Can Social Features Help Learning to Rank YouTube Videos?

Chelaru, Sergiu; Orellana-Rodriguez, Claudia; Altıngövde, İsmail Sengör

doi:10.1007/978-3-642-35063-4_40

Cited by 30 publications

(23 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We wish to gauge the quality of their correlation to see whether we can use the comment history as a surrogate to predict future views. While prior works [25,6] have shown that comments do exhibit a strong correlation with views, this is only done for a particular temporal snapshot (i.e., on some given day d, how highly correlated are the total cumulative view count with the comment count). To ensure the feasibility of our approach, we need to analyze how the histories of views and comments on individual items evolve over time.…”

Section: Correlation Of Comments and Viewsmentioning

confidence: 99%

Predicting the popularity of web 2.0 items based on user comments

Gao

Kan

et al. 2014

Proceedings of the 37th International ACM SIGIR Conference on Research &Amp; Development in Information Retrieval

View full text Add to dashboard Cite

In the current Web 2.0 era, the popularity of Web resources fluctuates ephemerally, based on trends and social interest. As a result, content-based relevance signals are insufficient to meet users' constantly evolving information needs in searching for Web 2.0 items. Incorporating future popularity into ranking is one way to counter this. However, predicting popularity as a third party (as in the case of general search engines) is difficult in practice, due to their limited access to item view histories.To enable popularity prediction externally without excessive crawling, we propose an alternative solution by leveraging user comments, which are more accessible than view counts. Due to the sparsity of comments, traditional solutions that are solely based on view histories do not perform well. To deal with this sparsity, we mine comments to recover additional signal, such as social influence. By modeling comments as a time-aware bipartite graph, we propose a regularization-based ranking algorithm that accounts for temporal, social influence and current popularity factors to predict the future popularity of items. Experimental results on three realworld datasets -crawled from YouTube, Flickr and Last.fmshow that our method consistently outperforms competitive baselines in several evaluation tasks.

show abstract

Section: Correlation Of Comments and Viewsmentioning

confidence: 99%

Predicting the popularity of web 2.0 items based on user comments

Gao

Kan

et al. 2014

Proceedings of the 37th International ACM SIGIR Conference on Research &Amp; Development in Information Retrieval

View full text Add to dashboard Cite

show abstract

“…Some works focus on analyzing different statistical patterns of UGC (user generated content) such as YouTube videos [13], or on how to improve IR effectiveness by exploiting these UGC, particularly users' actions, with their underlying social network [7,11,12,22,23,25].…”

Section: Time-independent Signals Approachesmentioning

confidence: 99%

“…In the last years, some works have concentrated on studying the richness and the possible use of these user-generated characteristics in search. Chelaru et al [11,12] studied the impact of social signals (like, dislike, comment, etc) on the effectiveness of search on YouTube. They showed that, although the basic criteria using the similarity of query with video title and annotations are effective for video search, social criteria are also useful and have improved the ranking of search results for 48% queries.…”

Section: Time-independent Signals Approachesmentioning

confidence: 99%

“…However, unlike previous work [3,7,11,12,19,23,25], that attempt to improve a search on specific social networks (e.g. YouTube, Twitter) by exploiting their own signals, our work specifically focuses on, firstly, exploiting various signals from different sources as document priors to enhance Web IR.…”

Section: Time-aware Social Signals Approachesmentioning

confidence: 99%

“…Most of existing approaches [11,12,22,25] exploit these signals to estimate the document prior by simply considering the quantity (the number) of signals related to a resource, or the polarity for some specific signals. In this paper, we hypothesize that the time (age) of signals (when they occur), signals diversity on a resource, and the resource age may affect on the prior estimation.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations