Pith. sign in

REVIEW 4 major objections 4 minor 61 references

Preference-Driven Online Adaptation for Personalized Interaction Initiation in Proactive AI Assistants

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Counting feedback beats LLM reflection for proactive assistant timing.

desk verdict EOPA is a coherent, well-ablated method for timing personalization that achieves large simulated gains, but its headline result depends on full-information feedback on silence steps that real users don't provide. read the letter →

arxiv 2608.04416 v1 pith:VGD5G3JA submitted 2026-08-05 cs.HC

Yufeng Wang , Wei Zhang , Zhiquan Wen , Jinwu Hu , Linhui Xiao , Tianlu Pan , Qingfang Zheng , Mingkui Tan This is my paper · ORCID
classification cs.HC
keywords proactiveAIassistantsinteractiontimingonlineadaptationuserfeedbackcontextualevidencetemporalpreferenceanchorsactivityprototypespersonalization
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 proactive assistant can personalize when it initiates interactions by converting online user feedback into countable contextual evidence, rather than by fine-tuning model parameters or by reflecting on past trajectories in natural language. It introduces EOPA, an evidence-driven online preference adaptation method, which keeps two evidence carriers: temporal preference anchors that count interaction-appropriate and silence-preferred outcomes by time-of-day and activity-progress stage, and evidence-bearing activity prototypes that carry the same counts under semantic activity clusters. At each polling step, the two evidence sources are turned into probability estimates smoothed by a global user prior, scaled by how uncertain they are, and fused with an adaptively calibrated threshold to decide between interacting and staying silent. In a 14-day simulated benchmark with 32 user personas and 160,632 polling steps, EOPA reports an interaction-timing F1 of 30.46, about 19.80 points above the strongest baseline, while cutting average silence-step latency from 688 ms to 1.96 ms and daily adaptation time from 11.41 to 0.39 seconds. If correct, this makes fine-grained, continuously updated timing personalization practical without invoking an LLM at every step.

What carries the argument

The load-bearing object is the evidence score built from two complementary counters. Temporal preference anchors $A_{i,r}=(N^{\mathrm{INT}}_{i,r},N^{\mathrm{SIL}}_{i,r})$ count interaction-appropriate and silence-preferred labels for each time interval $i$ and activity-progress stage $r$; evidence-bearing activity prototypes similarly store semantic centers plus progress-conditioned counts. Each branch converts its counts into a posterior probability using a user-prior-smoothed Beta-Binomial estimate, e.g. $p_{\mathrm{temp},t}=(N^{\mathrm{INT}}_{i,r}+\kappa p^u_t)/(N^{\mathrm{INT}}_{i,r}+N^{\mathrm{SIL}}_{i,r}+\kappa)$, divides by the posterior standard deviation plus a small constant, and the two scaled evidence scores are summed as $s_t=e_{\mathrm{temp},t}+\lambda_t e_{\mathrm{act},t}$ and compared with an adaptive threshold $\delta_t$. The mechanism preserves fine-grained differences among similar states because each state draws on its own matched anchor and prototype statistics rather than on a global rule; the adaptive $\lambda_t$ and $\delta_t$ replay recent history to keep the decision boundary aligned with the user.

What would settle it

Run EOPA online with real users but collect feedback only on interaction steps, leaving all silence steps unlabeled, and compare timing F1 against the simulated full-feedback setting; if the claimed gains require accurate silence counts, F1 should fall well below the reported 30.46, and precision or recall should shift noticeably as the unlabeled-silence bias accumulates.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that interaction-timing preferences are better treated as measurable contextual evidence than as textual or parametric experience. EOPA stores interaction and silence counts under fine-grained temporal and activity-semantic conditions; user-prior smoothing stabilizes sparse local counts, uncertainty-guided scaling lets better-supported evidence speak louder, and adaptive fusion with an online-calibrated threshold lets the decision policy itself adjust to the user. Response generation is decoupled: an LLM is called only when interaction is selected, using high-quality historical responses as demonstrations. The reported consequence is that the same accumulated feedback that outperforms LLM-based self-improvement methods on timing also yields shorter silence decisions and nearly zero adaptation cost, because no weight updates or self-reflection calls are needed.

