Usability studies often overlook valuable insights from moderator-to-participant interventions. This study proposes that these interventions, moderator-provided “nudges”, are a rich source of insights on user needs, cognitive load, and system design gaps. We analyzed transcripts from 86 sessions involving two digital library systems and 56 blue-collar (BC) and 30 white-collar (WC) participants, and identified 962 instances of moderator interventions (i.e., nudges). Findings show significant disparities in both the volume and composition of nudges across groups. BC participants required 21 times as many nudges (N=919) as WC participants (N=43). Of the BC participants’ nudges, 737 (80.2%) were system-related, and 182 (19.8%) were user-related. Treating nudges as usability insights for HCI research enables researchers to identify where system scaffolding, such as explicit language, progressive guidance, or simplified workflows, is necessary. The study contributes to inclusive design theory by demonstrating how occupational background influences interventions in user studies and provides design implications for making digital information services more inclusive across occupational groups.

As generative AI (GenAI) is increasingly applied in persona development to represent real users, understanding the implications and limitations of this technology is essential for establishing robust practices. This scoping review analyzes how 81 articles (2022-2025) use GenAI techniques for the creation, evaluation, and application of personas. The articles exhibited good level of reproducibility, with 61% of articles sharing resources (personas, code, or datasets). Furthermore, conversational persona interfaces are increasingly provided alongside traditional profiles. However, nearly half (45%) of the articles lack evaluation, and the majority (86%) use only GPT models. In some articles, GenAI use creates a risk of circularity, in which the same GenAI model both generates and evaluates outputs. Our findings also suggest that GenAI seems to reduce the role of human developers in the persona-creation process. To mitigate the associated risks, we propose actionable guidelines for the responsible integration of GenAI into persona development.

Emerging multimodal conversational search (MCS) tools (e.g., Gemini Live) allow users to search for spatiotemporal information through natural language dialogues as they move through urban space. Despite the growing popularity of these tools, there is limited understanding of how people engage with this technology. To address this gap, we developed UrbanSearch, an MCS technology probe designed to capture the user’s current geolocation, time, and visual surroundings. A contextual inquiry (N=23) revealed that MCS tools provide two core values: requiring low effort in forming queries while offering highly relevant responses, and functioning as a central information gateway. As a promising technology, MCS supports environmental learning, in-situ decision making, and personalized navigation. Participants also revealed unmet needs for spatial reasoning and transparent integration of multi-source information, along with concerns related to peripheral awareness, social context, and personal space. Drawing from the findings, we discuss design implications for future MCS tools in urban spaces.

Metaverse environments combined with large language models (LLMs) enable guided interaction through LLM-powered avatars that function as embodied conversational agents. In our study, we examined how scholars interact with an LLM-powered avatar modeled after a real professor during a virtual reality (VR) tour of a research lab. As little is known about how metaverse characteristics shape the user experience (UX) mechanisms that drive acceptance of such technologies, we conducted a 2 (avatar realism: abstract vs. hyperrealistic) × 2 (immersion: desktop vs. headset-based VR) within-subjects study (N= 30), where academic participants engaged in a virtual lab tour guided by the professor avatar. We conducted path analyses on three conceptual models and, based on the results, proposed the Virtual Lab Acceptance Model (VLAM), which features an experiential path (where perceived immersion increases empathy towards the avatar and task enjoyment) and a rational path (where perceived realism increases avatar credibility and task confidence). Flow states amplify these pathways by strengthening task experiences. Task enjoyment is the strongest predictor of behavioral intention. These findings inform HCI research on metaverse characteristics to drive technology acceptance through UX mechanisms, yielding design implications for developing LLM-powered avatars for virtual labs.

The proliferation of artificial intelligence (AI) technologies has led to the rise of conversational decision-making support systems, such as dialogue persona systems that provide conversational access to various user segments. For example, product managers can ask personas about features before implementing them, politicians can learn about the needs of local communities through personas, and so on. Nascent research has looked at challenges when users interact with AI personas, but has not framed it as a principal–agent problem, in which the AI represents a persona that itself represents real people in the data. This setting exposes unique interaction challenges that decision makers face when engaging with AI-generated conversational personas, which we examine through a user study with 56 participants using AI-generated conversational personas. Our results indicate seven interaction challenges: (1) Hidden Information, (2) Hidden Personas, (3) Hidden UI, (4) Lack of AI Agency, (5) AI’s Selective Attention, (6) Confusing Distributional Information, and (7) Conversational Cold Start that we conceptually link with agency theory. We discuss how the interaction challenges could be alleviated and suggest directions for future work. • This article explores how decision makers interact with AI-generated conversational personas derived from real survey data. • It conducts a comparative study between conversational personas and traditional profile personas in decision-support contexts. • The study employs a think-aloud user study with 56 participants to capture interaction experiences and challenges. • It identifies seven specific interaction challenges unique to conversational personas that may hinder effective decision making. • The article provides insights and recommendations for designing conversational decision support systems using AI-generated personas.

