Cognitive Aware Smart Service Composition

Internet正在成为承载异构服务的全球基础设施。一般来说，有两种类型的服务运行在互联网上：（1）网络服务（例如，在线商店）由服务器托管并且通过与其它web服务通信来向用户提供软件功能;以及（2）物联网（IoT）服务（例如，恒温器应用程序），其直接与硬件设备（例如，恒温器）并对环境产生直接影响（例如，改变室温）。尽管越来越多地使用互联网上的服务来支持人类活动，但大多数在线活动都是手动执行的。有有限的智能支持，以促进人类和机器之间的互动。 在这个项目中，我们与IBM加拿大的工业合作伙伴合作，建立一个智能服务组合框架，可以感知最终用户的情景需求，自动支持最终用户进行在线任务，并减少人类的认知负担。我们的目标是利用机器学习算法来学习最终用户的在线活动和偏好，以改善我们的服务推荐。该项目将集中在三个方面：1）构建一个智能个人代理框架，可以无缝集成传统的Web服务和物联网服务，并使代理能够代表最终用户执行在线任务; 2）创建分析和预测引擎，跟踪各种信息源，分析信息，并向最终用户推荐服务;以及3）开发多代理协作基础设施，该基础设施允许各种角色的个人代理代表其终端进行通信和交互，用户实现共同的目标。 我们提出的项目的结果可以让最终用户创建智能环境，以响应他们的需求，并帮助提高他们在电子健康，电子商务，金融服务和辅助生活等各个领域的生活质量。该项目将在软件工程的重要领域培训10名HQP（2名博士，4名硕士和4名学士），特别是SOA和物联网软件。

The Internet is becoming a global infrastructure to host heterogeneous services. Generally speaking, there are two types of services running on the Internet: (1) web services (e.g., on-line stores) are hosted by servers and provide users with software functionality by communicating with other web services; and (2) Internet of Things (IoT) services (e.g., thermostat apps) that directly communicate with hardware devices (e.g., thermostat) and make a direct impact to the environment (e.g., changing the room temperature). Despite the increasing use of the services on the Internet to support human activities, most on-line activities are performed manually. There is limited intelligent support to facilitate the interactions between human and machines. In this project, we collaborate with our industrial collaborators in IBM Canada to build a smart service composition framework that can perceive end-users' situational needs, automatically support end-users to conduct on-line tasks and reduce human cognitive burden. We aim to utilize machine learning algorithms to learn end-users' on-line activities and preferences over time to improve our service recommendation. The proposed project will focus on three aspects:1) Building a smart personal agent framework that can seamlessly integrate the traditional web services and IoT services, and enable an agent to act on behalf of an end-user in performing on-line tasks; 2) Creating analytics and predictive engines that track various information sources, analyze the information, and recommend services to end-users; and 3) Developing a multi-agents collaboration infrastructure that allows personal agents of various roles to communicate and interact on behalf of their end-users to achieve a common goal. The results of our proposed project can allow end-users to create smart environments to be responsive to their needs and help improve the quality of their lives in various domains, such as e-health, e-commerce, financial services, and assisted living. This project will train 10 HQP (2 PhD, 4MSc, and 4 BSc) in the important area of software engineering, specifically SOA and IoT software.