Introduction of graduation project(SSOK)

20 Feb 2018

I’ve finally finished a graduation project with 2 team-mates.
Since some articles from university web page were messed up, we decide to build an android application which categories and recommends articles to the users.
We scrapped the articles from the main bulletin board and boards from each department(Physics, Law etc).
We also used the message(mobile) we’ve received.

What we’ve done

Classification

1. Naive Baysian

data	3,xxx
input	Title of the articles
output	['공지', '취업]
accuracy	0.9x

1. MLP with keras

data	2x,xxx
input	Title of the articles
output	['공지', '학사', '학생'.. ] in 8 categories
accuracy	0.9 , 0.85

1. Message Classfication
   Messages that students received from university were sent by the phone numbers of department.
   The problem was that most of the departments use more than one numbers. To use the numbers for classification, normalization between department and phone numbers had to be done first. After that, match the recieved message’s number with the list of department numbers.

Recommendation

1. content based recommendation
2. coordination based recommendation
3. word recommendation working with word2vec