Pith. sign in

REVIEW 4 major objections 5 minor 54 references

AttenTrack: Mobile User Attention Awareness Based on Context and External Distractions

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that a user's current attention can be predicted from non-sensitive phone signals alone—context, activity, and how they handle a notification—with no personal history, so attention-aware interfaces can work for brand-new u

desk verdict Useful dataset and behavioral analysis, but the headline interruption-optimization claim rides on a feature that only exists after the notification is sent. read the letter →

arxiv 2509.01414 v1 pith:2YYXRXPT submitted 2025-09-01 cs.HC

Yutong Lin , Suyuan Liu , Kaiwen Guo , Haohua Du , Chao Liu , Xiang-Yang Li This is my paper
classification cs.HC
keywords attentionawarenessmobilenotificationsexternaldistractionexperiencesamplingmethodcold-startpredictioninterruptibilitycontext-awarecomputingnotificationresponsebehavior
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that a person's momentary attention state on a smartphone—how focused they are right now—can be read from ordinary, non-sensitive device signals: time of day, physical activity, the foreground app, and the way they respond to a notification. The move that makes this plausible is treating every incoming notification as a natural, pre-existing experiment in distraction: a highly focused person resists or misses it, while an unfocused person clicks, swipes, or responds readily. Drawing on 9,002 experience-sampling responses from 43 people in daily life, the authors learn group-level patterns and show the resulting predictor works on users it has never seen, reaching an average F1 of 80.09% and recall of 85.84% in the binary 'completely unfocused' task—over 23% above a random baseline—and raising AUC by about 20% in the other two tasks. If this holds, attention-aware computing—apps that wait for the right moment to notify, teaching systems that adapt pacing, health platforms that prompt breaks—becomes deployable on day one for any new user, without wearables, without personality profiles, and without reading notification content. The paper also contributes what it describes as the first public real-world smartphone dataset linking context, external distractions, and subjective attention.

What carries the argument

The load-bearing object is the notification event as a naturally occurring distraction probe, paired with the ESM survey it triggers. Each incoming notification produces an objective trace—whether the user clicks, clears, cancels, ignores, fails to notice, or adjusts settings, and how long they take—which the paper treats as an observable reaction to an external disturbance; this trace is fused with context features into the AttenTrack model (Random Forest or Gradient Boosting), predicting one of several attention labellings (I: less/more focused; II: completely unfocused vs. somewhat focused; III: low/medium/high). The supporting hypothesis is that resistance to distraction scales with enga

What would settle it

A validation study with an independent attention reference: run AttenTrack on a fresh cohort alongside objective markers (reaction-time probes, touch dynamics, or EEG in a subset), and specifically inspect the moments when the ESM prompt is ignored. If a substantial share of unresponded prompts occur in objectively low-focus contexts—phone face-down in a pocket, driving, early sleep—then the equation 'unresponsive = highly focused' fails and cold-start recall is inflated. The same protocol can test whether the attention→response-time slope generalizes across users, which would settle the respo

Watch

Extended reading notes

Core claim

AttenTrack's central claim: attention can be predicted from context (time period, weekday, physical activity, foreground app type) plus distraction features (six response behaviors toward the triggering notification, and response time), with the model trained on group data rather than the target user's history. The paper grounds this in a field study whose ESM surveys are triggered by real notifications, tying every attention label to an actual distraction event. Response behavior varies strongly with attention, response time rises with attention at a decelerating rate, and roughly 55% of response motivations are personal rather than content-driven—so notification handling carries genuine at

Load-bearing premise

The paper assumes that self-reported attention, collected only when a user is available to answer a prompt, is an unbiased picture of attention—and that moments when users cannot respond are the highest-focus moments; the paper itself notes in Section 7.3 that responsive users are by definition somewhat available, so deep-engagement states are inferred rather than directly observed.

