Providing awareness, explanation and control of personalized filtering in a social networking site

Abstract

Social networking sites (SNSs) have applied personalized filtering to deal with overwhelmingly irrelevant social data. However, due to the focus of accuracy, the personalized filtering often leads to Bthe filter bubble^problem where the users can only receive information that matches their pre-stated preferences but fail to be exposed to new topics. Moreover, these SNSs are black boxes, providing no transparency for the user about how the filtering mechanism decides what is to be shown in the activity stream. As a result, the user's usage experience and trust in the system can decline. This paper presents an interactive method to visualize the personalized filtering in SNSs. The proposed visualization helps to create awareness, explanation, and control of personalized filtering to alleviate the Bfilter bubble^problem and increase the users' trust in the system. Three user evaluations are presented. The results show that users have a good understanding about the filter bubble visualization, and the visualization can increase users' awareness of the filter bubble, understandability of the filtering mechanism and to a feeling of control over the data stream they are seeing. The intuitiveness of the design is overall good, but a context sensitive help is also preferred. Moreover, the visualization can provide users with better usage experience and increase users' trust in the system.

What do we want from Explainable Artificial Intelligence (XAI)? – A stakeholder perspective on XAI and a conceptual model guiding interdisciplinary XAI research

“…Debugability Identify and fix errors and bugs Developer [5, 7, 12, 13, 16, 25, 30, 33, 44, 47, 48, 51, 56, 58, 61-64, 67, 74, 83-88] [ 82,89,90] Education Learn how to use a system and system's peculiarities User [6,13,24,31,33,36,39,49,74,80,88,91,92] Effectiveness Assess and increase a system's effectiveness; work effectively with a system Developer, User [13,26,34,36,39,42,48,49,65,67,74,77,78,[93][94][95] [ 28,31,46] Efficiency Assess and increase a system's efficiency; work efficiently with a system Developer, User [13,28,39,55,74,78,[93]…”

“…Developer [5,14,55,105] Safety Assess and increase a system's safety Deployer, User [44,58,69,70,74,78,105] Satisfaction Have satisfying systems User [5,13,15,24,28,33,36,39,43,46,78,79,94,102] [31]…”

Explanations as Mechanisms for Supporting Algorithmic Transparency

“…A third type of mechanism to promote greater transparency is providing explanations, a common approach in recommender systems [51] that may help solve problems caused by lack Algorithms that make autonomous decisions and provide recommendations are a "mission critical" aspect of many online content and e-commerce platforms, including Facebook, Google, Netflix, and Amazon [28,29,39]. Algorithmic decision-making systems like Facebook, and recommender systems like Netflix, fundamentally use the same kinds of technologies and perform very similar functions: both involve matching users with items.…”

“…However, in an algorithmic decision-making system, users are not explicitly informed that the information they see is a subset of what is available. Unlike recommender systems, algorithmic decision-making systems typically do not provide visibility into what the technologies are doing [39].…”

Collaboration technology in teams and organizations: Introduction to the special issue

“…Furthermore, personalized filtering systems usually do not provide details about the filtering process, resulting in decline of user trust. In their paper, Providing awareness, explanation and control of personalized filtering in a social networking site, Nagulendra and Vassileva (2016) propose a visualization tool with a metaphoric view to alleviate the aforementioned issue. The tool provides awareness (what has been hidden or shown), explanation (why it has been filtered), and control (how to adjust filtering rules) of the personalized filtering process to its users.…”