How scales influence user rating behaviour in recommender systems

Fake or Credible? Towards Designing Services to Support Users’ Credibility Assessment of News Content

“…The qualitative evaluation uncovered users are willing to rate the information sources, which is in line with research on rating scales, for example, in the context of recommendation agents [33]. Furthermore, the participants rated the source as a piece of essential information for the credibility assessment, which was also found to be true in the context of fake news and its effects on behavioral intentions towards an advertised brand [18].…”

“…Nevertheless, deep learning can process a multitude of different features to reach high performance in this task [26]. In the context of recommender systems, research indicates that users prefer the communication of a prediction on different forms of rating scales [33]. Lastly, the design should provide supplemental information in combination with the content of a user's feed [2].…”

“…Consequently, the original source should be clearly positioned on top of the screen DF1. Moreover, the user interface should visualize the credibility of the source, which could be generated by fact-checking services, users, through data mining approaches or collaborative reputation systems [3,9,33,35]. Consequently, the source credibility should be visualized as a rating scale DF2.…”

“…Consequently, the source credibility should be visualized as a rating scale DF2. The outcome of the prediction for the present news article will be presented as a rating scale [12,33]. Consequently, the likelihood of the machine learning-based classification should be visualized as a rating scale DF3.…”

A User-Centric Diversity by Design Recommender System for the Movie Application Domain

“…Generally, individual averages tend to be more compact near 3.5 and 4 (in rating scale), following that users tend to give high ratings, but the presence of lower and higher averages suggests and clearly proves that users have different rating behaviours. The existence of varying rating scales is also supported by [19] and reinforced by an analysis performed on the effect of the rating scale granularity [20]. Moreover, we uncover the existence of subjective scales at which users adhere when rating movies [21]: as preferences are subjective, we need to standardize ratings so that individual biases can be removed to compare rating scales objectively and finally, to extract the positive preferences.…”