Editorial extensions

If this is right

  • Interruption optimization becomes deployable for new users on day one: AttenTrack II identifies 'completely unfocused' moments at 80.09% F1 and 85.84% recall with none of the user's own data, so ads or non-urgent pushes can be timed for low-focus windows.
  • Sustained-attention support (AttenTrack I) and three-level attention tracking (AttenTrack III) are both feasible from objective context alone, each lifting AUC by about 20% over random baselines—opening education, collaboration, and health-management use cases without wearables.
  • Cold start and personalization are complementary rather than competing: group models for similar lifestyles add over 20% recall for employed users, and roughly two weeks of personal data adds about 10% accuracy, so deployments can start generic and calibrate cheaply.
  • A sparse feature set suffices: coarse, system-recognizable response features perform on par with fine-grained self-reported behavior, easing real-world deployment.
  • The publicly released dataset—described as the first real-world smartphone attention dataset—provides an interpretable context–distraction–attention resource for future mobile attention research.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The response-time signal suggests a fully passive recalibration loop the paper does not run: because attention lengthens response time, a deployed system could quietly update its thresholds per user from natural notification latencies, with no ESM prompts at all.
  • If notification behavior reveals attention, the inference direction can be inverted: attention estimates could serve as a continuous, in-the-wild measurement instrument for evaluating how disruptive particular notification categories are—a use the paper hints at but leaves implicit.
  • The sampling design means 'highly focused' is partly an inference about missed surveys; a cheap external check would be a follow-up micro-survey asking users why they did not answer, to test whether missed prompts really coincide with deep engagement or with contexts like driving, sleep, or the phone being left behind.
  • Since the model needs only app category, activity, and response behavior, it could run entirely on-device and never transmit raw traces, substantially strengthening the privacy story for real deployments—an architectural consequence the paper does not develop.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper presents AttenTrack, a mobile attention-awareness model that predicts a user's self-reported attention state (1-5) from smartphone context (time, activity, foreground app) and response behaviors to notifications, with the goal of strong cold-start performance without personalized data. The authors ran two field studies with 43 participants, collecting 9,002 ESM-triggered entries, analyzed relationships among context, notification response, motivation, and attention, and evaluated three AttenTrack variants (binary less/more focused, binary completely/somewhat focused, three-level focus) under Leave-One-User-Out validation. They report, for the interruption-optimization variant, an average F1 of 80.09% and recall of 85.84%, and claim this outperforms the baseline by over 23%. The paper also announces a public dataset and discusses applications to collaboration, advertising, and health.

Significance. If the central claim were supported, this would be a meaningful contribution: a privacy-friendly, wearable-free attention predictor with cold-start capability would have broad applicability in interruptibility management and mobile health. The field-study dataset itself is valuable, and the use of open-ended motivation responses with LLM-assisted coding (Cohen's Kappa 0.93) is a methodological strength. The Leave-One-User-Out protocol is appropriate for the stated cold-start goal. However, the evaluation contains a temporal leakage issue that invalidates the interruption-optimization headline result as a deployable prediction, the baseline is only random prediction, and the ESM sampling biases the ground truth. These issues are load-bearing for the paper's primary claims, though they are fixable by re-framing or re-analysis.

