My tasks: 

• Propose tf-idf methods to detect “hot” topics using social network interation evaluation, resulting in obtaining trends within nearly 7 weeks which are “widely”concerned by 52/60 users (just over 86.67%) and removal of meaningless topics:

  • Improve tf score using the interest level on the topic of users who communicate with the comment​-owner, resulting in removing over 90% meaningless comments posted by famous users.

  • Propose a method that ensures topics are "locally" or widely discussed using adding weights to tf scores.

• Improve the detected trends using correlation coefficient, resulting in grouping 7/9 (nearly 80%) terms involving two outstanding trends within nearly 7 weeks.