The document proposes a Twitter-based recommender system called Thought Bubbles. It would filter tweets and assign users to topic-related bubbles. The system would use natural language processing and filtering by metrics like tweet frequency and followers to recommend relevant tweets. An initial test of recommending tweets to 10 accounts found over 60% of the recommendations were accepted, showing potential for the system to discover new information on Twitter for research purposes.
MOOCs, Learning Analytics and OER - a perfect triangle for the future of ed...
A Conceptual Prototype for a Twitter Based Recommender System for Research 2.0
1. #THOUGHTBUBBLES
@IKNOW12
THOUGHT BUBBLES
A CONCEPTUAL PROTOTYPE FOR A TWITTER BASED
RECOMMENDER SYSTEM FOR RESEARCH 2.0
• TWO BASIC QUESTIONS:
• IS TWITTER USEFUL FOR DISCOVERING NEW INFORMATION IN SENSE OF A USERS
THOUGHT
RESEARCH 2.0? BUBBLE
SPORTS
• IS IT POSSIBLE TO FILTER RELEVANT INFORMATION FROM TWEETS?
• CONCEPT:
IOS DEV
• LET’S IMAGINE EVERY TWITTER USER BELONGS TO SEVERAL DIFFERENT
SERVICE
TOPIC RELATED BUBBLES. USER
TW
• REALIZATION: RE ITT
ST ER T
SOCIAL MEDIA
BU HO
• NATURAL LANGUAGE PROCESSING AP
I
BB UG
LE HT
S
• FILTERING BY COMMON TWITTER EVALUATION RATIOS LIKE AP
I
TWEET-FREQUENCY, FOLLOWER-COUNT, ETC
PRE-
NLP
• SYSTEM LEARNS AND IMPROVES RECOMMENDATIONS BY ANALYZING FILTERING
ACCEPTED RECOMMENDATIONS CLUSTERING
• PRELIMINARY RESULTS: DB
CATEGORI ANALYZE
• 10 TWITTER ACCOUNTS WERE GIVEN TOP 25 RECOMMENDATIONS SATION RECS
• OVER 60% ACCEPTANCE RATE!!!! SERVER
BY PATRCK THONHAUSER
Thursday, August 30, 12