This repos contains code for an assignemnt for AUC course Text Mining and Collective Intelligence. This will revolutionize course selection at AUC. -Selection of relevant posts -Keywords could be used such as "workload" -Date of posts (tend to be around enrolment period) -Identification of Courses -Based on name, even fall and spring semester could be used to increase accuracy -Levensteihn distance based on keyboard distance to take into account possible spelling errors
-Processing of information -Sentiment Analysis -Summaries (keywords and how frequently they are used to describe this course) -Possibly take into account the average sentiment of this person to use as a measure of accuracy/bias of these comments
`
Issues: -focus on sent analysis of facebook comments -or unsupervised methods -We can use try to use methods of product reviews
-Take into account different lecturers/structures of courses -Perhaps use topic modelling to see if there is a topic that matche talking about courses
For presentation: -Mining such a web, how API works (copy this, idea is last year they had somebody that did presentation on his book that is called mining social web)
-Most frequent shared feedback on course, trying to find a way to describe each course
-Choose from either:
- Polarity + description why something is positive or negative
- Simply look for a way to describe this class/course without polarity and how to visualize
-Take into account whether poor rating is due to teacher or