arXiv subjects

Hins Hu

Publications and source records attributed to Hins Hu.

8 recordsLinked to original sources

arXiv2026

Toward User Preference Alignment in LLM Recommendation via Explicit Context Feedback

Traditional recommender systems (RecSys) primarily infer user preferences from implicit signals (such as clicks, watches, and purchases), often neglecting the rich explicit contextual feedback users provide through verbal text, like comments and reviews. This explicit context feedback captures the nuanced reasons behind user decisions regarding their preferences. In addition, it offers critical heterogeneous information for user preference alignment and more explainable recommendations. Overlooking such signals can lead to misaligned user preferences and further reinforce filter bubbles, as algorithms fail to understand the "semantic context" behind user choices. Recent advances in Large Language Models (LLMs) present new opportunities to harness user-generated content for more accurate and diverse recommendations, yet current LLM-based recommendations still focus on using item meta-data and underutilize this resource. In this paper, we advocate for prioritizing explicit context feedback in the next generation of LLM-based RecSys. We review the evolution of recommendation paradigms, highlight the value of context-rich feedback, call for new benchmarks and metrics, and introduce frameworks for integrating explicit user signals into scalable LLM-driven RecSys. Centering on user-preference modeling, we aim to foster more personalized, transparent, and explainable RecSys online platforms.

cs.IR↗

arXiv2026

Column Generation for the Micro-Transit Zoning Problem

Along with the rapid development of new urban mobility options like ride-sharing over the past decade, on-demand micro-transit services stand out as a middle ground, bridging the gap between fixed-line mass transit and single-request ride-hailing, balancing ridership maximization and travel time minimization. Micro-transit adoption can have significant social impact. It improves urban sustainability, through lower energy consumption and reduced emissions, while enhancing equitable mobility access for disadvantaged communities, thanks to its lower vehicle miles per passenger, flexible schedules, and affordable pricing. However, effective operation of micro-transit services requires planning geo-fenced zones in advance, which involves solving a challenging combinatorial optimization problem. Existing approaches enumerate candidate zones first and selects a fixed number of optimal zones in the second step. In this paper, we generalize the Micro-Transit Zoning Problem (MZP) to allow a global budget rather than imposing a size limit for candidate zones. We also design a Column Generation (CG) framework to solve the problem and several pricing heuristics to accelerate computation. Extensive numerical experiments across major U.S. cities demonstrate that our approach produces higher-quality solutions more efficiently and scales better in the generalized setting.

math.OC↗

arXiv2025

Optimal Micro-Transit Zoning via Clique Generation and Integer Programming

Micro-transit services offer a promising solution to enhance urban mobility and access, particularly by complementing existing public transit. However, effectively designing these services requires determining optimal service zones for these on-demand shuttles, a complex challenge often constrained by operating budgets and transit agency priorities. This paper presents a novel two-phase algorithmic framework for designing optimal micro-transit service zones based on the objective of maximizing served demand. A key innovation is our adaptation of the shareability graph concept from its traditional use in dynamic trip assignment to the distinct challenge of static spatial zoning. We redefine shareability by considering geographical proximity within a specified diameter constraint, rather than trip characteristics. In Phase 1, the framework employs a highly scalable algorithm to generate a comprehensive set of candidate zones. In Phase 2, it formulates the selection of a specified number of zones as a Weighted Maximum Coverage Problem, which can be efficiently solved by an integer programming solver. Evaluations on real-world data from Chattanooga, TN, and synthetic datasets show that our framework outperforms a baseline algorithm, serving 27.03% more demand in practice and up to 49.5% more demand in synthetic settings.

math.OC↗

arXiv2025

HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial Optimization

While Large Language Models (LLMs) have demonstrated significant advancements in reasoning and agent-based problem-solving, current evaluation methodologies fail to adequately assess their capabilities: existing benchmarks either rely on closed-ended questions prone to saturation and memorization, or subjective comparisons that lack consistency and rigor. In this work, we introduce HeuriGym, an agentic framework designed for evaluating heuristic algorithms generated by LLMs for combinatorial optimization problems, characterized by clearly defined objectives and expansive solution spaces. HeuriGym empowers LLMs to propose heuristics, receive evaluative feedback via code execution, and iteratively refine their solutions. We evaluate nine state-of-the-art models on nine problems across domains such as computer systems, logistics, and biology, exposing persistent limitations in tool use, planning, and adaptive reasoning. To quantify performance, we propose the Quality-Yield Index (QYI), a metric that captures both solution pass rate and quality. Even top models like GPT-o4-mini-high and Gemini-2.5-Pro attain QYI scores of only 0.6, well below the expert baseline of 1. Our open-source benchmark aims to guide the development of LLMs toward more effective and realistic problem-solving in scientific and engineering domains.