major comments (4)
  1. [§6.2, Table 10] The headline result for interruption optimization is not a deployable prediction. The feature set includes the response behavior to the triggering notification (Table 1, §6.1) and response time (Table 1). In the Leave-One-User-Out evaluation, each test instance contains the user's actual response to the notification whose attention label is predicted. At the moment a system must decide whether to interrupt, that response has not yet occurred. Thus the F1=80.09 for AttenTrack II measures post-hoc recognition of an already-observed reaction, not the ability to time an interruption. Moreover, the response-behavior feature is self-reported in the same ESM survey as the attention label (Q3 vs. Q1, §3.2.2), so shared method variance can inflate apparent association. To support the interruption-optimization claim, evaluate using only features observable before notification delivery, or lagged r
  2. [§6.2, baseline] The only comparison is 'random prediction as the Baseline.' No existing interruptibility or attention-prediction model is used as a baseline, so the 'over 23%' improvement over random gives no evidence that AttenTrack advances the state of the art. At minimum, compare against a context-only model (currently only in Fig. 7 without reported numbers) and against standard published baselines (e.g., interruptibility models using context features). The context-only vs. all-features ablation in Fig. 7 is not quantified in the text; those numbers are needed to calibrate how much performance comes from the post-outcome response-behavior features.
  3. [§7.3] The ESM design censors exactly the high-focus states the model is intended to detect. Labels exist only for moments when the user responded to the ESM survey; the authors acknowledge this and assume unresponsive moments are highly focused. This assumption is untested, and the model is trained on a sample that systematically under-represents full engagement. Consequently, the reported cold-start generalization may not transfer to real deployment, where unresponsive moments are common. Provide a sensitivity analysis or validation of the assumption (e.g., comparing inferred attention on unresponsive vs. responsive episodes), or temper the deployment claims.
  4. [§6.1, Appendix A] The three attention thresholds are defined after inspecting the full dataset's attention distributions (level 3 for AttenTrack I, level 1 for AttenTrack II), and the same full dataset is used in LOUO. This is a subtle target leakage: the test user's self-reports influence the threshold definitions. Also, participants with near-constant labels are excluded (Appendix A), even though Table 10 says 'all participants.' Both choices should be disclosed in the main text and, ideally, thresholds should be chosen within training folds only.
minor comments (5)
  1. [§6.1] The exact feature vector used in each AttenTrack variant is not specified. Please provide a table listing each feature and whether it is observable before or after notification delivery.
  2. [Throughout] Typographical errors: 'sill' (Table 5), 'Vedio' (Table 3), 'studing', 'We design' in the abstract, and inconsistent capitalization. A proofread pass is needed.
  3. [Figure 7] The feature-ablation figure has no numeric labels or axis values; the reader cannot judge the magnitude of the differences. Report the underlying numbers in the text or in a table.
  4. [Table 10] No statistical tests are reported for differences among variants. Given the large standard deviations relative to some differences (e.g., AttenTrack I_GB vs. AttenTrack I_response_GB), paired significance tests would help the reader interpret the comparison.
  5. [§7.3] The statement that 'AttenTrack treats cases where users are completely unresponsive as highly focused' is presented as a deployment property, but the model training and evaluation do not include such cases. Clarify how this mapping would be implemented and how it is evaluated or justified.

Circularity Check

1 steps flagged · score 6.0 of 10

Interruption-optimization evaluation leaks the outcome: response-behavior features are measured on the same notification the model is supposed to time, so the reported F1/recall is post-hoc recognition, not pre-interruption prediction.

  1. other [Section 6.1 (response-behavior features), Section 6.2 (LOUO evaluation, Table 10), Section 7.2 (interruption-optimization case)]
    "In the AttenTrack II(“completely unfocused” vs. “somewhat focused”), the RF model performs better, achieving an average F1 score of 80.09% and a recall rate of 85.84% for unknown users... AttenTrack II focuses on identifying users' complete lack of focus... For example, when the system detects a user is unfocused, it can push consumer ads or entertainment content... we collect six types of fine-grained notification response behaviors ... aligning the feature design with the system's ability to recognize coarser behaviors."

    The model's inputs include the user's response behavior (clicked, ignored, no response) to the notification that triggered the ESM survey. In the claimed interruption-optimization deployment (Section 7.2), the system must decide whether to push an ad before the user responds to it; the user's response to the candidate notification does not yet exist. The Leave-One-User-Out evaluation (Section 6.2) nevertheless supplies that post-outcome response as a feature for the same instance whose attention state is predicted. Thus the reported F1=80.09/recall=85.84 measures recognition of an already observed reaction, not a prospective prediction of interruptibility. The model is rewarded for reading off the very behavior (click/ignore) that the interruption-optimization system is supposed to influen

full rationale

The derivation chain itself is not mathematically circular: AttenTrack is a standard classifier, and the paper does not fit a parameter and then predict it. No load-bearing self-citations or imported uniqueness theorems were found. The circularity is concentrated in the evaluation for the paper's own headline scenario. The response-behavior feature is a post-outcome measurement of the same notification that the system is supposed to time; the Leave-One-User-Out protocol removes user identity but not this temporal leak, so the reported F1/recall overstates deployable interruption-optimization performance. A secondary concern is that in the collected data, both the response-behavior feature and the attention label come from the same self-reported ESM survey (Section 3.2.2), introducing correlated measurement error; however, the paper's argument that these behaviors are system-recognizable mitigates this for deployment. The threshold choices in Section 6.1 are data-informed label definitions, not circularity. The acknowledged ESM limitation (Section 7.3) is a validity caveat but not a circular derivation. Overall, the central cold-start claim retains some independent content (context features alone could support attention-maintenance tasks), but the headline interruption-optimization result is partially circular because the 'prediction' reduces to an already observed outcome. Score 6 reflects this partial circularity.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim depends on the validity of self-reported attention as ground truth, the assumed response model, and the representativeness of notification-triggered sampling. No new physical entities are introduced.