We examined the accessibility of online information services (OISs) for underserved communities through a user study involving 56 blue collar participants interacting with a website and an app for five tasks. The blue collar participants were generally unsuccessful on both platforms, with 12.7% (n=7) unable to successfully complete any tasks; a hundred percent required at least minor assistance. Participants were also inefficient, taking 28.62 more steps than optimal (143.1%) on the website and 10.41 more steps (47.3%) on the app. Time inefficiency was also noteworthy, with 535.76 more seconds than optimal (248.0%) on the website and 266.55 more seconds (142.4%) on the app. Though still poor, the app yielded better outcomes with higher success rates and usability ratings. Digital proficiency correlated with success on both platforms, which is good news as this is addressable by OIS providers. Qualitative analysis revealed that many in this underserved population were unaware that these valuable OISs were available to them. Findings underscore the need for OIS providers to prioritize targeted outreach to inform underserved communities that OISs are open and welcoming. Designing OISs with accessibility and simplicity targeted for mobile devices is crucial for bridging the digital literacy gap and empowering underserved communities to engage effectively with OISs.

The development of persona systems provides a possibility for end users to interact with different persona modalities. In a 54-participant randomized controlled experiment, we compare two persona interaction modalities, document and dialogue personas, both generated using AI approaches from survey data. Overall, dialogue personas appear to be perceived more favorably than document personas. However, document personas exhibit a wider range of perceptions, suggesting that experiences with document personas are more polarizing among users. The document personas had higher transparency and were perceived as more complete, but the task completion was perceived as more difficult, although the task success rate was higher. The dialogue personas were perceived as more usable, with a higher System Usability Scale score, and more enjoyable. Our findings provide critical insights into the increasingly important area of persona interaction modalities and the broad paradigm of human-persona interaction.

With nearly two billion Muslims worldwide, designing technology that serves their needs is a significant task, especially during Ramadan, the season of spiritual reflection and change in lifestyle. We present a data-driven approach to generating personas representing differing religious views of Ramadan, based on a large-scale survey conducted in four Muslim-majority countries: Egypt, Indonesia, the United Arab Emirates, and Saudi Arabia. Our approach furthers the representation of underrepresented groups by embedding religious and cultural factors in persona development, aligning with the principles of value-sensitive design. Via correlation analysis, we identified seven distinct groups that reflect diverse practices and values during the Islamic holy month. These findings informed persona creation, capturing variations in spiritual engagement, digital media consumption, and the planning of Ramadan. The resulting personas provide actionable insights for developers designing inclusive applications such as charitable platforms and ecommerce systems aligned with Muslim values. We discuss the implications of employing personas for culturally aware system design and demonstrate how AI-generated personas can support inclusive design in religiously diverse settings. Ultimately, this work contributes to the growing research on human-centered technologies for value-aligned and context-aware systems.

We analyzed 83 persona prompts from 27 research articles that used large language models (LLMs) to generate user personas. Findings show that the prompts predominantly generate single personas. Several prompts express a desire for short or concise persona descriptions, which deviates from the tradition of creating rich, informative, and rounded persona profiles. Text is the most common format for generated persona attributes, followed by numbers. Text and numbers are often generated together, and demographic attributes are included in nearly all generated personas. Researchers use up to 12 prompts in a single study, though most research uses a small number of prompts. Comparison and testing multiple LLMs is rare. More than half of the prompts require the persona output in structured format, such as JSON, and $74 \%$ of the prompts insert data or dynamical variables. We discuss the implications of increased use of computational personas for user representation.

User engagement (UE) is widely discussed in HCI articles, but its definition, reliability, and application remain elusive. This research conducts a systematic literature review of 241 articles from 1993 to 2023 to analyze how UE is defined and measured within the domain of HCI. Our findings reveal significant definitional inconsistencies that hinder UE's practical application in HCI research and system design. Based on our findings, we recommend using UE as a categorical label rather than a unified construct until more systematic frameworks are established. We also highlight the need for divergent views of UE across HCI research communities as a valuable avenue to pursue. This divergent view approach can help HCI researchers focus on specific, measurable aspects of UE that align with specific community practices and norms. Our findings also suggest that until such a framework emerges, researchers should be aware of its limitations when using UE as a research construct.

Nascent research on human-computer interaction concerns itself with fairness of content moderation systems. Designing globally applicable content moderation systems requires considering historical, cultural, and socio-technical factors. Inspired by this line of work, we investigate Arab users' perception of Facebook's moderation practices. We collect a set of 448 deleted Arabic posts, and we ask Arab annotators to evaluate these posts based on (a) Facebook Community Standards (FBCS) and (b) their personal opinion. Each post was judged by 10 annotators to account for subjectivity. Our analysis shows a clear gap between the Arabs' understanding of the FBCS and how Facebook implements these standards. The study highlights a need for discussion on the moderation guidelines on social media platforms about who decides the moderation guidelines, how these guidelines are interpreted, and how well they represent the views of marginalised user communities.

We investigated the presence and acceptance of hallucinations (i.e., accidental misinformation) of an AI-generated persona system that leverages large language models for persona creation from survey data in a 54-user within-subjects experiment. After interacting with the personas, users were given a task to ask the personas a series of questions, including an unanswerable question, meaning the personas lacked the data to answer the question. The AI-generated persona system provided a plausible but incorrect answer half (52%) of the time, and more than half of the time (57%), the users accepted the incorrect answer, and the rest of the time, users answered the unanswerable question correctly (no answer). We found that when the AI-generated persona hallucinated, the user was significantly more likely to answer the unanswerable question incorrectly. Also, for genders separately, when the AI-generated persona hallucinated, it was significantly more likely for the female user and the male users to answer the unanswerable question incorrectly. We identified four themes in the AI-generated persona's answers and found that users perceive AI-generated persona's answers as long and unclear for the unanswerable question. Findings imply that personas leveraging LLMs require guardrails to ensure that personas clearly state the possibility of data restrictions and hallucinations when asked unanswerable questions.

