Presenter : Chuang, Kai-Ting Authors : Seok Jong YU 2012, IS

Intelligent Database Systems Lab

Presenter : Chuang, Kai-Ting

Authors : Seok Jong YU

2012, IS

The dynamic competitive recommendation algorithm in social network services


Outlines Motivation Objectives Methodology Experiments Conclusions Comments


Motivation• Existing recommendation algorithms depend on

single algorithm.


Objectives• To overcome this limitation and to provide consistent

and stable recommendation in social networks.


Methodology-Framework


MethodologyCreation of initial candidate pool

u → a → c

u → a ↗ ↘c b


Number of tweets

MethodologyCreation of initial candidate poolApplication of component algorithm

Number of followers

Common follower

PageRank

Profile matching

u → a → c

u → v ↗ ↘c a


Application of component algorithmCompetition algorithm

Methodology


Competition algorithm

Methodology

Integration of candidate list

Suggestion of top ranked candidates


Experiments-dataset


Experiments-RHR


Experiments-RP


Experiments-STDEV


Experiments-CV


Experiments-Overall


Conclusions

• A novel dynamic competitive approach for SNS

recommendation, for the purpose of overcoming the

limitation of single approaches.

• DCR algorithm enabled comprehensively more stable

and consistent recommendation.


Comments• Advantages– The subject is interesting.– The experiment is precise.

• Applications– Social network service.

Presenter : Chuang, Kai-Ting Authors : Seok Jong YU 2012, IS

Documents