free parameters (1)
  • Attention thresholds for AttenTrack I/II/III = level 3 split; level 1 vs >1; levels 1, 2-3, 4-5
    The thresholds are chosen in Section 6.1 based on the observed distribution of self-reported attention in the collected dataset, then evaluated by leave-one-user-out. This is a post hoc data-driven selection, not a parameter derived from theory or prior work.
assumptions (3)
  • domain assumption Self-reported 1-5 Likert responses to ESM Q1 are a valid measure of the user's actual attention state.
    Invoked throughout, first in Section 3.2.2 where Q1 is defined, and used as ground truth for all analyses and model training. There is no objective physiological validation.
  • domain assumption A user's response to a notification can be modeled as d = a f(A) + beta g(S) + epsilon, where S is distraction attributes and epsilon is noise.
    Introduced in Section 1 as an abstraction of the response mechanism. This linear decomposition is assumed, not derived, and underlies the feature design where response behavior is treated as an indicator of attention.
  • domain assumption Notification-triggered ESM samples are representative of attention states generally, and non-response during deep engagement can be treated as high focus.
    The sampling design triggers surveys only when notifications arrive, and Section 7.3 explicitly acknowledges that users who cannot respond are missing. The paper assumes in deployment that unresponsiveness corresponds to high attention, without direct evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AttenTrack: Mobile User Attention Awareness Based on Context and External Distractions." pith.science (2026). https://pith.science/paper/2YYXRXPT

@misc{pith2026250901414,
  author       = {Pith},
  title        = {Pith review of: AttenTrack: Mobile User Attention Awareness Based on Context and External Distractions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2YYXRXPT}},
  note         = {Machine review of arXiv:2509.01414}
}
read the original abstract

In the mobile internet era, managing limited attention amid information overload is crucial for enhancing collaboration and information delivery. However, current attention-aware systems often depend on wearables or personalized data, limiting their scalability and cross-context adaptability. Inspired by psychological theories, we attempt to treat mobile notifications as naturally occurring external distractions and infer users' attention states based on their response behaviors and contextual information. Our goal is to build an attention-aware model that does not rely on personalized historical data or complex subjective input, while ensuring strong cold-start capability and cross-context adaptability. To this end, We design a field study framework integrating subjective and objective data, closely aligned with real-world external distractions (i.e., mobile notifications). Through field studies, we construct a fine-grained and interpretable dataset centered on the relationship among current context - external distractions - subjective attention. Through our field studies, we conduct an in-depth analysis of the relationships among users' response behaviors, response motivations, contextual information, and attention states. Building on our findings, we propose AttenTrack, a lightweight, privacy-friendly attention awareness model with strong cold-start capability. The model relies solely on non-privacy-sensitive objective data available on mobile devices, and can be applied to a variety of attention management tasks. In addition, we will publicly release the constructed dataset to support future research and advance the field of mobile attention awareness.

Figures

Figures reproduced from arXiv: 2509.01414 by the authors.

Figure 1
Figure 1. The figure depicts the timeline for data collection through ESM surveys. The ESM survey indicates the [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. The figure illustrates the questionnaire content of the ESM survey, the automatically collected data, and the [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. The bar chart illustrates the distribu￾tion of all data collected in two rounds across dif￾ferent time periods, cov￾ering most hours of the day [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The figure shows the distribution of attention levels among all participants. [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 7
Figure 7. Figure 7: The figure shows the average AUC when utilizing only context features, only external distrac￾tion features, and all features [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: The figure shows the aver￾age cold-start performance of the group model and the general model for em￾ployed users [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 52 canonical work pages

  1. [1]

    Attention

    Steven J Luck and Shaun P Vecera. Attention. Stevens’ handbook of experimental psychology, 1:235–286, 2002

    work page 2002

  2. [2]

    The psychology of attention

    Elizabeth Styles. The psychology of attention. Psychology Press, 2006

    work page 2006

  3. [3]

