TU Berlin

Database Systems and Information Management GroupDesign and Implementation of a Flink-Based News Recommendation System on Tweeter Information

Logo FG DIMA-new  65px

Page Content

to Navigation

Basic Information

Student: YangChao Liu

Advisors: Sebastian Schelter, Prof. Dr. Volker Markl

Degree: Master


The Internet in the Web 2.0 era develops towards a more intelligent, personalized
and socialized direction, influences and changes peoples way of life. The most typical example is the SNS (Social Networking Services). Twitter stands out among
numerous SNS due to its real-time information and information sharing features.
Since Twitter has a huge amount of users who update Tweets information frequently, large amounts of user data are produced on Twitter every day. Therefore,
it has become an important study topic to explore ways to tap in-depth useful
information from these Tweets from users, and to provide users with personalized recommendation services. This paper applied content-based recommendation
algorithm to Twitter after researching the personalized news recommendation system. It implemented algorithms such as TF-IDF, TextRank, cosine similarity,
Euclidean-distance on the Flink platform, got their news recommendation ranks
and recommended personalized news service for each user. It also compared and
analyzed different results of these algorithms, and validated the accuracy of the


Quick Access

Schnellnavigation zur Seite über Nummerneingabe