Large Language Models (LLMs) are emerging as a powerful tool for AI-generated personas. This study evaluates the usability of AI-generated personas, comparing chat and profile formats. The findings indicate chat personas tend to be perceived more favourably, and profile personas exhibit greater variability in user perception. The increased difficulty and longer dwell time experienced by users with the profile persona, despite negative usability metrics, paradoxically resulted in better task performance. Usability issues indicate that many current limitations of AI, including verbosity, hallucinations, and empty rhetoric which was described as the persona having ‘no soul’, are inherited in AI-generated chat personas. However, there are also new issues. For one, the risk of information overload in an AI-generated profile persona implies that the AI does not consider human users’ cognitive limitations when designing the persona (but usability scores for profile personas increase with dwell time, implying that users get used to the longer format the more time they spend). Another is the ‘fourth wall’ effect of AI-generated chat personas in which the user feels they are talking to someone describing the persona rather than the persona itself. Future work could address the usability paradox and the fourth wall effect of using personas.CCS CONCEPTS Human-centered computing Human computer interaction (HCI)

Deepfake technology is increasingly integrated into global mobile and web services when human representation is not feasible or cost-effective. Our analysis of 202 deepfake avatars from three deepfake providers reveals significant demographic disparities with 18 out of 48 possible demographic groups unrepresented. Deepfake avatars' gender distribution was nearly balanced (49.01% male, 50.99% female), but older age groups (Baby Boomers and Silent Generation) were substantially underrepresented by 64.36% and 76.24%, respectively, relative to the average number of all deepfake avatars. Differences in language representation were present in deepfake avatar providers with only 1.06% of global languages covered. The findings indicate that current deepfake technology lacks diversity, primarily favoring young white individuals, neglecting older demographics, Asians, and Middle Eastern populations, with underrepresentation of 40.59% and 52.48%, respectively, relative to the average number of all deepfake avatars. Only 15.27% of deepfake avatars portray any occupational characteristics. Addressing these diversity gaps is crucial for better serving varied user groups and warrants attention from deepfake providers and caution from those using deepfakes.

• Shows how GenAI fundamentally transforms existing persona development issues through evolutionary amplification rather than creating entirely new problems, with traditional biases becoming algorithmic discrimination and manual inconsistencies becoming convincing AI hallucinations. • Reveals how traditional limitations manifest differently in GenAI contexts across transparency, fairness, reliability, and control domains, with expert validation showing 60% of challenges are more problematic for GenAIPs than conventional approaches. • Documents how GenAI transforms not just technical challenges but harm distribution, with persona developers facing operational complexity while target user groups bear severe consequences through systematic misrepresentation and exclusion. • Provides evidence that while GenAIPs appear to solve traditional limitations, they transform existing challenges into more complex forms requiring novel validation approaches and human-AI collaboration frameworks for responsible implementation. Generative AI personas (GenAIPs) promise user-centred design efficiency, but their impact on different persona challenges remains unexplored. Inspired by Dijkstra’s classic essay on harmful programming constructs, we analyze twenty challenges in persona development using Human-Centered AI principles. Through literature review and expert survey (n=17), we find that GenAIPs transform rather than eliminate traditional persona challenges. Experts rated all challenges as problematic for GenAIPs (M > 4.0), with the highest concerns for hallucinations (M=5.94), over-sanitization (M=5.82), and lack of standardization (M=5.59). 12 out of 20 challenges are considered more problematic for GenAIPs than conventional personas, particularly bias amplification, validation challenges, and accessibility without expertise. We provide HCAI-grounded guidelines demonstrating that effective GenAIP implementation requires human-AI collaboration rather than automation and prioritizing user welfare over technical efficiency.

Demographics are considered foundational information in most persona profiles. However, the effect of persona ethnicity and gender on designers’ identification with the persona has limited evaluation in the human-computer interaction literature. We conducted a study with 64 professional designers from the United States, Indian, Korean, and Mexican nationalities to investigate the effects of AI-generated persona ethnicity and gender on persona identification. The personas were created using Generative AI in the persona narratives and the persona video creation. The contribution of this work is that, against assumptions, neither persona ethnicity nor gender play a major role in persona identification among designers with different ethnic backgrounds. While there were some insinuations of ethnicity and gender in the open-ended feedback from the designers, the emergent qualitative themes describing persona identification were overwhelmingly universal and applicable regardless of ethnicity or gender. This implies that professional designers can effectively use personas with different demographic backgrounds, and effects of demographic attributes in personas leading to stereotyping are less impactful than presumed.

Stakeholders like designers use personas to learn about users. After persona development, stakeholders are usually presented with a persona set. However, there is little research on how stakeholders select a persona from a persona set. A think-aloud analysis with 37 stakeholders who were asked to select a persona for a content design task reveals that persona selection is influenced by comparative, non-comparative, and subjective elements. Persona choice is often made with task compatibility in mind: interests, professions, and education were important contextual factors in our focal task. Storifying is commonly applied by stakeholders, reflecting personas’ narrative nature. The persona’s picture is often evoked, in addition to nationality and name, though demographics do not play a decisive role. Stakeholders refer to a host of persona attributes when explicating their persona choice. Overall, reasonings for persona choice are multifaceted and individualistic, as we might expect given the information-richness of personas.