Load-bearing premise

The load-bearing assumption is that after every polling step, including steps where the assistant stayed silent, the assistant receives a binary timing label; in real use, feedback is usually given only when the assistant actually interacts, so the silence counts that the evidence scores depend on may be missing or systematically biased.

Editorial extensions

If this is right

  • Silence decisions become nearly free: EOPA's reported average silence-step latency is 1.96 ms versus 688 ms for an LLM-reflection method, because keeping silent requires no LLM call.
  • Personalization can continue all day without a training budget: daily adaptation drops from 11.41 s to 0.39 s, so the assistant can re-calibrate after every feedback signal rather than batching updates.
  • Fine-grained distinctions survive: on state pairs that share time interval, progress stage, or activity description but differ in preferred timing, EOPA's average decision accuracy is 38.60 percent versus 22.72 percent for the strongest baseline, supporting the claim that countable evidence preserves differences that textual reflections blur.
  • Moderate feedback sparsity is tolerable: retaining only 75 percent or 50 percent of the binary feedback keeps 89.6 percent and 83.9 percent of the full-feedback F1, and even 25 percent feedback outperforms the full-feedback LLM-reflection baseline.
  • Timing and content become separable problems: the response-generation LLM runs only when interaction is selected, so response quality can be optimized independently of the timing policy.

Reading between the lines

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

  • A natural next test is to drop the assumption of a label at every silence step and treat unobserved silences as missing data; the paper's 25-percent-feedback result hints the method degrades gradually, but real users' silence feedback would likely be systematically missing, not random.
  • The same evidence-carrier design could be lifted to other sparse-decision domains such as notification scheduling, interruption management, or content moderation, where the cost of a false positive is high and non-event feedback is scarce.
  • Because the labels in the benchmark are model-generated and the personas are simulated, the 19.80-point gain is evidence about the mechanism's discriminative power under clean labels; a field deployment would test whether human feedback about silence is obtainable enough to feed the counters.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes EOPA, an evidence-driven online adaptation method for deciding when a proactive AI assistant should initiate an interaction. EOPA maintains two evidence carriers, temporal preference anchors and evidence-bearing activity prototypes, which accumulate interaction-appropriate and silence-preferred counts from user feedback. At each polling step, it estimates temporal and activity evidence via user-prior-smoothed Beta-Binomial posteriors with uncertainty-guided scaling, fuses them with an adaptive weight, and compares the sum to an online-calibrated threshold. Response generation is decoupled from timing decisions and only invoked when an interaction is selected. On a ProPerSim-based benchmark with 32 simulated user personas over 14 days, EOPA achieves 30.46 interaction-timing F1, improves over the strongest LLM-based baseline by 19.80 F1 points, reduces silence-step latency from 688 ms to 1.96 ms, and lowers daily adaptation time from 11.41 s to 0.39 s. Incremental ablations, partial-feedback robustness, evidence-discriminability analysis, and similar-state decision accuracy are provided.

Significance. The contribution is timely and practically motivated. A decisive strength is that timing decisions are made by deterministic evidence statistics rather than by an LLM, giving large latency and adaptation-time reductions while keeping response generation decoupled. The incremental ablations (Table 2), the evidence-discriminability curves (Table 4), and the similar-state decision accuracy analysis (Table 5) are well-designed and internally consistent. If the feedback-availability assumption is repaired, the method offers a credible lightweight alternative to LLM self-reflection for proactive assistants. The main limitation on significance is external validity: the benchmark protocol supplies timing labels at every step, and the paper does not yet show that the strong F1 gain survives realistic missingness of silence feedback. The skeptical concern about structured label missingness is therefore load-bearing and should be addressed before the headline gain is interpreted as a deployable-policy result.

