A Video Recommendation Algorithm Based on Hyperlink-Graph Model

A Video Recommendation Algorithm Based on Hyperlink-Graph Model

Songtao Shang, Wenqian Shang, Minyong Shi, Shuchao Feng, Zhiguo Hong
Copyright: © 2017 |Pages: 15
DOI: 10.4018/IJSI.2017070104
(Individual Articles)
No Current Special Offers


The traditional graph-based personal recommendation algorithms mainly depend the user-item model to construct a bipartite graph. However, the traditional algorithms have low efficiency, because the matrix of the algorithms is sparse and it cost lots of time to compute the similarity between users or items. Therefore, this paper proposes an improved video recommendation algorithm based on hyperlink-graph model. This method cannot only improve the accuracy of the recommendation algorithms, but also reduce the running time. Furthermore, the Internet users may have different interests, for example, a user interest in watching news videos, and at the same time he or she also enjoy watching economic and sports videos. This paper proposes a complement algorithm based on hyperlink-graph for video recommendations. This algorithm improves the accuracy of video recommendations by cross clustering in user layers.
Article Preview

2. Classical Graph-Based Recommendation Algorithm

The classical graph-based recommendation algorithm (Li, Su & Wang, 2012) mainly depends on constructing a resource allocation matrix over a graph and following the random walk algorithm. The detailed description is as follows:

  • Step 1.

    Build a bipartite graph. Assume that there are U users and N videos in the recommendation system. The system can be expressed by a U+N nodes bipartite graph. It can be described as Figure 1.

Figure 1.

A user-video-based bipartite graph

  • Step 2.

    Construct the metric matrix. This process includes two parts. One is resource allocation from videos to videos. The other is resource allocation from user to videos. The resource allocation weight IJSI.2017070104.m01 from video j to user i can be described as follows:


where, Dj means how many users browse video j. Dk is the number of videos that user k has ever browsed. Hence, the metric matrix is a square matrix IJSI.2017070104.m03.

Complete Article List

Search this Journal:
Volume 12: 1 Issue (2024)
Volume 11: 1 Issue (2023)
Volume 10: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 9: 4 Issues (2021)
Volume 8: 4 Issues (2020)
Volume 7: 4 Issues (2019)
Volume 6: 4 Issues (2018)
Volume 5: 4 Issues (2017)
Volume 4: 4 Issues (2016)
Volume 3: 4 Issues (2015)
Volume 2: 4 Issues (2014)
Volume 1: 4 Issues (2013)
View Complete Journal Contents Listing