Personalized App Service Composition based on Usage Behavior of Smartphone Owners in IoT Environment

Abstract

Smartphone owners use many smartphone applications to request the services needed in their daily
lives such as shopping, payment or entertainment. Despite the behavioral change of the human,
there are very few research initiatives to analyze the behavioral patterns of the apps usage in order
to improve the utility of apps. As a result, smartphone owners should make decisions to select the
appropriate apps in their context, and even perform to compose the services among different apps
when they need to interact with each other. Unfortunately, the average number of apps installed on
smartphone is about 86 in Korea, Japan, and America. The cumbersome of apps usage may be severe
in the future. To overcome this, we propose the App Service Composition (ASC) borrowing the concept
of Web service. In ASC, we find the app usage patterns with the user context using the machine
learning techniques and automatically discover and select the heterogeneous app services using the
Linked Open Data (LOD) technology. To prove the superiority of the framework, we perform the
experiments to evaluate performance of the framework that can predict the apps usage patterns of the
users.