    Oasis: A framework for linking notification delivery to the perceptual structure of goal-directed tasks

    Shamsi T Iqbal and Brian P Bailey. Oasis: A framework for linking notification delivery to the perceptual structure of goal-directed tasks. ACM Transactions on Computer-Human Interaction (TOCHI), 17(4):1–28, 2010

    work page 2010

  4. [4]

    Don’t bother me

    Chunjong Park, Junsung Lim, Juho Kim, Sung-Ju Lee, and Dongman Lee. Don’t bother me. i’m socializing! a breakpoint-based smartphone notification system. In Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing, pages 541–554, 2017

    work page 2017

  5. [5]

    Controlling interruptions: awareness displays and social motivation for coordination

    Laura Dabbish and Robert E Kraut. Controlling interruptions: awareness displays and social motivation for coordination. In Proceedings of the 2004 ACM conference on Computer supported cooperative work , pages 182–191, 2004

    work page 2004

  6. [6]

    Estimating work engagement from online chat logs

    Hiroaki Tanaka, Wataru Yamada, and Keiichi Ochiai. Estimating work engagement from online chat logs. In Proceedings of the Asian CHI Symposium 2021, pages 70–73, 2021

    work page 2021

  7. [7]

    Objective measurement of hyperactivity using mobile sensing and machine learning: Pilot study

    Oliver Lindhiem, Mayank Goel, Sam Shaaban, Kristie J Mak, Prerna Chikersal, Jamie Feldman, and Jordan L Harris. Objective measurement of hyperactivity using mobile sensing and machine learning: Pilot study. JMIR Formative Research, 6(4):e35803, 2022

    work page 2022

  8. [8]

    Edusense: Practical classroom sensing at scale

    Karan Ahuja, Dohyun Kim, Franceska Xhakaj, Virag Varga, Anne Xie, Stanley Zhang, Jay Eric Townsend, Chris Harrison, Amy Ogan, and Yuvraj Agarwal. Edusense: Practical classroom sensing at scale. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 3(3):1–26, 2019

    work page 2019

Show all 54 references
  1. [9]

    Real-time attention monitoring using smart wearable for adhd patients

    Smitha G Prabhu, Joydeep Das, Lavina Moncy, PS Lalith Kumar, and M Chaitra. Real-time attention monitoring using smart wearable for adhd patients. In 2023 International Conference on Data Science and Network Security (ICDSNS), pages 1–5. IEEE, 2023

    2023

  2. [10]

    Unobtrusive assessment of students’ emotional engagement during lectures using electrodermal activity sensors

    Elena Di Lascio, Shkurta Gashi, and Silvia Santini. Unobtrusive assessment of students’ emotional engagement during lectures using electrodermal activity sensors. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2(3):1–21, 2018

    2018

  3. [11]

    Safedriving: An effective abnormal driving behavior detection system based on emg signals

    Yuanzhao Fan, Fei Gu, Jin Wang, Jianping Wang, Kejie Lu, and Jianwei Niu. Safedriving: An effective abnormal driving behavior detection system based on emg signals. IEEE Internet of Things Journal, 9(14):12338–12350, 2021

    2021

  4. [12]

    Mindful interruptions: a lightweight system for managing interruptibility on wearables

    Claudio Forlivesi, Utku Günay Acer, Marc Van Den Broeck, and Fahim Kawsar. Mindful interruptions: a lightweight system for managing interruptibility on wearables. In Proceedings of the 4th ACM Workshop on Wearable Systems and Applications, pages 27–32, 2018

    2018

  5. [13]

    Using context-aware computing to reduce the perceived burden of interruptions from mobile devices

    Joyce Ho and Stephen S Intille. Using context-aware computing to reduce the perceived burden of interruptions from mobile devices. pages 909–918, 2005

    2005

  6. [14]

    Attelia: Reducing user’s cognitive load due to interruptive notifications on smart phones

    Tadashi Okoshi, Julian Ramos, Hiroki Nozaki, Jin Nakazawa, Anind K Dey, and Hideyuki Tokuda. Attelia: Reducing user’s cognitive load due to interruptive notifications on smart phones. In 2015 IEEE international conference on pervasive computing and communications (PerCom), pag...

