Computer Science > Information Retrieval
[Submitted on 4 Jun 2015 (this version), latest version 29 Jan 2016 (v2)]
Title:Socially Driven News Recommendation
View PDFAbstract:The question of how to recommend and manage information available in the Internet is an active area of research, namely concerning news recommendations. The methods used to produce news rankings by the most popular recommender systems are not public and it is unclear if they reflect the real importance assigned by the readers. Also, the latency period between the publication of a news item and acknowledging its importance prevents recent news from being quickly presented as relevant or otherwise. This paper presents an approach to news recommendation that is based on trying to anticipate the importance that will be assigned to each news item by the readers upon publication. To avoid the latency problem our approach uses models that predict this importance before we have feedback from readers. We use the number of tweets of a news as a proxy to the readers assigned importance. Through the improvement and extension of previous work, we describe an experimental comparison of different approaches to the task of forecasting the amount of future tweets of news items using resampling strategies, and we develop and evaluate approaches to translate the resulting predictions into concrete news rankings. The key distinguishing aspect of our work concerning this predictive task is the focus on accuracy concerning highly popular news, which are rare but are the more relevant for news recommendation. In this context, the main contribution of this work is a study and a proposal to the problem of predicting highly popular news upon their publication and their application in the context of news recommendation. Additionally, we developed an application of this work and present a comparison between our rankings and Google News.
Submission history
From: Nuno Moniz [view email][v1] Thu, 4 Jun 2015 22:32:40 UTC (33 KB)
[v2] Fri, 29 Jan 2016 12:45:08 UTC (39 KB)
References & Citations
Bookmark
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.