Large language models (LLMs) can generate personas based on prompts that describe the target user group. To understand what kind of personas LLMs generate, we investigate the diversity and bias in 450 LLM-generated personas with the help of internal evaluators (n=4) and subject-matter experts (SMEs) (n=5). The research findings reveal biases in LLM-generated personas, particularly in age, occupation, and pain points, as well as a strong bias towards personas from the United States. Human evaluations demonstrate that LLM persona descriptions were informative, believable, positive, relatable, and not stereotyped. The SMEs rated the personas slightly more stereotypical, less positive, and less relatable than the internal evaluators. The findings suggest that LLMs can generate consistent personas perceived as believable, relatable, and informative while containing relatively low amounts of stereotyping.

Cipherbot, an educational chatbot using large language models to answer student questions concerning learning materials uploaded by the educator, was pilot tested in a classroom setting. Forty-four students used Cipherbot for seven weeks, sending 8077 messages. The average number of messages sent per student was 184 (SD = 80), with an average length of 98 characters (SD = 80). The engagement followed a non-normal distribution, with few power users, implying that most students are still hesitant to adopt tools like Cipherbot. Cipherbot was able to answer 82.5% of the student questions, demonstrating a scalable ability to address students' learning queries, with some room for improvement.

The evolution of generative artificial intelligence (AI) technologies, particularly large language models (LLMs), has lead to consequences for the field of Human-Computer Interaction (HCI) in areas such as personalization, predictive analytics, automation, and data analysis. This research aims to evaluate LLM-generated topics derived from survey responses in comparison with topics suggested by humans, particularly participants recruited through a crowdsourcing experiment. We present an evaluation results to compare LLM-generated topics with human-generated topics in terms of Quality, Usefulness, Accuracy, Interestingness, and Completeness. This involves three stages: (1) Design and Generate Topics with an LLM (OpenAI’s GPT-4); (2) Crowdsourcing Human-Generated Topics; and (3) Evaluation of Human-Generated Topics and LLM-Generated Topics. However, a feasibility study with 33 crowdworkers indicated challenges in using participants for LLM evaluation, particularly in inviting humans participants to suggest topics based on open-ended survey answers. We highlight several challenges in recruiting crowdsourcing participants for generating topics from survey responses. We recommend using well-trained human experts rather than crowdsourcing to generate human baselines for LLM evaluation.

Personas inform design by representing diverse user needs. Since their initial application in commercial technology contexts, personas have been adopted in several research domains for public good, such as health, accessibility, politics and civic society, education, sustainability, cybersecurity, and criminology. In this review paper, we analyzed 58 research studies that created personas in these domains, referred to as Personas for Social Good (PFSG). In most studies, PFSG was primarily exploratory and focused on initial methodology development. More than half (59%) neglected to discuss concerns with stereotyping or evaluate how personas contributed to improving social concerns in their respective domains. To facilitate a shift towards more socially conscious persona applications, we identified and critically examined the most comprehensive PFSG domain applications in our sample. Based on their strengths, we present an ecological framework to guide researchers in holistically aligning persona creation efforts with addressing critical social challenges.

In this article, we present METRIC. Measuring Engagement Through Remote Interactions of Customers (METRIC) (https://metric.qcri.org/) is a tool for collecting, measuring, analyzing, and reporting the engagement of online systems through actual interactions of customers or users, either remote or in the lab. METRIC enables system stakeholders to enhance their understanding of their audience, customer, or users' actual behavior on pages, images, videos, interfaces, and online systems, including the gaze and interaction with sub-elements on a page within a system or comparisons via A/B testing. Along with eye-tracking devices, METRIC uses a webcam-based eye-tracking JavaScript library for the ability to monitor the users' real visual attention during interaction with the online system. METRIC provides sophisticated reporting features throughout the collecting, measuring, and analyzing process. METRIC can also be deployed in user experiments toward the design of better cooperation technologies, primarily due to its online nature.

Mobile Health applications facilitate users in managing their health and receiving health care services. Despite a proven utility, several privacy concerns have been raised. At this frontier, we should be responsible for enhancing our understanding of user privacy concerns and enabling privacy by design. This study uncovers and compares individual privacy concerns within health applications. Our scenario-based approach considers 432 different health application data exchanges. Participants (n=189) assessed 27 unique scenarios across three categories: data type, data sharing, and data retention. The findings show user concerns in the three categories, most prominently, data sharing. From this position, we propose design approaches to improve user awareness and control of health application data. We discuss the finding’s implications, including pre-emptive and predictive strategies for health app developers.

Compensating crowdworkers for their research participation often entails paying a flat rate to all participants, regardless of the amount of time they spend on the task or skill level. If the actual time required varies considerably between workers, flat rates may yield unfair compensation. To study this matter, we analyzed three survey studies with varying complexity. Based on the United Kingdom minimum wage and actual task completion times, we found that more than 3 in 4 (76.5%) of the crowdworkers studied were paid more than the intended hourly wage, and around one in four (23.5%) was paid less than the intended hourly wage when using a flat rate compensation model based on estimated completion time. The results indicate that the popular flat rate model falls short as a form of equitable remuneration, when perceiving fairness in the form of compensating one’s time. Flat rate compensation would not be problematic if the workers’ completion times were similar, but this is not the case in reality, as skills and motivation can vary. To overcome this problem, the study proposes three alternative compensation models: Compensation by Normal Distribution, Multi-Objective Fairness, and Post-Hoc Bonuses.

