Group Recommender Systems Enhanced by Social Elements

My research lines consist of improving current group recommendation techniques by introducing two novel factors: the personality of each individual and trust between group members. In this way I can simulate in a more realistic way the argumentation process followed by groups of people when agreeing on a common activity.


HappyMovie is a Facebook application where we provide social group recommendations for groups of people that wish to go together to the movies.

DataSet of Group Recommender Systems Enhanced by Social Elements

Through the several experiments that have been performed to validate my work I have gathered user related data, regarding users’ preferences and personality, and also group related information like trust between group members or group preferences and decisions.

In order to help other group recommender application developers and designers I here provide an anonymized dataset.

In case you manage to obtain good results with my dataset and publish you should cite my work and the origin of the dataset.

Download HappyMovie dataset

Creative Commons Licence
DataSet of Group Recommender Systems Enhanced by Social Elements by Lara Quijano is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
Based on a work at Group Recommender Systems Enhanced by Social Elements related papers.

To gather the data here presented I used my Facebook application HappyMovie. This dataset is a small sample of 58 users and 15 groups of different sizes.

During the experiment I employed the following instruments:

  • TKI metaphor: It is a personality test based on the TKI test [Thomas and Kilmann 1974]. This test is used to obtain a personality profi le about the way a person deals with conflict.
  • Movie listing: A list of 15 recent movies (of year 2009), which represents a movie listing from a cinema. This movie listing was chosen heterogeneously from movies from the MovieLens data set [Bobadilla et al. 2009].
  • Movies to rate: A list of 50 movies selected from the MovieLens data set.

The experiment was carried out as follows:

  1. Every user completed two di fferent questionnaires (personality and preferences test). The answers to these questionnaires were analyzed to defi ne each participants’ user profi le, which contains information about their personality and their individual movie preferences.
  2. The trust between users was automatically computed by extracting social information from users’ Facebook profiles.
  3. I managed to form groups of users. Afterwards, I asked each group to choose which 3 movies from the Movie Listing they would actually watch together.




If you are interested in any of these areas do not hesitate to contact me for collaborations.

Los comentarios están cerrados.