Unified YouTube Video Recommendation via Cross-network Collaboration

Yan, Ming; Sang, Jitao; Xu, Changsheng

doi:10.1145/2671188.2749344

Cited by 73 publications

(32 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Yan et. al., [8], [9] introduced users' information on Twitter into Youtube video recommendation. Li,et.al.,[7] extracted sound and visual information of items to build a multiple kernel SVM recommendation algorithm.…”

Section: Sparsity and Cold-start Problem In Recommendationmentioning

confidence: 99%

“…To deal with the sparsity problems, there are basically two types of methods. One type of methods takes into account some auxiliary information, e.g., the sound, visual, textual or content information of videos [5]- [7] or the attribute of users extracted from cross network [8], [9]. Another type fills the missing ratings with default values [10], [11], e.g., the average of ratings or video popularity.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Improving Recommendation via Inference of User Popularity Preference in Sparse Data Environment

Tan

Guo

Chen

et al. 2018

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

The Collaborative Filtering (CF) algorithms work fairly well in personalized recommendation except in sparse data environment. To deal with the sparsity problem, researchers either take into account auxiliary information extracted from additional data resources, or set the missing ratings with default values, e.g., video popularity. Nevertheless, the former often costs high and incurs difficulty in knowledge transference whereas the latter degrades the accuracy and coverage of recommendation results. To our best knowledge, few literatures take advantage of users' preference on video popularity to tackle this problem. In this paper, we intend to enhance the performance of recommendation algorithm via the inference of the users' popularity preferences (PPs), especially in a sparse data environment. We propose a scheme to aggregate users' PPs and a Collaborative Filtering based algorithm to make the inference of PP feasible and effective from a small number of watching records. We modify a k-Nearest-Neighbor recommendation algorithm and a Matrix Factorization algorithm via introducing the inferred PP. Experiments on a large-scale commercial dataset show that the modified algorithm outperforms the original CF algorithms on both the recommendation accuracy and coverage. The significance of improvement is significant especially with the data sparsity. key words: user popularity preference, video recommendation, sparse data

show abstract

Section: Sparsity and Cold-start Problem In Recommendationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Improving Recommendation via Inference of User Popularity Preference in Sparse Data Environment

Tan

Guo

Chen

et al. 2018

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…Contemporary web services provide many important applications ranging from e-commerce websites [4,6,21,31] to online video/news platforms [9,36]. One of the most important modules of 2:2 X.…”

Section: Introductionmentioning

confidence: 99%

Understanding Assimilation-contrast Effects in Online Rating Systems

Zhang

Xie

Zhao

et al. 2019

ACM Trans. Inf. Syst.

View full text Add to dashboard Cite

Kong "Unbiasedness," which is an important property to ensure that users' ratings indeed reflect their true evaluations of products, is vital both in shaping consumer purchase decisions and providing reliable recommendations in online rating systems. Recent experimental studies showed that distortions from historical ratings would ruin the unbiasedness of subsequent ratings. How to "discover" historical distortions in each single rating (or at the micro-level), and perform the "debiasing operations" are our main objective. Using 42M real customer ratings, we first show that users either "assimilate" or "contrast" to historical ratings under different scenarios, which can be further explained by a well-known psychological argument: the "Assimilate-Contrast" theory. This motivates us to propose the Historical Influence Aware Latent Factor Model (HIALF), the "first" model for real rating systems to capture and mitigate historical distortions in each single rating. HIALF allows us to study the influence patterns of historical ratings from a modelling perspective, which perfectly matches the assimilation and contrast effects observed in experiments. Moreover, HIALF achieves significant improvements in predicting subsequent ratings and characterizing relationships in ratings. It also contributes to better recommendations, wiser consumer purchase decisions, and deeper understanding of historical distortions in both honest rating and misbehaving rating settings. CCS Concepts: • Information systems → Data mining; • Social and professional topics → User characteristics;

show abstract

“…Recommender systems are a typical information filtering technology, which have been widely used to offer users personalized recommendations among numerous potential choices [3,4]. In recent years, recommender systems have started to appear in a number of applications, such as recommending academic articles [5], videos [6], movies [7], locations [8], music [9], and services for e-business and egovernment [10].…”

Section: Introductionmentioning

confidence: 99%

Diffusion-based recommendation with trust relations on tripartite graphs

et al. 2017

View full text Add to dashboard Cite

Abstract. The diffusion-based recommendation approach is a vital branch in recommender systems, which successfully applies physical processes to make recommendations for users on bipartite or tripartite graphs. Trust links indicate users' social relations and can provide the benefit of reducing data sparsity. However, traditional diffusion-based algorithms only consider rating links when making recommendations. In this paper, the complementarity of users' implicit and explicit trust is exploited, and a novel resource-allocation strategy is proposed, which integrates these two kinds of trust relations on tripartite graphs. Through empirical studies on three benchmark datasets, our proposed method obtains better performance than most of the benchmark algorithms in terms of accuracy, diversity and novelty. According to the experimental results, our method is an efficient and reasonable way to integrate additional features into the diffusion-based recommendation approach.

show abstract

Unified YouTube Video Recommendation via Cross-network Collaboration

Cited by 73 publications

References 24 publications

Improving Recommendation via Inference of User Popularity Preference in Sparse Data Environment

Improving Recommendation via Inference of User Popularity Preference in Sparse Data Environment

Understanding Assimilation-contrast Effects in Online Rating Systems

Diffusion-based recommendation with trust relations on tripartite graphs

Contact Info

Product

Resources

About