Friday, February 3, 2012

Online Friend Recommendation through Personality Matching

If you had read the paper "Online Friend Recommendation through Personality Matching and Collaborative Filtering"

It is useless at all for Online Dating because that recommender system called MatchMaker does not assess personality of members using a normative personality test.
Abstract: Most social network websites rely on people's proximity on the social graph for friend recommendation. In this paper, we present MatchMaker, a collaborative filtering friend recommendation system based on personality matching. The goal of MatchMaker is to leverage the social information and mutual understanding among people in existing social network connections, and produce friend recommendations based on rich contextual data from people's physical world interactions. MatchMaker allows users' network to match them with similar TV characters, and uses relationships in the TV programs as parallel comparison matrix to suggest to the users friends that have been voted to suit their personality the best. The system’s ranking schema allows progressive improvement on the personality matching consensus and more diverse branching of users' social network connections. Lastly, our user study shows that the application can also induce more TV content consumption by driving users' curiosity in the ranking process.

(the authors had several problems to calculate similarity)

"....While "People you may know" from Facebook recommends a friend to a user based upon the number of their mutual friends, work and education information, MatchMaker recommends a friend to a user based upon the matching in personality and characteristics that their social network friends, and TV show story writers, have collectively concluded.
Facebook uses proximity matching whereas MatchMaker uses personality matching for friend recommendation.
... our research on Facebook profiles has shown that the profile information is usually kept at a minimal level and gives not much information on the actual personality and characteristics of the recommended friends."

No comments:

Post a Comment