Employing customer information from one of the world's largest airline companies, we develop a price elasticity model (PREM) using machine learning to identify customers likely to purchase an upgrade offer from economy to premium class and predict a customer's acceptable price range. A simulation of 64.3 million flight bookings and 14.1 million email offers over three years mirroring actual data indicates that PREM implementation results in approximately 1.12 million (7.94%) fewer non-relevant customer email messages, a predicted increase of 72,200 (37.2%) offers accepted, and an estimated $72.2 million (37.2%) of increased revenue. Our results illustrate the potential of automated pricing information and targeting marketing messages for upselling acceptance. We also identified three customer segments: (1) Never Upgrades are those who never take the upgrade offer, (2) Upgrade Lovers are those who generally upgrade, and (3) Upgrade Lover Lookalikes have no historical record but fit the profile of those that tend to upgrade. We discuss the implications for airline companies and related travel and tourism industries.

Deepfakes, realistic portrayals of people that do not exist, have garnered interest in research and industry. Yet, the contributions of deepfake technology to human-computer interaction remain unclear. One possible value of deepfake technology is to create more immersive user personas. To test this premise, we use a commercial-grade service to generate three deepfake personas (DFs). We also create counterparts of the same persona in two traditional modalities: classic and narrative personas. We then investigate how persona modality affects the perceptions and task performance of the persona user. Our findings show that the DFs were perceived as less empathetic, credible, complete, clear, and immersive than other modalities. Participants also indicated less willingness to use the DFs and less sense of control, but there were no differences in task performance. We also found a strong correlation between the uncanny valley effect and other user perceptions, implying that the tested deepfake technology might lack maturity for personas, negatively affecting user experience. Designers might also be accustomed to using traditional persona profiles. Further research is needed to investigate the potential and downsides of DFs.

Personas represent the needs of users in diverse populations and impact design by endearing empathy and improving communication. While personas have been lauded for their benefits, we could locate no prior review of persona use cases in design, prompting the question: how are personas actually used to achieve these benefits? To address this question, we review 95 articles containing persona application across multiple domains, and identify software development, healthcare, and higher education as the top domains that employ personas. We then present a three-stage design hierarchy of persona usage to describe how personas are used in design tasks. Finally, we assess the increasing trend of persona initiatives aimed towards social good rather than solely commercial interests. Our findings establish a roadmap of best practices for how practitioners can innovatively employ personas to increase the value of designs and highlight avenues of using personas for socially impactful purposes.

Much of the reported work on personas suffers from the lack of empirical evidence. To address this issue, we introduce Persona Analytics (PA), a system that tracks how users interact with data-driven personas. PA captures users’ mouse and gaze behavior to measure users’ interaction with algorithmically generated personas and use of system features for an interactive persona system. Measuring these activities grants an understanding of the behaviors of a persona user, required for quantitative measurement of persona use to obtain scientifically valid evidence. Conducting a study with 144 participants, we demonstrate how PA can be deployed for remote user studies during exceptional times when physical user studies are difficult, if not impossible.

We conduct two studies to evaluate the suitability of artificially generated facial pictures for use in a customer-facing system using data-driven personas. STUDY 1 investigates the quality of a sample of 1,000 artificially generated facial pictures. Obtaining 6,812 crowd judgments, we find that 90% of the images are rated medium quality or better. STUDY 2 examines the application of artificially generated facial pictures in data-driven personas using an experimental setting where the high-quality pictures are implemented in persona profiles. Based on 496 participants using 4 persona treatments (2 × 2 research design), findings of Bayesian analysis show that using the artificial pictures in persona profiles did not decrease the scores for Authenticity, Clarity, Empathy, and Willingness to Use of the data-driven personas.

Studies in human-computer interaction recommend creating fewer than ten personas, based on stakeholders’ limitations to cognitively process and use personas. However, no existing studies offer empirical support for having fewer rather than more personas. Investigating this matter, thirty-seven participants interacted with five and fifteen personas using an interactive persona system, choosing one persona to design for. Our study results from eye-tracking and survey data suggest that when using interactive persona systems, the number of personas can be increased from the conventionally suggested ‘less than ten’, without significant negative effects on user perceptions or task performance, and with the positive effects of increasing engagement with the personas, having a more diverse representation of the end-user population, as well as users accessing personas from more varied demographic groups for a design task. Using the interactive persona system, users adjusted their information processing style by spending less time on each persona when presented with fifteen personas, while still absorbing a similar amount of information than with five personas, implying that more efficient information processing strategies are applied with more personas. The results highlight the importance of designing interactive persona systems to support users’ browsing of more personas.