    2015

  7. [15]

    Engagement-aware computing: Modelling user engagement from mobile contexts

    Akhil Mathur, Nicholas D Lane, and Fahim Kawsar. Engagement-aware computing: Modelling user engagement from mobile contexts. In Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing, pages 622–633, 2016

    2016

  8. [16]

    User engagement in mobile health applications

    Babaniyi Yusuf Olaniyi, Ana Fernández del Río, África Periáñez, and Lauren Bellhouse. User engagement in mobile health applications. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 4704–4712, 2022. 18

    2022

  9. [17]

    Engagemon: Multi-modal engagement sensing for mobile games

    Sinh Huynh, Seungmin Kim, JeongGil Ko, Rajesh Krishna Balan, and Youngki Lee. Engagemon: Multi-modal engagement sensing for mobile games. Proceedings of the ACM on interactive, mobile, wearable and ubiquitous technologies, 2(1):1–27, 2018

    2018

  10. [18]

    What and how long: Prediction of mobile app engagement

    Yuan Tian, Ke Zhou, and Dan Pelleg. What and how long: Prediction of mobile app engagement. ACM Transactions on Information Systems (TOIS), 40(1):1–38, 2021

    2021

  11. [19]

    How busy are you? predicting the interruptibility intensity of mobile users

    Fengpeng Yuan, Xianyi Gao, and Janne Lindqvist. How busy are you? predicting the interruptibility intensity of mobile users. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, pages 5346–5360, 2017

    2017

  12. [20]

    Interruptme: designing intelligent prompting mechanisms for pervasive applications

    Veljko Pejovic and Mirco Musolesi. Interruptme: designing intelligent prompting mechanisms for pervasive applications. In Proceedings of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing, pages 897–908, 2014

    2014

  13. [21]

    Smartphone interruptibility using density-weighted uncertainty sampling with reinforcement learning

    Robert Fisher and Reid Simmons. Smartphone interruptibility using density-weighted uncertainty sampling with reinforcement learning. In 2011 10th international conference on machine learning and applications and workshops, volume 1, pages 436–441. IEEE, 2011

    2011

  14. [22]

    Designing content-driven intelligent notification mechanisms for mobile applications

    Abhinav Mehrotra, Mirco Musolesi, Robert Hendley, and Veljko Pejovic. Designing content-driven intelligent notification mechanisms for mobile applications. In Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing, pages 813–824, 2015

    2015

  15. [23]

    Cognition

    Thomas A Farmer and Margaret W Matlin. Cognition. John Wiley & Sons, 2019

    2019

  16. [24]

    The principles of psychology

    William James. The principles of psychology. Henry Holt, 1890

  17. [25]

    Busybody: creating and fielding personalized models of the cost of interruption

    Eric Horvitz, Paul Koch, and Johnson Apacible. Busybody: creating and fielding personalized models of the cost of interruption. In Proceedings of the 2004 ACM conference on Computer supported cooperative work, pages 507–510, 2004

    2004

  18. [26]

    If not now, when? the effects of interruption at different moments within task execution

    Piotr D Adamczyk and Brian P Bailey. If not now, when? the effects of interruption at different moments within task execution. In Proceedings of the SIGCHI conference on Human factors in computing systems, pages 271–278, 2004

    2004

  19. [27]

    Passive remote monitoring using phone and wearable sensors to identify digital markers of adhd

    Heet Sankesara, Hayley Denyer, Shaoxiong Sun, Qigang Deng, Yatharth Ranjan, Pauline Conde, Zulqarnain Rashid, Philip Asherson, Andrea Bilbow, Maddie Groom, et al. Passive remote monitoring using phone and wearable sensors to identify digital markers of adhd

  20. [28]

    Learning to detect user activity and availability from a variety of sensor data

    Martin Muhlenbrock, Oliver Brdiczka, Dave Snowdon, and J-L Meunier. Learning to detect user activity and availability from a variety of sensor data. In Second IEEE Annual Conference on Pervasive Computing and Communications, 2004. Proceedings of the, pages 13–22. IEEE, 2004

    2004

  21. [29]

    Learning and reasoning about interruption

    Eric Horvitz and Johnson Apacible. Learning and reasoning about interruption. In Proceedings of the 5th international conference on Multimodal interfaces, pages 20–27, 2003

