiment classification.

Insights and Visualization:

Once we obtained sentiment results, we explored the insights we could extract from the data. We plotted graphs to visualize sentiments across different categories, including positive, negative, strongly positive, strongly negative, and neutral. This graphical representation provided a comprehensive overview of sentiment distribution.

Demographic Analysis and User Insights:

In addition to sentiment analysis, we incorporated demographics information to gain deeper insights into the users. By analyzing their locations, we could take targeted actions based on geographic trends and user preferences. This feature allowed us to customize our approach and better understand our audience.

Building the Flutter App:

To make sentiment analysis accessible to everyone, we developed a powerful Flutter application. Flutter is a cross-platform framework that enabled us to create a seamless user interface for our sentiment analysis tool. It provided us with a rich set of UI components, fast performance, and the ability to deploy the app on both iOS and Android platforms. The Flutter app enhanced the user experience, allowing them to easily query any topic of interest and receive valuable sentiment insights.

Conclusion:

The Twitter sentiment analysis project using SVM and the accompanying Flutter app offered a unique perspective on sentiment analysis. By employing multiclassification techniques and leveraging the capabilities of Flutter, we achieved more accurate sentiment classification and delivered a user-friendly interface for our application. The visualization of sentiment data, the inclusion of demographic information, and the power of Flutter made our project impactful and valuable in the world of social media analysis.