Hi, I am Junxian "Rupert" Wu

I am a senior Computer Science student in Georgia Institute of Technology, focusing on Artifical Intelligence and Media.

Right now, I am developing a college class scheduler named "Auto-Scheduler" with my team. It is based on Web and it can generate class schedule for students.

I am interested in Machine Learning and Computer Vision, and will continue to study in Georgia Institute of Technology for the Master degree in the next year.

I am also a fan of Japanese Anime, it is my dream that I can live and work in Japan to make my own Anime.

In the summer of 2016, I will work at Yahoo! Inc at Sunnyvale, CA as an intern Software Engineer.


Junxian "Rupert" Wu



Machine Learning on Customer Reviews to Predict Business Categories


pizza, chinese, rice, italian, bread, waitress, pasta, sandwich, salad, burger, bar, meal, server, dinner, soup, dish, service, table, shrimp, fried, dishes, good, fries, sauce, place, food, restaurant, wine, came, breakfast, plate, cheese, drinks, great, menu, night, steak, chicken, prices, excellent, went, try, nice, cooked, atmosphere, ok, delicious, served, asked, drink, friend, think, beer, friendly, coffee, dining, lunch, price, wasnt, ordered, restaurants, special, definitely, selection, check, did, experience, said, best, say, little, friends, bad, fish, feel, people, flavor, really, going, family, recommend, pretty, meat, time, home, favorite, perfect, decent, large, love, old, tried, visit, looked, local, worth, beef, huge, awesome, course

List of keywords after filtering in the review dataset

A system was built to predict and suggest missing categories of a restaurant from Yelp.com solely based on customer reviews.

To filter out significant keywords, the data was pre-processed with a NLP algorithm which calculated information gain per word and eliminate stop words. Then the system predicted the subcategories of restaurant businesses using an ensemble of Decision Tree, Bayes Network, K-NN and Neural Network algorithms based on statistics of keywords selected from the reviews.

Android App development: Shopping with Friends


An online mobile application in which registered users can share sales information with their friends. The application uses Firebase as backend database.

The application has following features:

  • User Service: registered users can add other users as friend though email and username registered users can post their interested goods. registered users can share sales with their friends.
  • Sales sharing: users can share sales with their friend with the good on sale, the price of good and the location of the sale
  • Notification: user can receive notification when user's friend send user's a sale that user is interested in.

On Going Project: College Class Scheduler "Auto Scheduler"


Class registration has always been an important issue for students. The known course arrangement web­app, CourseOff, can hardly help students find the best schedules to cater to their personal needs. Therefore, our team decides to build a web application which can automatically generate schedules for students based on their specific requirements. Our project, Auto­scheduler, is a web application that provides a convenient and efficient interface for college students to create their class schedules for the upcoming semester. Our application will provide the class information of different schools with the aid of CourseOff API, which means that our web application can be used by college students from different schools

The application is under development. The retrieval of course details was achieved and the basic template of HTML interface has been constructed. Right now, our team are working on the algorithm of scheduling classes with the model of Constrain Satisfaction Problem. Also, the backend Server and Database is nearly finished, more update can be viewed direactly from the Github repository.

The application is not published yet and we are still looking of hosting Server.