Large commercial sentiment analysis tools are often deployed in software engineering due to their ease of use. However, it is not known how accurate these tools are, and whether the sentiment ratings given by one tool agree with those given by another tool. We use two datasets - (1) NEWS consisting of 5,880 news stories and 60K comments from four social media platforms: Twitter, Instagram, YouTube, and Facebook; and (2) IMDB consisting of 7,500 positive and 7,500 negative movie reviews - to investigate the agreement and bias of four widely used sentiment analysis (SA) tools: Microsoft Azure (MS), IBM Watson, Google Cloud, and Amazon Web Services (AWS). We find that the four tools assign the same sentiment on less than half (48.1%) of the analyzed content. We also find that AWS exhibits neutrality bias in both datasets, Google exhibits bi-polarity bias in the NEWS dataset but neutrality bias in the IMDB dataset, and IBM and MS exhibit no clear bias in the NEWS dataset but have bi-polarity bias in the IMDB dataset. Overall, IBM has the highest accuracy relative to the known ground truth in the IMDB dataset. Findings indicate that psycholinguistic features - especially affect, tone, and use of adjectives - explain why the tools disagree. Engineers are urged caution when implementing SA tools for applications, as the tool selection affects the obtained sentiment labels.

There has been little research into whether a persona's picture should portray a happy or unhappy individual. We report a user experiment with 235 participants, testing the effects of happy and unhappy image styles on user perceptions, engagement, and personality traits attributed to personas using a mixed-methods analysis. Results indicate that the participant's perceptions of the persona's realism and pain point severity increase with the use of unhappy pictures. In contrast, personas with happy pictures are perceived as more extroverted, agreeable, open, conscientious, and emotionally stable. The participants’ proposed design ideas for the personas scored more lexical empathy scores for happy personas. There were also significant perception changes along with the gender and ethnic lines regarding both empathy and perceptions of pain points. Implications are the facial expression in the persona profile can affect the perceptions of those employing the personas. Therefore, persona designers should align facial expressions with the task for which the personas will be employed. Generally, unhappy images emphasize realism and pain point severity, and happy images invoke positive perceptions.

Though photographs of real people are typically used to portray personas, there is little research into the potential advantages or disadvantages of using such images, relative to other image styles. We conducted an experiment with 149 participants, testing the effects of six different image styles on user perceptions and personality traits that are attributed to personas by the participants. Results show that perceptions of clarity, completeness, consistency, credibility, and empathy for a persona increase with picture realism. Personas with more realistic pictures are also perceived as more agreeable, open, and emotionally stable, with higher confidence in these assessments. We also find evidence of the uncanny valley effect, with realistic cartoon personas experiencing a decrease in the user perception scores.

Controlling the quality of social media feeds poses an issue for many users. Platforms such as Twitter give users some options to influence their feeds. Still, the selection of content predominantly relies on implicit rather than explicit user actions, as manual options for "cleaning the feed" are often cumbersome and difficult to use for most users. Here, we present Take Back Control, a web browser extension that gives users control to hide undesirable content from their social media feeds. The extension combines JavaScript (for hiding the content) and machine learning (for deciding what content to hide). Our current demonstration includes three filter types: Toxic, Political, and Negative content, with a possibility to add more filters, all of this with the overarching aim of helping end users control the information visible in their social media feeds.

Data-driven persona development unifies methodologies for creating robust personas from the behaviors and demographics of user segments. Data-driven personas have gained popularity in human-computer interaction due to digital trends such as personified big data, online analytics, and the evolution of data science algorithms. Even with its increasing popularity, there is a lack of a systematic understanding of the research on the topic. To address this gap, we review 77 data-driven persona research articles from 2005–2020. The results indicate three periods: (1) Quantification (2005–2008), which consists of the first experiments with data-driven methods, (2) Diversification (2009–2014), which involves more pluralistic use of data and algorithms, and (3) Digitalization (2015–present), marked by the abundance of online user data and the rapid development of data science algorithms and software. Despite consistent work on data-driven personas, there remain many research gaps concerning (a) shared resources, (b) evaluation methods, (c) standardization, (d) consideration for inclusivity, and (e) risk of losing in-depth user insights. We encourage organizations to realistically assess their data-driven persona development readiness to gain value from data-driven personas.

This research explores the value of personas for supporting professional advertisers to design adverts for social media. We test if a personified user group (PUG), when provided to online ad designers, results in better ad performance than when using a non-personified user group (NUG) that had no face picture or name. Our experiment has 30 participants that created Facebook ads using both PUG and NUG. We found that using PUG did increase advertising click performance of ads created by people who are more experienced with ads and personas. Moreover, an analysis of the ad texts showed that the use of PUG increased the empathy of the created ads, supporting the foundational empathy benefit cited in HCI literature. However, the use of PUG did not significantly increase purchase intent. The results imply that using PUG for online ad design evokes more empathy and improves click-through performance. More empathetic ads can have a positive impact on social media users, given that they appear to increase relevance.

Quantitative persona creation (QPC) has tremendous potential, as HCI researchers and practitioners can leverage user data from online analytics and digital media platforms to better understand their users and customers. However, there is a lack of a systematic overview of the QPC methods and progress made, with no standard methodology or known best practices. To address this gap, we review 49 QPC research articles from 2005 to 2019. Results indicate three stages of QPC research: Emergence, Diversification, and Sophistication. Sharing resources, such as datasets, code, and algorithms, is crucial to achieving the next stage (Maturity). For practitioners, we provide guiding questions for assessing QPC readiness in organizations.

Personas are a well-known technique in human computer interaction. However, there is a lack of rigorous empirical research evaluating personas relative to other methods. In this 34-participant experiment, we compare a persona system and an analytics system, both using identical user data, for efficiency and effectiveness for a user identification task. Results show that personas afford faster task completion than the analytics system, as well as outperforming analytics with significantly higher user identification accuracy. Qualitative analysis of think-aloud transcripts shows that personas have other benefits regarding learnability and consistency. However, the analytics system affords insights and capabilities that personas cannot due to inherent design differences. Findings support the use of personas to learn about users, empirically confirming some of the stated benefits in the literature, while also highlighting the limitations of personas that may necessitate the use of accompanying methods.

