Queueing Network-Model Human Processor (QN-MHP): A Computational Model and a Simulation Technology for Analyzing Human Multitask Performance in Human-Computer Systems

人类行为的综合计算模型对人机系统设计具有重要的科学意义和实用价值。该项目将在HCI建模领域做出重大贡献。PI将扩展他一直在开发的计算模型和称为排队网络模型人类处理器（QN-MHP）的仿真技术（实现）。QN-MHP是独特的，因为它集成了两种互补的认知建模方法：过程和生产系统方法（例如MHP/GOMS系列模型，ACT-R， CAPS， EPIC和SOAR），以及排队网络方法。程序和生产系统模型在建模方面取得了巨大的成功，并在生成一个人在执行各种任务时可能采取的详细程序和行动方面取得了巨大的成功；他们的缺点是，虽然他们使用数学来分析模型的特定方面，但他们缺乏数学理论来表示模型的整体结构。排队网络模型能够将大量有影响的心理结构数学模型作为特殊情况集成（如Sternberg的系列阶段模型、McClelland的级联模型、Schweikert的关键路径网络模型），一般适合于对动态复杂的任务和过程的体系结构安排进行建模；然而，作为一个单独的数学理论，排队网络不能用于生成人在特定任务情况下的详细行动，缺乏人在完成其特定目标时可能使用的程序知识。QN-MHP通过将MHP的三个离散串行阶段扩展为队列网络的三个连续传输子网，通过使用过程函数定义每个服务器，以及使用goms风格的任务分析方法，集成了这两种方法。在这个项目中，PI将使用大量的驾驶模拟器数据作为测试平台：(1)研究用QN-MHP对并发任务建模的不同方法；(2)将最优网络负载均衡、服务器调度等排队网络理论应用于多任务建模；(3)对网络停留时间、服务器拥塞等排队网络指标与时间、误差、心理工作量等人类性能数据之间的关系进行时间序列建模和可视化；(4)进一步发展QN-MHP，以涵盖更广泛的人类性能，例如为某些服务器提供解决问题的生产能力；(5)进一步发展QN-MHP仿真技术。更广泛的影响：QN-MHP模拟器是在ProModel中实现的，这是一个广泛使用的软件包，需要最少的学习时间，并允许分析师实时可视化QN-MHP内部网络过程，以及最终的统计结果。这些特征不仅对界面分析很有价值，而且对促进认知分析和建模的教学和培训也很有价值。该软件和其他研究成果将由PI随时提供给从业人员、研究人员和教育工作者。随着生命攸关型多模态、多任务HCI接口（如车载和航空设备）的激增，QN-MHP将显著影响系统安全性和产品可用性。

Comprehensive and computational models of human performance have both scientific and practical importance to human-computer system design. This project will make a significan contribution in the area of HCI modeling. The PI will expand a computational model and a simulation technology (implementation) called Queueing Network-Model Human Processor (QN-MHP) he has been developing. QN-MHP is unique, in that it integrates two complementary approaches to cognitive modeling: the procedure and production systems approach (exemplified by the MHP/GOMS family of models, ACT-R, CAPS, EPIC, and SOAR), and the queueing network approach. Procedure and production systems models have achieved great success in modeling, and in generating the detailed procedures and actions that a person might take in performing, a wide range of tasks; their shortcoming is that although they employ mathematics to analyze specific aspects of their models, they lack mathematical theories to represent the overall structure of their models. The queueing network model is able to integrate a large number of influential mathematical models of mental structure as special cases (such as Sternberg's serial stages model, McClelland's cascade model, and Schweikert's critical path network model), and is in general well suited for modeling dynamic and complex tasks and architectural arrangements of processes; as a mathematical theory alone, however, queueing networks cannot be used to generate detailed actions of a person in specific task situations, lacking procedural knowledge a person may employ in accomplishing his/her specific goals. QN-MHP integrates these two approaches by expanding the three discrete serial stages of the MHP into three continuous-transmission subnetworks of a queueing network, by defining each server with procedure functions, and by using a GOMS-style method for task analysis. In this project, the PI will use extensive driving simulator data as a testbed to (1) study different methods of modeling concurrent tasks with QN-MHP; (2) apply queueing network theory such as optimal network load balancing and server scheduling in multitask modeling; (3) perform time-series modeling and visualization of the relationship between queueing network indices such as network sojourn time and server congestion with human performance data such as time, error, and mental workload; (4) further develop QN-MHP to cover a broader range of human performance such as providing certain servers with production capabilities for problem solving; and (5) further develop the simulation technology of QN-MHP.Broader Impacts: The QN-MHP simulator is implemented in ProModel, a widely used software package that requires minimal learning time and allows an analyst to visualize in real time the QN-MHP internal network processes, in addition to the final statistical outcomes. These features are valuable not only for interface analysis, but also for promoting teaching and training in cognitive analysis and modeling. This software and other research results will be made readily available by the PI to practitioners, researchers, and educators. With the proliferation of life-critical multimodal, multitask HCI interfaces such as in-vehicle and aviation devices, QN-MHP will significantly impact system safety and product usability.