major comments (4)
  1. [Section 5.1; Eqs. (8) and (15)] The online protocol assumes that after each decision the assistant receives user feedback from which a timing label y_t in {INT, SIL} is derived. In a deployed proactive assistant, a user who was not interrupted gives no explicit feedback, and missed opportunities (steps whose correct label is INT but where EOPA chose SIL) also produce no label. The counters N^SIL and H^SIL, and therefore the estimates p_temp and p_act in Eqs. (5) and (12), would not grow as modeled under realistic feedback. Table 3 randomly drops a fraction of all labels, which does not reproduce missingness concentrated on SIL-labeled and false-negative steps. This is load-bearing because the claimed 19.80-point F1 gain is obtained under the full-information protocol. Please add a structured-missingness experiment in which (i) SIL steps receive no label, or (ii) labels are available only for steps where the assistant interacted, and report the resulting F1 and adaptation curves. If the method relies on imputed or synthetic SIL labels, state that assumption explicitly as a limitation of the deployment claim.
  2. [Section 4.3; Section 5.2] The adaptive fusion weight lambda_t and decision threshold delta_t are calibrated by replaying decisions under candidate parameter pairs and optimizing F1 on recent historical data. The same accumulated stream is subsequently scored for the reported per-user metrics, so the evaluation does not cleanly separate the calibration signal from the test signal. Please state explicitly that each day's lambda_t and delta_t are computed only from steps strictly before the evaluation window, and ideally report a walk-forward variant in which the parameters are frozen before the target day. Without this clarification, the 30.46 F1 number could partly reflect tuning to the evaluated stream rather than a deployable online policy.
  3. [Section 5.1; Appendix I] The results depend on several hand-set hyperparameters (kappa=2.5, K=12, gamma=0.1, rho=0.96, number of time intervals B, and progress-index cap R). The manuscript states that hyperparameter analysis is in Appendix I, but no appendix is included in the provided text. Because EOPA's advantage over baselines is large and concentrated in the timing branch, the absence of the promised sensitivity curves leaves open the possibility that performance is peaked at the reported settings. Please include the appendix or add a sensitivity table in the main text, reporting the F1 range across reasonable values of kappa, K, gamma, rho, and B.
  4. [Section 5.2] The comparison set contains only an LLM parameter-learning method and LLM-based self-improvement baselines. A simple non-LLM baseline, such as always-silence, interaction at the most frequent time intervals for each persona, or a threshold on the global user prior, would clarify whether the 30.46 F1 reflects effective use of feedback or simply the difficulty of the class-imbalanced task. This is inexpensive to add and would strengthen the claim that the evidence carriers, rather than the mere absence of an LLM for timing decisions, drive the observed gain.
minor comments (4)
  1. [Figures 3 and 4] Parts of the figure captions and labels render as uninterpretable glyph sequences (for example, '/uni00000014/uni00000015/...'); please replace these with readable textual labels in the final version.
  2. [Table 2] The row headers use dashes to denote disabled components without a legend; adding check/cross marks or explicit component names would make the incremental ablation unambiguous.
  3. [Section 5.1] The definitions of F1@1 and F1@2 count predictions within one and two polling steps of an annotated interaction moment; please clarify how duplicate matches are handled when one prediction is near two annotated moments, and whether precision and recall are computed per moment or per prediction.
  4. [Equation (16)] Equation (16) introduces the fusion weight lambda_t but the main text does not state its initialization, positivity constraint, or allowed range; these details are presumably in the missing appendix and should be stated in the main text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: EOPA is a causal online learning procedure whose predictions are evaluated against external benchmark timing labels, not against its own fitted outputs.

full rationale

EOPA's derivation chain is self-contained and causal with respect to the benchmark protocol. At each polling step the decision uses only statistics accumulated from previous steps: Eqs. (3), (5), (7), (9)-(14), and (16) are functions of past interaction/silence counts and a global prior, and the counts are updated after the decision using the externally supplied timing label (Eqs. (8) and (15)). The label y_t is not defined in terms of the model's output; it comes from the ProPerSim-based simulated user and the separately constructed annotation. The F1 evaluation compares the model's decisions to these same external labels, which is the standard protocol for an online learner evaluated on a stream. The fusion weight lambda_t and threshold delta_t are described as being calibrated by replaying historical decisions and choosing parameters that improve historical F1; nothing in Section 4.3 or the experiments indicates that future labels are used to make a past decision, so the calibration is online model selection rather than in-sample prediction. The partial-feedback robustness study (Table 3) randomly drops feedback labels, and the paper explicitly notes users may not provide feedback after every decision; this is a limitation about structured missingness in real deployments, but it is an external validity concern, not a circular derivation. There are no load-bearing self-citations: the paper builds on ProPerSim, Beta-Binomial estimation, and prototype learning as independent prior work, and no uniqueness or ansatz is imported from the authors' own prior publications. The performance claims therefore do not reduce by construction to fitted constants or to the paper's own definitions.

