This paper provides a comprehensive review of explanations in recommender systems. We highlight seven possible advantages an explanation facility, and describe how existing measures can be used to evaluate the quality explanations. Since are not independent recommendation process, we consider ways recommendations presented may affect Next, look at different interacting with The is illustrated examples throughout, where from applications.

A common approach to designing Recommender Systems (RS) consists of asking users explicitly rate items in order collect feedback about their preferences. However, have been shown be inconsistent and introduce a non-negligible amount natural noise ratings that affects the accuracy predictions. In this paper, we present novel improve RS by reducing input data via preprocessing step. quantitatively understand impact noise, first analyze response recommendation algorithms noise. Next, propose...

This paper characterizes general properties of useful, or Effective, explanations recommendations. It describes a methodology based on focus groups, in which we elicit what helps moviegoers decide whether not they would like movie. Our results highlight the importance personalizing to individual user, as well considering source recommendations, user mood, effects group viewing, and effect expectations.

research-article Free Access Share on Recommender systems under European AI regulations Authors: Tommaso Di Noia Politecnico di Bari, Italy ItalyView Profile , Nava Tintarev TU Delft, The Netherlands NetherlandsView Panagiota Fatourou University of Crete, Greece GreeceView Markus Schedl Linz Institute Technology Lab LabView Authors Info & Claims Communications the ACMVolume 65Issue 4April 2022 pp 69–73https://doi.org/10.1145/3512728Online:19 March 2022Publication History...

Recent research claims that information cues and system attributes of algorithmic decision-making processes affect decision subjects' fairness perceptions. However, little is still known about how these factors interact. This paper presents a user study (N = 267) investigating the individual combined effects explanations, human oversight, contestability on informational procedural perceptions for high- low-stakes decisions in loan approval scenario. We find explanations contribute to...

This thesis focuses on explanations of recommendations. Explanations can have many advantages, from inspiring user trust to helping users make good decisions. We identified seven different aims explanations, and in this we will consider how be optimized for some these aims. both an explanation's content its presentation. As a domain, are currently investigating movie recommender, developing prototype system. paper summarizes the goals thesis, methodology using, work done so far our intended...

The growing volume of digital data stimulates the adoption recommender systems in different socioeconomic domains, including news industries. While recommenders help consumers deal with information overload and increase their engagement, use also raises an increasing number societal concerns, such as "Matthew effects", "filter bubbles", overall lack transparency. We argue that focusing on transparency for content-providers is under-explored avenue. As such, we designed a simulation framework...

A common criteria for Explainable AI (XAI) is to support users in establishing appropriate trust the – rejecting advice when it incorrect, and accepting correct. Previous findings suggest that explanations can cause an over-reliance on (overly advice). Explanations evoke are even more challenging decision-making tasks difficult humans AI. For this reason, we study by non-experts high-uncertainty domain of stock trading. We compare effectiveness three different explanation styles (influenced...

Selecting reviewers for code changes is a critical step an efficient review process. Recent studies propose automated reviewer recommendation algorithms to support developers in this task. However, the evaluation of algorithms, when done apart from their target systems and users (i.e., tools change authors), leaves out important aspects: perception recommendations, influence recommendations on human choices, effect user experience. This study first evaluate recommender vivo. We compare...

The way pages are ranked in search results influences whether the users of engines exposed to more homogeneous, or rather diverse viewpoints. However, this viewpoint diversity is not trivial assess. In paper, we use existing and novel ranking fairness metrics evaluate result rankings. We conduct a controlled simulation study that shows how can be used for diversity, their outcome should interpreted, which metric most suitable depending on situation. This paper lays out important groundwork...

Psychological factors such as personality, emotions, social connections , and decision biases can significantly affect the outcome of a process. These are also prevalent in existing literature related to inclusion psychological aspects recommender system development. Personality emotions users have strong with their interests decision-making behavior. Hence, integrating these into systems help better predict users’ item preferences increase satisfaction recommended items. In scenarios where...

Music preferences are likely to depend on contextual characteristics such as location and activity. However, most recommender systems do not allow users adapt recommendations their current context. We therefore built ContextPlay, a context-aware music that enables user control for both preferences. By conducting mixed-design study (N=114) with four typical scenarios of listening, we investigate the effect controlling in system aspects: perceived quality, diversity, effectiveness, cognitive...

Existing eXplainable Artificial Intelligence (XAI) techniques support people in interpreting AI advice. However, although previous work evaluates the users’ understanding of explanations, factors influencing decision are largely overlooked literature. This article addresses this gap by studying impact user uncertainty , correctness and interaction between explanation logic-styles for classification tasks. We conducted two separate studies: one requesting participants to recognize handwritten...

In this paper we consider how to help users better understand their consumption profiles by examining two approaches visualising user - chord diagrams, and bar charts aimed at revealing those regions of the recommendation space that are unknown them, i.e. blind-spots. Both visualisations do connecting profile preferences with a filtered space. We compare contrast in live study (n = 70). The results suggest that, although can both visualisations, diagrams particularly effective helping...

When recommendations become increasingly personalized, users are often presented with a narrower range of content. To mitigate this issue, diversity-enhanced user interfaces for recommender systems have in the past found to be effective increasing overall satisfaction recommendations. However, may different requirements diversity, and consequently visualization requirements. In paper, we evaluate two visual interfaces, SimBub ComBub, present diversity music system from perspectives. is...

Recent research in Augmented and Alternative Communication (AAC) has begun to make use of Natural Language Generation (NLG) techniques. This creates an opportunity for constructing stories from sensor data, akin existing work life-logging. paper examines the potential using NLG merge AAC life-logging domains. It proposes a four stage hierarchy that categorises levels complexity output text. formulates key subproblem clustering data into narrative events describes three approaches resolving...