A dynamic flow of unstructured, fragmentary, and potentially unreliable stories characterizes the news landscape. Quickly finding relevant information challenges readers, who rely on tools to filter the stream of news. The spread of increasing concerns about disinformation coupled with privacy concerns necessitate improving these tools.
This workshop primarily addresses news recommender systems and analytics. The news streams may originate in large media companies, but may also come from social sites, where user models are needed to decide how user generated content is to be taken into account while minimizing the privacy concerns and the spread of disinformation. In this workshop we aim to bring researchers, media companies, and practitioners together, in order to exchange ideas about how to create and maintain a trusted and sustainable environment for digital news production and consumption. This year, we also provide the opportunity for the researchers who would like to test their ideas on real world news settings by using our datasets and evaluation platforms.
Workshop Theme and Topics
Topics of interests for this workshop include but are not limited to:
• News Recommendation
- News context modeling
- Deep learning
- Word embeddings
- News diversity and filter bubbles
- Big data technologies for news streams
- News evolution and trends
- Practical applications
- News recommendation on mobile platforms
- Group recommendations for news
- Gamification in news recommender systems
• News Analytics
- News semantics and ontologies
- News summarization, classification and sentiment analysis
- Large-scale news mining and analytics
- News from social media
• Fake News and Disinformation
- Detection and analysis of disinformation and/or misinformation
- Fake news
- Spread mechanisms of news disinformation
• User Experience Issues
- User behavior analysis
- User profiling
- Privacy and security in news recommender systems
- User perception analysis (e.g. privacy, disinformation etc.)
• Evaluation Platforms, Methods and Datasets
- Experiences with evaluation platforms
- News datasets
- Evaluation methods
For a more enhanced research and the reproducibility of research results, we believe the importance of readily available datasets and evaluation platforms. In this context, we provide the opportunity for our participants to evaluate their systems by using the following datasets and platforms if they would like to:
- NewsREEL has been part of the CLEF initiative in 2014–2017. It offers a dataset as well as access to a recommender systems in form of a “living lab.” The data in both settings include articles’ features, context-related features, and interactions of users on the platform.
- Adressa News Dataset: This dataset was published in 2017 with two versions available (1 week (2,286,835) entries and 10 weeks (27,223,576 entries) of data collection). The dataset can be used for both content- based and collaborative filtering methods. The content of the news articles (in Norwegian) is available on request. The dataset is available at: http://reclab.idi.ntnu.no/dataset
