Effect of the time window on the heat-conduction information filtering model

Recommendation systems have been proposed to filter out the potential tastes and preferences of the normal users online, however, the physics of the time window effect on the performance is missing, which is critical for saving the memory and decreasing the computation complexity. In this paper, by gradually expanding the time window, we investigate the impact of the time window on the heat-conduction information filtering model with ten similarity measures. The experimental results on the benchmark dataset Netflix indicate that by only using approximately 11.11% recent rating records, the accuracy could be improved by an average of 33.16% and the diversity could be improved by 30.62%. In addition, the recommendation performance on the dataset MovieLens could be preserved by only considering approximately 10.91% recent records. Under the circumstance of improving the recommendation performance, our discoveries possess significant practical value by largely reducing the computational time and shortening the data storage space.

[1]  Yang Li,et al.  INFORMATION FILTERING BASED ON USERS' NEGATIVE OPINIONS , 2013 .

[2]  Linyuan Lü,et al.  Predicting missing links via local information , 2009, 0901.0553.

[3]  Liu Jian-guo,et al.  Information Filtering via Improved Similarity Definition , 2010 .

[4]  Qiang Guo,et al.  Improved hybrid information filtering based on limited time window , 2014 .

[5]  A. Barabasi,et al.  Hierarchical Organization of Modularity in Metabolic Networks , 2002, Science.

[6]  Yi-Cheng Zhang,et al.  Bipartite network projection and personal recommendation. , 2007, Physical review. E, Statistical, nonlinear, and soft matter physics.

[7]  P. Jaccard,et al.  Etude comparative de la distribution florale dans une portion des Alpes et des Jura , 1901 .

[8]  Lu Qi-shao,et al.  Dynamical Effect of Calcium Pump on Cytosolic Calcium Bursting Oscillations with IP 3 Degradation , 2010 .

[9]  George Karypis,et al.  Item-based top-N recommendation algorithms , 2004, TOIS.

[10]  Qiang Guo,et al.  Heat conduction information filtering via local information of bipartite networks , 2012, The European Physical Journal B.

[11]  Tian Qiu,et al.  Information Filtering via a Scaling-Based Function , 2013, PloS one.

[12]  Lior Rokach,et al.  Recommender Systems Handbook , 2010 .

[13]  Lada A. Adamic,et al.  Friends and neighbors on the Web , 2003, Soc. Networks.

[14]  M. Newman,et al.  Vertex similarity in networks. , 2005, Physical review. E, Statistical, nonlinear, and soft matter physics.

[15]  Yi-Cheng Zhang,et al.  Effect of user tastes on personalized recommendation , 2009, ArXiv.

[16]  Tao Zhou,et al.  Evolution of the Internet and its cores , 2008 .

[17]  Tao Zhou,et al.  An item-oriented recommendation algorithm on cold-start problem , 2011 .

[18]  Mohsen Ramezani,et al.  A pattern mining approach to enhance the accuracy of collaborative filtering in sparse data domains , 2014 .

[19]  Sophie Ahrens,et al.  Recommender Systems , 2012 .

[20]  An Zeng,et al.  Extracting the Information Backbone in Online System , 2013, PloS one.

[21]  Michael McGill,et al.  Introduction to Modern Information Retrieval , 1983 .

[22]  Qiang Guo,et al.  Information filtering via biased heat conduction , 2011, Physical review. E, Statistical, nonlinear, and soft matter physics.

[23]  Yi-Cheng Zhang,et al.  Solving the apparent diversity-accuracy dilemma of recommender systems , 2008, Proceedings of the National Academy of Sciences.

[24]  Lucia Reining,et al.  Plasmon satellites in valence-band photoemission spectroscopy , 2012 .