A new task, Behavior-Aware Travel Planning, which infers user preferences directly from past behaviors and generates personalized travel plans and proposes B2T-Agent, a reinforcement learning-based agent that leverages user behavior trajectories, interacts with external tools for preference-aligned retrieval, and maintains an internal memory module.
Abstract
Travel planning agents assist users in generating personalized travel plans by modeling their individual preferences. Existing agents either rely on explicit user instructions or engage in multi-turn clarification to elicit user preferences. However, both approaches overlook the rich behavioral signals latent in users'past behaviors, which implicitly encode their preferences. This over-reliance on active user input increases interaction burden and limits plan personalization. To bridge this gap, we introduce a new task, Behavior-Aware Travel Planning, which infers user preferences directly from past behaviors and generates personalized travel plans. To facilitate research on this task, we introduce Behavior2Trip, a benchmark constructed from one of the largest Chinese online travel platforms, comprising 11,400 instances. Each instance represents an average of 39.8 past user behaviors spanning 14 attributes across 5 preference dimensions. We further propose B2T-Agent, a reinforcement learning-based agent that leverages user behavior trajectories, interacts with external tools for preference-aligned retrieval, and maintains an internal memory module. Experiments on Behavior2Trip show that GPT-4.1 achieves a full-constraint pass rate of only 0.5\% on the hardest tasks, while B2T-Agent built upon Qwen3-8B outperforms all baselines, highlighting the substantial challenge of this task. Moreover, Qwen3-8B trained with B2T-Agent also outperforms GPT-4.1 on the TravelPlanner benchmark, demonstrating strong generalization. Code and data are available at https://github.com/BUAA-IRIP-LLM/Behavior2Trip
DASH is a decision-aware user simulator that jointly generates thinking traces and predicts behavioral actions from heterogeneous cross-domain histories and tailors a rubric-based reward model that evaluates thinking traces along form, content, and logic for RL training.
Urban door-to-door (D2D) mobility planning is a core task for AI-powered smart cities, requiring models to capture individual mobility behavior and generate optimized plans under real-world urban constraints such as network connectivity and service schedules. Existing methods face fundamental limitations. Optimization-based approaches rely on static costs and fail to capture individual-specific preferences. LLM-agent-based approaches often have weak spatio-temporal reasoning and unstable constraint tracking, which reduces feasibility and reliability. In this study, we propose CityWeave, a VLM-based framework for urban D2D mobility planning that integrates the Who--When--Where--How (3W1H) reasoning paradigm with a two-stage training scheme. CityWeave learns this paradigm through supervised fine-tuning and is further improved by reinforcement learning based enhancement. A dataset of 180,000 real-world samples from 80,000 users is constructed to support training and evaluation. The model learns to identify user needs (Who), reason over departure and arrival time windows (When), read maps and spatial topology (Where), and invoke routing tools (How) to generate feasible plans. We further introduce a unified User--World Grounding (UWG) module that enforces navigation-based world constraints and evaluates personalization with respect to the user profile. Extensive experiments show that CityWeave achieves a state-of-the-art Final Pass Rate of 64.7% and a Commonsense Pass Rate of 92.4%, outperforming both conventional non-LLM planning pipelines and strong LLM-agent baselines. These results demonstrate that structured reasoning over human mobility behavior, combined with explicit user and world grounding, offers a practical path toward reliable and personalized planning agents for smart urban transportation systems.
Ao Wang, Zhiwen Chen, Shen Wang et al.· Proceedings of the 32nd ACM...· 0 citations
This work introduces AcCoRD, a user-agent collaboration benchmark requiring agents to handle diverse user preference dynamics in two domains: online shopping and travel planning, and evaluates five frontier LLMs under two prompting strategies: vanilla ReAct, and an uncertainty-guided variant that prompts models to identify and resolve ambiguity about user preferences.
Agentic recommender systems increasingly employ large language model-based UserAgents to evaluate candidate items through simulated feedback before recommendations are delivered. However, existing UserAgents typically reason in isolation based on limited personal histories, which may lead to perspective narrowing: the agent evaluates candidates from a local and incomplete view, overlooks relevant preference facets, and consequently produces inaccurate judgments. A natural way to alleviate this problem is to introduce other users as advisor agents, whose diverse histories provide complementary evidence that helps the target user reconsider overlooked preference signals. Nevertheless, a generic user-advisor communication process is insufficient, as different user decision states require different forms of external advice. Based on this insight, we propose AgentCom, a personalized communication skill framework for agentic recommender systems. AgentCom organizes reusable communication skills into a shared why--what--how--who skill bank: why identifies the decision deficiency that necessitates communication, what specifies the information task, how determines the advisor interaction protocol, and who retrieves advisors capable of executing that protocol. To make the shared skill bank personalized at use time and adaptive over time, AgentCom introduces two complementary mechanisms: personalized skill routing and failure-driven skill evolution. Personalized skill routing constructs a communication path by sequentially selecting suitable skills for each user and recommendation context. Failure-driven skill evolution learns from unsuccessful communication cases and enriches the shared bank with reusable skills that address previously uncovered communication needs. Experiments show that AgentCom consistently improves recommendation performance across traditional, social, and agentic recommenders.
Zongwei Wang, Min Gao, Guang-Yu Hu et al.· 0 citations
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.
Yufeng Wang, Wei Zhang, Z. Wen et al.· 0 citations
ODYSSE is presented, a Reinforced Fine-Tuning (RFT) framework for personalized agentic reasoning designed to address long action horizons and strong cross-step dependencies in personalized agentic reasoning, and an episodic batch sampler that groups actions from the same episode into unified training batches, facilitating coherent optimization under ESPO.