    2003

  22. [30]

    Notifications and awareness: a field study of alert usage and preferences

    Shamsi T Iqbal and Eric Horvitz. Notifications and awareness: a field study of alert usage and preferences. In Proceedings of the 2010 ACM conference on Computer supported cooperative work, pages 27–30, 2010

    2010

  23. [31]

    Classifying attention types with thermal imaging and eye tracking

    Yomna Abdelrahman, Anam Ahmad Khan, Joshua Newn, Eduardo Velloso, Sherine Ashraf Safwat, James Bailey, Andreas Bulling, Frank Vetere, and Albrecht Schmidt. Classifying attention types with thermal imaging and eye tracking. Proceedings of the ACM on Interactive, Mobile, Wearabl...

    2019

  24. [32]

    n-gage: Predicting in-class emotional, behavioural and cognitive engagement in the wild

    Nan Gao, Wei Shao, Mohammad Saiedur Rahaman, and Flora D Salim. n-gage: Predicting in-class emotional, behavioural and cognitive engagement in the wild. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 4(3):1–26, 2020

    2020

  25. [33]

    A multi-perspective analysis of social context and personal factors in office settings for the design of an effective mobile notification system

    Seyma Kucukozer Cavdar, Tugba Taskaya-Temizel, Mirco Musolesi, and Peter Tino. A multi-perspective analysis of social context and personal factors in office settings for the design of an effective mobile notification system. Proceedings of the ACM on Interactive, Mobile, Weara...

    2020

  26. [34]

    A mobile intelligent interruption management system

    Sina Zulkernain, Praveen Madiraju, Sheikh Iqbal Ahamed, and Karl Stamm. A mobile intelligent interruption management system. Journal of Universal Computer Science, 2010

    2010

  27. [35]

    My phone and me: understanding people’s receptivity to mobile notifications

    Abhinav Mehrotra, Veljko Pejovic, Jo Vermeulen, Robert Hendley, and Mirco Musolesi. My phone and me: understanding people’s receptivity to mobile notifications. In Proceedings of the 2016 CHI conference on human factors in computing systems, pages 1021–1032, 2016

    2016

  28. [36]

    Beyond interruptibility: Predicting opportune moments to engage mobile phone users

    Martin Pielot, Bruno Cardoso, Kleomenis Katevas, Joan Serrà, Aleksandar Matic, and Nuria Oliver. Beyond interruptibility: Predicting opportune moments to engage mobile phone users. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 1(3):1–25, 2017. 19

    2017

  29. [37]

    Understanding and managing notifications

    Swadhin Pradhan, Lili Qiu, Abhinav Parate, and Kyu-Han Kim. Understanding and managing notifications. In IEEE INFOCOM 2017-IEEE Conference on Computer Communications, pages 1–9. IEEE, 2017

    2017

  30. [38]

    Study of user interruptibility estimation based on focused application switching

    Takahiro Tanaka and Kinya Fujita. Study of user interruptibility estimation based on focused application switching. In Proceedings of the ACM 2011 conference on Computer supported cooperative work, pages 721–724, 2011

    2011

  31. [39]

    Xensible interruptions from your mobile phone

    G Henri Ter Hofte. Xensible interruptions from your mobile phone. In Proceedings of the 9th international conference on Human computer interaction with mobile devices and services, pages 178–181, 2007

    2007

  32. [40]

    Attention and engagement-awareness in the wild: A large-scale study with adaptive notifications

    Tadashi Okoshi, Kota Tsubouchi, Masaya Taji, Takanori Ichikawa, and Hideyuki Tokuda. Attention and engagement-awareness in the wild: A large-scale study with adaptive notifications. In 2017 ieee international conference on pervasive computing and communications (percom), pages...

    2017

  33. [41]

    Using electrodermal activity to recognize ease of engagement in children during social interactions

    Javier Hernandez, Ivan Riobo, Agata Rozga, Gregory D Abowd, and Rosalind W Picard. Using electrodermal activity to recognize ease of engagement in children during social interactions. In Proceedings of the 2014 ACM international joint conference on pervasive and ubiquitous com...