Assumptions & free parameters 8 free parameters · 5 assumptions · 2 invented entities

The central claim relies on the availability of unbiased per-step feedback labels, on the assumption that LLM-generated labels approximate real user preferences, and on several hand-chosen hyperparameters. The evidence carriers are algorithmic constructs with no external falsifiable handle beyond the benchmark.

free parameters (8)
  • lambda_t (activity evidence weight) = adapted online via grid search
    Controls the contribution of activity evidence in Eq. (16); calibrated to improve F1 on recent history (Section 4.3).
  • delta_t (decision threshold) = adapted online via grid search
    Determines interaction versus silence in Eq. (17); calibrated jointly with lambda_t.
  • kappa (prior smoothing strength) = 2.5
    Set in Section 5.1 for both evidence branches; controls shrinkage toward the global user prior in Eqs. (5) and (12).
  • K (retrieved prototype count) = 12
    Number of nearest activity prototypes aggregated in Eq. (11).
  • gamma (prototype temperature) = 0.1
    Softmax temperature in Eq. (10).
  • rho (prototype creation threshold) = 0.96
    Similarity threshold for creating a new prototype (Section 4.2).
  • B (time interval count) = 48 (30-minute intervals)
    Discretization of the day for temporal anchors (Section 5.1).
  • R (progress index cap) = not specified
    Maximum number of progress stages for anchors and prototypes; value is omitted in the main text.
assumptions (5)
  • standard math Beta-binomial conjugacy with Jeffreys prior
    Used to derive the smoothed estimate in Eq. (5) and variance in Eq. (6), following Gelman et al. (2013).
  • domain assumption Timing feedback labels are available for all polling steps and are unbiased
    Section 5.1 states the assistant receives binary timing feedback at each step; Table 3 only drops labels randomly, keeping them unbiased.
  • domain assumption LLM-generated timing labels approximate real user preferences
    The fixed labels are produced by Qwen3.5-397B-A17B conditioned on ProPerSim persona preferences (Section 5.1).
  • domain assumption User timing preferences are consistent within time-progress bins and activity-semantic prototypes
    The entire EOPA method assumes these bins capture meaningful variation in interaction appropriateness (Sections 4.1 and 4.2).
  • ad hoc to paper The evidence score e = p / sqrt(sigma^2 + epsilon) is a valid measure of interaction appropriateness
    Eqs. (7) and (14) define this scaling; it is a heuristic t-statistic-like transformation, not derived from a formal decision objective.
invented entities (2)
  • Temporal preference anchors
    purpose: Store interaction and silence counts per time interval and activity progress stage (Eq. 3).
    Internal data structure evaluated only on the paper's benchmark; no outside falsifiable prediction.
  • Evidence-bearing activity prototypes
    purpose: Store semantic centers plus interaction and silence counts per progress stage (Eq. 9).
    Same as above; the semantic centers are updated online and the statistics support evidence estimation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Preference-Driven Online Adaptation for Personalized Interaction Initiation in Proactive AI Assistants." pith.science (2026). https://pith.science/paper/VGD5G3JA

@misc{pith2026260804416,
  author       = {Pith},
  title        = {Pith review of: Preference-Driven Online Adaptation for Personalized Interaction Initiation in Proactive AI Assistants},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VGD5G3JA}},
  note         = {Machine review of arXiv:2608.04416}
}
read the original abstract

AI assistants are typically reactive, relying on users to initiate interactions. Proactive assistants go beyond this paradigm by autonomously initiating interactions based on users' activity contexts. However, appropriate interaction timing is user-specific and difficult to determine in advance, while online feedback offers valuable signals for personalization. Direct feedback-driven adaptation is therefore appealing, but remains challenging due to sparse interaction-worthy moments scattered across fine-grained user states. To address the issues, we propose Evidence-driven Online Preference Adaptation (EOPA), which grounds a user's interaction-timing preferences in measurable contextual evidence through two evidence carriers: temporal preference anchors and evidence-bearing activity prototypes. At each polling step, EOPA derives temporal and activity evidence from the carriers through user-prior-smoothed evidence estimation and uncertainty-guided evidence scaling, and adaptively fuses the evidence for interaction-or-silence decisions. When interaction is selected, an LLM uses high-quality historical responses as demonstrations to generate a context-aware response that better reflects user preferences. EOPA updates its evidence carriers and decision parameters from received online feedback without LLM-based reasoning or retraining. Extensive experiments on a ProPerSim-based benchmark show that EOPA improves the interaction-timing F1 score by 19.80 points over the strongest baseline in our experiments, substantially reduces inference latency for both silence and interaction steps, and lowers the average daily adaptation time from 11.41 to 0.39 seconds.