cs.LG↗

arXiv2025

Ride-pool Assignment Algorithms: Modern Implementation and Swapping Heuristics

On-demand ride-pooling has emerged as a popular urban transportation solution, addressing the efficiency limitations of traditional ride-hailing services by grouping multiple riding requests with spatiotemporal proximity into a single vehicle. Although numerous algorithms have been developed for the Ride-pool Assignment Problem (RAP) -- a core component of ride-pooling systems, there is a lack of open-source implementations, making it difficult to benchmark these algorithms on a common dataset and objective. In this paper, we present the implementation details of a ride-pool simulator that encompasses several key ride-pool assignment algorithms, along with associated components such as vehicle routing and rebalancing. We also open-source a highly optimized and modular C++ codebase, designed to facilitate the extension of new algorithms and features. Additionally, we introduce a family of swapping-based local-search heuristics to enhance existing ride-pool assignment algorithms, achieving a better balance between performance and computational efficiency. Extensive experiments on a large-scale, real-world dataset from Manhattan, NYC reveal that while all selected algorithms perform comparably, the newly proposed Multi-Round Linear Assignment with Cyclic Exchange (LA-MR-CE) algorithm achieves a state-of-the-art service rate with significantly reduced computational time. Furthermore, an in-depth analysis suggests that a performance barrier exists for all myopic ride-pool assignment algorithms due to the system's capacity bottleneck, and incorporating future information could be key to overcoming this limitation.

cs.AI↗

arXiv2025

Vehicle Routing Problems in the Age of Semi-Autonomous Driving

We are in the midst of a semi-autonomous era in urban transportation in which varying forms of vehicle autonomy are gradually being introduced. This phase of partial autonomy is anticipated by some to span a few decades due to various challenges, including budgetary constraints to upgrade the infrastructure and technological obstacles in the deployment of fully autonomous vehicles (AV) at scale. In this study, we introduce the vehicle routing problem in a semi-autonomous environment (VRP-SA) where the road network is not fully AV-enabled in the sense that a portion of it is either not suitable for AVs or requires additional resources in real-time (e.g., remote control) for AVs to pass through. Moreover, such resources are scarce and usually subject to a budget constraint. An exact mixed-integer linear program (MILP) is formulated to minimize the total routing cost of service in this environment. We propose a two-phase algorithm based on a family of feasibility recovering sub-problems (FRP) to solve the VRP-SA efficiently. Our algorithm is implemented and tested on a new set of instances that are tailored for the VRP-SA by adding stratified grid road networks to the benchmark instances. The result demonstrates a reduction of up to 37.5% in vehicle routing costs if the fleet actively exploits the AV-enabled roads in the environment. Additional analysis reveals that cost reduction is higher with more budget and longer operational hours.

math.OC↗

arXiv2024

Integrating On-demand Ride-sharing with Mass Transit at-Scale

We are in the midst of a technology-driven transformation of the urban mobility landscape. However, unfortunately these new innovations are still dominated by car-centric personal mobility, which leads to concerns such as environmental sustainability, congestion, and equity. On the other hand, mass transit provides a means to move large amounts of travelers very efficiently, but is not very versatile and depends on an adequate concentration of demand. In this context, our overarching goal is to explore opportunities for new technologies such as ride-sharing to integrate with mass transit and provide a better service. More specifically, we envision a hybrid system that uses on-demand shuttles in conjunction with mass transit to move passengers efficiently, and provide an algorithmic framework for operational optimization. Our approach extends a state-of-the-art trip-vehicle assignment model to the multi-modal setting, where we develop a new integer-linear programming formulation to solve the problem efficiently. A comprehensive study covering five major cities in the United States based on real-world data is carried out to verify the advantages of such a system and the effectiveness of our algorithms. We show that our hybrid system provides significant improvements in comparison to a purely on-demand model by exploiting the efficiencies of the mass transit system.

cs.DM↗

arXiv2022

Cyber Risk Assessment for Capital Management

This paper introduces a two-pillar cyber risk management framework to address the pervasive challenges in managing cyber risk. The first pillar, cyber risk assessment, combines insurance frequency-severity models with cybersecurity cascade models to capture the unique nature of cyber risk. The second pillar, cyber capital management, facilitates informed allocation of capital for a balanced cyber risk management strategy, including cybersecurity investments, insurance coverage, and reserves. A case study, based on historical cyber incident data and realistic assumptions, demonstrates the necessity of comprehensive cost-benefit analysis for budget-constrained companies with competing objectives in cyber risk management. In addition, sensitivity analysis highlights the dependence of the optimal strategy on factors such as the price of cybersecurity controls and their effectiveness. The framework's implementation across a diverse range of companies yields general insights on cyber risk management.

q-fin.RM↗