VeracityVerify

Inspiration

The inspiration for this project stems from the growing concerns about misinformation and fake news in today's digital age. With the rapid spread of information through social media and other online platforms, it has become increasingly challenging for the general public to discern between fact and fiction. This project aims to leverage machine learning techniques to help identify factual statements that are worth fact-checking, thus empowering individuals to make more informed decisions. Learning

Throughout the development of this project, valuable insights into natural language processing (NLP) techniques, including text preprocessing, feature extraction, and sentiment analysis, were gained. Additionally, a deeper understanding of machine learning algorithms, particularly support vector machines (SVM), and their applications in text classification tasks was acquired. Building the Project

The project began with data collection and preprocessing. Utilizing a dataset containing text data along with labels indicating whether each statement was factual or not, the data was cleaned by removing unnecessary columns and handling missing values.

Next, feature extraction was performed using TF-IDF (Term Frequency-Inverse Document Frequency) vectorization. This technique converts the text data into numerical features that can be used by machine learning algorithms.

Once the model was trained and evaluated, it was tested on new sentences to determine whether they were check-worthy or not. Finally, a CSV file containing the predictions for the new sentences, including the sentence text and corresponding category (Yes or No), was generated. Challenges Faced

One of the main challenges encountered during the project was handling the imbalance in the dataset. With the dataset containing more non-check-worthy statements than check-worthy ones, there was a risk of bias towards the majority class. To address this, techniques such as oversampling the minority class or using class weights during model training were experimented with.

Another challenge was fine-tuning the model hyperparameters to achieve the best performance. This involved optimizing parameters such as the maximum number of features for TF-IDF vectorization, the choice of kernel for the SVM classifier, and regularization parameters. Conclusion

In conclusion, this project provided valuable hands-on experience in applying machine learning techniques to tackle real-world problems related to misinformation and fact-checking. By leveraging NLP, the developed model can assist in identifying factual statements that merit further scrutiny, thereby promoting critical thinking and informed decision-making in the digital age.

Built With

• colab
• jupyter
• kaggle
• python