Figures

Figures reproduced from arXiv: 2608.04416 by the authors.

Figure 1
Figure 1. Comparison between reactive AI assistants and [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. EOPA grounds a user’s interaction-timing preferences in measurable contextual evidence from two carriers: temporal [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. User-averaged daily F1 over the 14-day online [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Inference latency and daily adaptation time cost. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 37 canonical work pages

  1. [1]

    Nature , volume=

    Towards conversational diagnostic artificial intelligence , author=. Nature , volume=. 2025 , publisher=

    work page 2025

  2. [2]

    Journal of Educational Data Mining , volume=

    Designing safe and relevant generative chats for math learning in intelligent tutoring systems , author=. Journal of Educational Data Mining , volume=

    work page

  3. [3]

    Nature medicine , volume=

    Closing the accessibility gap to mental health treatment with a personalized self-referral chatbot , author=. Nature medicine , volume=. 2024 , publisher=

    work page 2024

  4. [4]

    Scientific Reports , year=

    A behaviour-adaptive AI assistant enhancing accessibility and usability for blind users through real-time interaction personalization , author=. Scientific Reports , year=

    work page

  5. [5]

    2025 IEEE International Conference on Interdisciplinary Approaches in Technology and Management for Social Innovation (IATMSI) , volume=

    Natural language processing for conversational AI: Chatbots and virtual assistants , author=. 2025 IEEE International Conference on Interdisciplinary Approaches in Technology and Management for Social Innovation (IATMSI) , volume=. 2025 , organization=

    work page 2025

  6. [6]

    ContextAgent: Context-Aware Proactive LLM Agents with Open-world Sensory Perceptions , volume =

    Yang, Bufang and Xu, Lilin and Zeng, Liekang and Liu, Kaiwei and Jiang, Siyang and Lu, Wenrui and Chen, Hongkai and Jiang, Xiaofan and Xing, Guoliang and Yan, Zhenyu , booktitle =. ContextAgent: Context-Aware Proactive LLM Agents with Open-world Sensory Perceptions , volume =

    work page

  7. [7]

    The Thirteenth International Conference on Learning Representations , year=

    Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance , author=. The Thirteenth International Conference on Learning Representations , year=

    work page

  8. [8]

    ProPerSim: Developing Proactive and Personalized

    Jiho Kim and Junseong Choi and Woosog Chay and Daeun Kyung and Yeonsu Kwon and Yohan Jo and Edward Choi , booktitle=. ProPerSim: Developing Proactive and Personalized

    work page

Show all 61 references
  1. [9]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Expel: Llm agents are experiential learners , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  2. [10]

    arXiv preprint arXiv:2512.15374 , year=

    Scope: Prompt evolution for enhancing agent effectiveness , author=. arXiv preprint arXiv:2512.15374 , year=

    arXiv

  3. [11]

    Bert: Pre-training of deep bidirectional transformers for language understanding , author=. Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers) , pages=

    2019

  4. [12]

    Journal of the American statistical Association , volume=

    Parametric empirical Bayes inference: theory and applications , author=. Journal of the American statistical Association , volume=. 1983 , publisher=

    1983

  5. [13]

    Proceedings of the Royal Society of London

    An invariant form for the prior probability in estimation problems , author=. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences , volume=. 1946 , publisher=

    1946

  6. [14]

    Proceedings of the 36th annual acm symposium on user interface software and technology , pages=

    Generative agents: Interactive simulacra of human behavior , author=. Proceedings of the 36th annual acm symposium on user interface software and technology , pages=

  7. [15]

    The Fourteenth International Conference on Learning Representations , year=

    EvoTest: Evolutionary Test-Time Learning for Self-Improving Agentic Systems , author=. The Fourteenth International Conference on Learning Representations , year=