    2014

  34. [42]

    Measuring the engagement level of tv viewers

    Javier Hernandez, Zicheng Liu, Geoff Hulten, Dave DeBarr, Kyle Krum, and Zhengyou Zhang. Measuring the engagement level of tv viewers. In 2013 10th IEEE international conference and workshops on automatic face and gesture recognition (FG), pages 1–7. IEEE, 2013

    2013

  35. [43]

    Predicting audience responses to movie content from electro-dermal activity signals

    Fernando Silveira, Brian Eriksson, Anmol Sheth, and Adam Sheppard. Predicting audience responses to movie content from electro-dermal activity signals. In Proceedings of the 2013 ACM international joint conference on Pervasive and ubiquitous computing, pages 707–716, 2013

    2013

  36. [44]

    Experience sampling method: Measuring the quality of everyday life

    Joel M Hektner, Jennifer Anne Schmidt, and Mihaly Csikszentmihalyi. Experience sampling method: Measuring the quality of everyday life. Sage, 2007

    2007

  37. [45]

    Chatgpt for automated qualitative research: content analysis

    Rimke Bijker, Stephanie S Merkouris, Nicki A Dowling, and Simone N Rodda. Chatgpt for automated qualitative research: content analysis. Journal of medical Internet research, 26:e59050, 2024

    2024

  38. [46]

    Exploring the use of ai in qualitative analysis: A comparative study of guaranteed income data

    Leah Hamilton, Desha Elliott, Aaron Quick, Simone Smith, and Victoria Choplin. Exploring the use of ai in qualitative analysis: A comparative study of guaranteed income data. International journal of qualitative methods, 22:16094069231201504, 2023

    2023

  39. [47]

    Thematic analysis and artificial intelligence: A step-by-step process for using chatgpt in thematic analysis

    Muhammad Naeem, Tracy Smith, and Lorna Thomas. Thematic analysis and artificial intelligence: A step-by-step process for using chatgpt in thematic analysis. International Journal of Qualitative Methods , 24:16094069251333886, 2025

    2025

  40. [48]

    Safety layers in aligned large language models: The key to llm security

    Shen Li, Liuyi Yao, Lan Zhang, and Yaliang Li. Safety layers in aligned large language models: The key to llm security. arXiv preprint arXiv:2408.17003, 2024

    2024 arXiv

  41. [49]

    Double-i watermark: Protecting model copyright for llm fine-tuning

    Shen Li, Liuyi Yao, Jinyang Gao, Lan Zhang, and Yaliang Li. Double-i watermark: Protecting model copyright for llm fine-tuning. arXiv preprint arXiv:2402.14883, 2024

    2024 arXiv

  42. [50]

    Security tensors as a cross-modal bridge: Extending text-aligned safety to vision in lvlm

    Shen Li, Liuyi Yao, Wujia Niu, Lan Zhang, and Yaliang Li. Security tensors as a cross-modal bridge: Extending text-aligned safety to vision in lvlm. arXiv preprint arXiv:2507.20994, 2025

    2025 arXiv

  43. [51]

    Effects of intelligent notification management on users and their tasks

    Shamsi T Iqbal and Brian P Bailey. Effects of intelligent notification management on users and their tasks. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pages 93–102, 2008

    2008

  44. [52]

    I think it’s her: Investigating smartphone users’ speculation about phone notifications and its influence on attendance

    Yung-Ju Chang, Yi-Ju Chung, and Yi-Hao Shih. I think it’s her: Investigating smartphone users’ speculation about phone notifications and its influence on attendance. In Proceedings of the 21st International Conference on Human-Computer Interaction with Mobile Devices and Servi...

    2019

  45. [53]

    To transform or not to transform: Using generalized linear mixed models to analyse reaction time data

    Steson Lo and Sally Andrews. To transform or not to transform: Using generalized linear mixed models to analyse reaction time data. Frontiers in psychology, 6:1171, 2015

    2015

  46. [54]

    r1” indicates that the participant participated in the first round of the field study. “r2

    Stephanie Rosenthal, Anind K Dey, and Manuela Veloso. Using decision-theoretic experience sampling to build personalized mobile phone interruption models. In Pervasive Computing: 9th International Conference, Pervasive 2011, San Francisco, USA, June 12-15, 2011. Proceedings 9,...

    2011

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.