In an experiment, we present 38 marketing and data analysts professionals with two online AI-driven persona interfaces, one using numbers and the other using text. We employ eye tracking, think-aloud, and a post-engagement survey for data collection to measure perception and visual engagement with the personas along 7 constructs. Results show that the use of numbers has a mixed effect on the perceptions and visual engagement of the persona profile, with job role as a determining factor on whether numbers/text affect end users for 2 of the constructs. The use of numbers has a significant positive effect on user perceptions of usefulness by analysts but a significantly negative effect on user perceptions of completeness for both marketers and analysts. The use of numbers decreases the perceived completeness of the personas for both marketer and analysts. This research has both theoretical and practical consequences for AI-driven persona development and their interface design, suggesting that the inclusion of numbers can have a desirable effect for certain roles but with possible negative effects on user perceptions.

Access to social media often enables users to engage in conversation with limited accountability. This allows a user to share their opinions and ideology, especially regarding public content, occasionally adopting offensive language. This may encourage hate crimes or cause mental harm to targeted individuals or groups. Hence, it is important to detect offensive comments in social media platforms. Typically, most studies focus on offensive commenting in one platform only, even though the problem of offensive language is observed across multiple platforms. Therefore, in this paper, we introduce and make publicly available a new dialectal Arabic news comment dataset, collected from multiple social media platforms, including Twitter, Facebook, and YouTube. We follow two-step crowd-annotator selection criteria for low-representative language annotation task in a crowdsourcing platform. Furthermore, we analyze the distinctive lexical content along with the use of emojis in offensive comments. We train and evaluate the classifiers using the annotated multi-platform dataset along with other publicly available data. Our results highlight the importance of multiple platform dataset for (a) cross-platform, (b) cross-domain, and (c) cross-dialect generalization of classifier performance.

Understanding the causes or triggers of toxicity adds a new dimension to the prevention of toxic behavior in online discussions. In this research, we define toxicity triggers in online discussions as a non-toxic comment that lead to toxic replies. Then, we build a neural network-based prediction model for toxicity trigger. The prediction model incorporates text-based features and derived features from previous studies that pertain to shifts in sentiment, topic flow, and discussion context. Our findings show that triggers of toxicity contain identifiable features and that incorporating shift features with the discussion context can be detected with a ROC-AUC score of 0.87. We discuss implications for online communities and also possible further analysis of online toxicity and its root causes.

We analyze the effect of using smiling/non-smiling and stock photo/non-stock photo pictures in persona profiles on four key persona perceptions, including credibility, likability, similarity, and willingness to use. For this, we collect data from an experiment with 2,400 participants using a 16-item survey instrument and multiple persona profile treatments of which half have a smiling photo/stock photo and half do not. The results from structural equation modeling, supplemented by a qualitative analysis, show that a smile enhances the perceived similarity with the persona, similar personas are more liked, and that likability increases the willingness to use a persona. In contrast, the use of stock photos decreases the perceived similarity with the persona as well as persona credibility, both of which are significant predictors to a willingness to use a persona. These professionally crafted stock-photos seem to diminish the sense of identification with the persona. The above effects are consistent across the tested ages, genders, and races of the persona picture, although the effect sizes tend to be small. The results suggest that persona creators should use smiling pictures of real people to evoke positive perceptions toward the personas. In addition to presenting quantitative evidence on the predictors of willingness to use a persona, our research has implications for the design of persona profiles, showing that the picture choice influences individuals’ persona perceptions even when the other persona information is identical.

Computational techniques are becoming more common in persona development. However, users of personas may question the information in persona profiles because they are unsure of how it was created. This problem is especially vexing for data-driven personas because their creation is an opaque algorithmic process. In this research, we analyze the effect of increased transparency – i.e., explanations of how the information in data-driven personas was produced – on user perceptions. We find that higher transparency through these explanations increases the perceived completeness and clarity of the personas. Contrary to our hypothesis, the perceived credibility of the personas decreases with the increased transparency, possibly due to the technical complexity of the persona profiles disrupting the facade of the personas being real people. This finding suggests that explaining the algorithmic process of data-driven persona creation involves a “transparency trade-off”. We also find that the gender of the persona affects the perceptions, with transparency increasing perceived completeness and empathy of the female persona, but not for the male persona. Therefore, transparency may specifically assist in the acceptance of female personas. We provide practical implication for persona creators regarding transparency in persona profiles.

Analyzing 5,665 crowd ratings on 1,133 social media comments, we find that individuals tend to agree on the extremes of a hate rating scale more than in the middle when evaluating the hatefulness of online comments. The agreement is higher for less hateful comments and lowest on moderately hateful comments. The results have implications for researchers developing machine learning models for online hate processing, as the extreme classes are likely to require fewer annotations for reaching statistical stability. Our findings suggest that the models developed in this domain should consider the distributions of hate ratings rather than average hate scores.