  8. [16]

    Sentence-bert: Sentence embeddings using siamese bert-networks , author=. Proceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP) , pages=

    2019

  9. [17]

    Advances in neural information processing systems , volume=

    Reflexion: Language agents with verbal reinforcement learning , author=. Advances in neural information processing systems , volume=

  10. [18]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Graph-grounded goal planning for conversational recommendation , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2022 , publisher=

    2022

  11. [19]

    Digest of Papers

    The conference assistant: Combining context-awareness with wearable computing , author=. Digest of Papers. Third International Symposium on Wearable Computers , pages=. 1999 , organization=

    1999

  12. [20]

    Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence , pages=

    Towards proactive interactions for in-vehicle conversational assistants utilizing large language models , author=. Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence , pages=

  13. [21]

    Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems , pages=

    Proactive conversational agents with inner thoughts , author=. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems , pages=

    2025

  14. [22]

    Proceedings of the 33rd ACM International Conference on the Foundations of Software Engineering , pages=

    Codinggenie: A proactive llm-powered programming assistant , author=. Proceedings of the 33rd ACM International Conference on the Foundations of Software Engineering , pages=

  15. [23]

    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies , volume=

    Socialmind: Llm-based proactive ar social assistive system with human-like perception for in-situ live interactions , author=. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies , volume=. 2025 , publisher=

    2025

  16. [24]

    arXiv preprint arXiv:2512.06721 , year=

    ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems , author=. arXiv preprint arXiv:2512.06721 , year=

    arXiv

  17. [25]

    arXiv preprint arXiv:2601.22964 , year=

    EvoClinician: A Self-Evolving Agent for Multi-Turn Medical Diagnosis via Test-Time Evolutionary Learning , author=. arXiv preprint arXiv:2601.22964 , year=

  18. [26]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    A survey of context-aware recommender systems: from an evaluation perspective , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2022 , publisher=

    2022

  19. [27]

    Advances in Neural Information Processing Systems , volume=

    Improving context-aware preference modeling for language models , author=. Advances in Neural Information Processing Systems , volume=

  20. [28]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    An adaptive, context-aware, and stacked attention network-based recommendation system to capture users’ temporal preference , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2022 , publisher=

    2022

  21. [29]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Continuous-time user preference modelling for temporal sets prediction , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2023 , publisher=

    2023

  22. [30]

    Leveraging Long Short-Term User Preference in Conversational Recommendation via Multi-agent Reinforcement Learning , year=

    Deng, Yang and Li, Yaliang and Ding, Bolin and Lam, Wai , journal=. Leveraging Long Short-Term User Preference in Conversational Recommendation via Multi-agent Reinforcement Learning , year=

  23. [31]

    Proceedings of the ACM Web Conference 2024 , pages=

    Rella: Retrieval-enhanced large language models for lifelong sequential behavior comprehension in recommendation , author=. Proceedings of the ACM Web Conference 2024 , pages=

    2024

  24. [32]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Improving conversational recommender system via contextual and time-aware modeling with less domain-specific knowledge , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2024 , publisher=

    2024

  25. [33]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Improving conversational recommendation system through personalized preference modeling and knowledge graph , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2024 , publisher=

    2024

  26. [34]

    Proceedings of the 32nd ACM International Conference on Multimedia , pages=

    Tut4crs: Time-aware user-preference tracking for conversational recommendation system , author=. Proceedings of the 32nd ACM International Conference on Multimedia , pages=

  27. [35]

    Proceedings of the 33rd ACM International Conference on Information and Knowledge Management , pages=

    Memocrs: Memory-enhanced sequential conversational recommender systems with large language models , author=. Proceedings of the 33rd ACM International Conference on Information and Knowledge Management , pages=

  28. [36]

    Empowering retrieval-based conversational recommendation with contrasting user preferences , author=. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)...

