Have you ever thought of predicting what's going on around your social media presence? Do you have many employees taking care of that? Time for automating the vibechecks around your #hashtags.
In this project we tried to identify the key themes around which tweets regarding certain hashtags moved. We also analyzed the important themes along with the emotions associated with those themes.
We created a dashboard in using django to automatically update the data in real time for any organization to see.
Github: https://github.com/farabimahmud/howdyhack-overview
DevPosts: https://devpost.com/software/howdy-vibe-check
We tried to simulate different cache replacement policies using the Gem5 microarchitectural simulator to determine which one performs better. We used a the Bimodal Re-reference Interval Prediction, First In First Out and Random cache replacement policy in the Last Level Cache. The result is shown in the graph where we could see no replacement policies are performing better on all the cases.
Github: https://github.com/farabimahmud/secure-cache
The main customer for this Project is the CSE Department at Texas A&M University. Dr. Duncan Walker is the stakeholder. Our team WeCode was involved in the development of the project. The Project involved enhancing the legacy CSE Ph.D. Qualification Practice Application. The customer requirement included features to improve the application logging, timer for the quiz, Password recovery, bookmarking questions, email verification during registration, statistics for the quiz, ability to quit the quiz, and testing over different computing platforms like Mobile Phone, Desktop. These features were intended to improve the user authentications so that only people with a valid email can register, ability to recover password, ability to log in using Google, FB so that you don’t need to register and start using the application as a user, get more insights into your quiz performance and ability to give a timed quiz, improve the UI for Mobile phones.
GitHub: https://github.com/Shruthi-Sampathkumar/PhD-Qual-Practice/
Every other computer scientists tried so we gave our shot as well.
Some feature extraction techniques used with Fischer score -
Grayscale: 0.546655
Mean pixel: 0.546658
Horizontal edges: 0.909772
Vertical edges: 0.749883
HOG: 0.77018
With five fold cross validation we get the accuracy in different models -
CNN 91.6% AdaBoost 62.4%
GitHub: https://github.com/ColbyMainard/633-HW-5
In collaboration with TAMU School of Public Health, determined the key themes of COVID Vaccine hesitancy among Twitter users with 1,286,659 tweets posted within the timeline of July 19, 2020, to August 19, 2020.
Paper Link: https://osf.io/preprints/socarxiv/vc9jb/