Increased access to data and computational techniques enable innovations in the space of automated customer analytics, for example, automatic persona generation. Automatic persona generation is the process of creating data-driven representations from user or customer statistics. Even though automatic persona generation is technically possible and provides advantages compared to manual persona creation regarding the speed and freshness of the personas, it is not clear (a) what information to include in the persona profiles and (b) how to display that information. To query into these aspects relating information design of personas, we conducted a user study with 38 participants. In the findings, we report several challenges relating to the design of automatically generated persona profiles, including usability issues, perceptual issues, and issues relating to information content. Our research has implications for the information design of data-driven personas.

In this research, we investigate if and how more photos than a single headshot can heighten the level of information provided by persona profiles. We conduct eye-tracking experiments and qualitative interviews with variations in the photos: a single headshot, a headshot and images of the persona in different contexts, and a headshot with pictures of different people representing key persona attributes. The results show that more contextual photos significantly improve the information end users derive from a persona profile; however, showing images of different people creates confusion and lowers the informativeness. Moreover, we discover that choice of pictures results in various interpretations of the persona that are biased by the end users' experiences and preconceptions. The results imply that persona creators should consider the design power of photos when creating persona profiles.

We present Automatic Persona Generation (APG), a methodology and system for quantitative persona generation using large amounts of online social media data. The system is operational, beta deployed with several client organizations in multiple industry verticals and ranging from small-to-medium sized enterprises to large multi-national corporations. Using a robust web framework and stable back-end database, APG is currently processing tens of millions of user interactions with thousands of online digital products on multiple social media platforms, such as Facebook and YouTube. APG identifies both distinct and impactful user segments and then creates persona descriptions by automatically adding pertinent features, such as names, photos, and personal attributes. We present the overall methodological approach, architecture development, and main system features. APG has a potential value for organizations distributing content via online platforms and is unique in its approach to persona generation. APG can be found online at https://persona.qcri.org.

To more effectively convey relevant information to end users of persona profiles, we conducted a user study consisting of 29 participants engaging with three persona layout treatments. We were interested in confusion engendered by the treatments on the participants, and conducted a within-subjects study in the actual work environment, using eye-tracking and talk-aloud data collection. We coded the verbal data into classes of informativeness and confusion and correlated it with fixations and durations on the Areas of Interests recorded by the eye-tracking device. We used various analysis techniques, including Mann-Whitney, regression, and Levenshtein distance, to investigate how confused users differed from non-confused users, what information of the personas caused confusion, and what were the predictors of confusion of end users of personas. We consolidate our various findings into a confusion ratio measure, which highlights in a succinct manner the most confusing elements of the personas. Findings show that inconsistencies among the informational elements of the persona generate the most confusion, especially with the elements of images and social media quotes. The research has implications for the design of personas and related information products, such as user profiling and customer segmentation.

We investigate the alignment of international attention of news media organizations within 193 countries with the expressed international interests of the public within those same countries from March 7, 2016 to April 14, 2017. We collect fourteen months of longitudinal data of online news from Unfiltered News and web search volume data from Google Trends and build a multiplex network of media attention and public attention in order to study its structural and dynamic properties. Structurally, the media attention and the public attention are both similar and different depending on the resolution of the analysis. For example, we find that 63.2% of the country-specific media and the public pay attention to different countries, but local attention flow patterns, which are measured by network motifs, are very similar. We also show that there are strong regional similarities with both media and public attention that is only disrupted by significantly major worldwide incidents (e.g., Brexit). Using Granger causality, we show that there are a substantial number of countries where media attention and public attention are dissimilar by topical interest. Our findings show that the media and public attention toward specific countries are often at odds, indicating that the public within these countries may be ignoring their country-specific news outlets and seeking other online sources to address their media needs and desires.

We develop a methodology to automate creating imaginary people, referred to as personas, by processing complex behavioral and demographic data of social media audiences. From a popular social media account containing more than 30 million interactions by viewers from 198 countries engaging with more than 4,200 online videos produced by a global media corporation, we demonstrate that our methodology has several novel accomplishments, including: (a) identifying distinct user behavioral segments based on the user content consumption patterns; (b) identifying impactful demographics groupings; and (c) creating rich persona descriptions by automatically adding pertinent attributes, such as names, photos, and personal characteristics. We validate our approach by implementing the methodology into an actual working system; we then evaluate it via quantitative methods by examining the accuracy of predicting content preference of personas, the stability of the personas over time, and the generalizability of the method via applying to two other datasets. Research findings show the approach can develop rich personas representing the behavior and demographics of real audiences using privacy-preserving aggregated online social media data from major online platforms. Results have implications for media companies and other organizations distributing content via online platforms.

In this work, we describe a methodology for leveraging large amounts of customer interaction data with online content from major social media platforms in order to isolate meaningful customer segments. The methodology is robust in that it can rapidly identify diverse customer segments using solely online behaviors and then associate these behavioral customer segments with the related distinct demographic segments, presenting a holistic picture of the customer base of an organization. We validate our methodology via the implementation of a working system that rapidly and in near real-time processes tens of millions of online customer interactions with content posted on major social media platforms in order to identify both the distinct behavioral segments and corresponding impactful demographic segments. We illustrate the functionality of the methodology with real data from a major online content provider with millions of online interactions from more than thirty countries. We further show one possible use for such information via the automatic generation of personas for an organization, which can be used for the formulation of marketing strategy, implementation of advertising plans, or development of products. The research results offer insights into competitive marketing and product preferences for the consumers of online digital content. We conclude with a discussion of areas for future work.