    2025

  29. [37]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    User preference learning for online social recommendation , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2016 , publisher=

    2016

  30. [38]

    Proceedings of the 32nd ACM International Conference on Information and Knowledge Management , pages=

    Cpmr: Context-aware incremental sequential recommendation with pseudo-multi-task learning , author=. Proceedings of the 32nd ACM International Conference on Information and Knowledge Management , pages=

  31. [39]

    IEEE Transactions on knowledge and data engineering , volume=

    Learning from imbalanced data , author=. IEEE Transactions on knowledge and data engineering , volume=. 2009 , publisher=

    2009

  32. [40]

    2013 , publisher=

    Bayesian Data Analysis , author=. 2013 , publisher=

    2013

  33. [41]

    Breakthroughs in Statistics: Foundations and basic theory , pages=

    An empirical Bayes approach to statistics , author=. Breakthroughs in Statistics: Foundations and basic theory , pages=. 1992 , publisher=

    1992

  34. [42]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Online prototype learning for online continual learning , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  35. [43]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Robust prototype-based learning on data streams , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2017 , publisher=

    2017

  36. [44]

    International Symposium on Handheld and Ubiquitous Computing , pages=

    Cybreminder: A context-aware system for supporting reminders , author=. International Symposium on Handheld and Ubiquitous Computing , pages=. 2000 , organization=

    2000

  37. [45]

    , author=

    SenSay: A Context-Aware Mobile Phone. , author=. ISWC , volume=

  38. [46]

    Proceedings of the SIGCHI conference on Human factors in computing systems , pages=

    Using context-aware computing to reduce the perceived burden of interruptions from mobile devices , author=. Proceedings of the SIGCHI conference on Human factors in computing systems , pages=

  39. [47]

    Journal of Cognitive Engineering and Decision Making , volume=

    Considering etiquette in the design of an adaptive system , author=. Journal of Cognitive Engineering and Decision Making , volume=. 2012 , publisher=

    2012

  40. [48]

    Advances in Neural Information Processing Systems , volume=

    Aligning to thousands of preferences via system message generalization , author=. Advances in Neural Information Processing Systems , volume=

  41. [49]

    Advances in Neural Information Processing Systems , volume=

    Personalizing reinforcement learning from human feedback with variational preference learning , author=. Advances in Neural Information Processing Systems , volume=

  42. [50]

    Advances in neural information processing systems , volume=

    Aligning llm agents by learning latent preference from user edits , author=. Advances in neural information processing systems , volume=

  43. [51]

    BMC medical research methodology , volume=

    A comparison of Cohen’s Kappa and Gwet’s AC1 when calculating inter-rater reliability coefficients: a study conducted with personality disorder samples , author=. BMC medical research methodology , volume=. 2013 , publisher=

    2013

  44. [52]

    arXiv preprint arXiv:2602.15763 , year=

    Glm-5: from vibe coding to agentic engineering , author=. arXiv preprint arXiv:2602.15763 , year=

    arXiv

  45. [53]

    arXiv preprint arXiv:2606.19348 , year=

    Deepseek-v4: Towards highly efficient million-token context intelligence , author=. arXiv preprint arXiv:2606.19348 , year=

  46. [54]

    Information Systems Frontiers , volume=

    Charting the evolution and future of conversational agents: A research agenda along five waves and new frontiers , author=. Information Systems Frontiers , volume=. 2024 , publisher=

    2024

  47. [55]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Streaming dense video captioning , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  48. [56]

    arXiv preprint arXiv:2505.09388 , year=

    Qwen3 technical report , author=. arXiv preprint arXiv:2505.09388 , year=

    arXiv

  49. [57]

    Journal of Image and Graphics , pages=

    A survey on multimodal real-time interactive digital humans , author=. Journal of Image and Graphics , pages=

  50. [58]

    Machine Intelligence Research , volume=

    Challenges and trends in egocentric vision: A survey , author=. Machine Intelligence Research , volume=. 2026 , publisher=

    2026

  51. [59]

    Thirty-seventh Conference on Neural Information Processing Systems , year=

    Direct Preference Optimization: Your Language Model is Secretly a Reward Model , author=. Thirty-seventh Conference on Neural Information Processing Systems , year=

  52. [60]

    Biometrics bulletin , volume=

    Individual comparisons by ranking methods , author=. Biometrics bulletin , volume=. 1945 , publisher=

    1945

  53. [61]

    Scandinavian journal of statistics , pages=

    A simple sequentially rejective multiple test procedure , author=. Scandinavian journal of statistics , pages=. 1979 , publisher=

    1979

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.