Recently, rapidly advanced IoT technologies have enabled intelligent services that provide users with the services or information needed in situations, and the demand for them is increasing. The intelligence service provided by the server has sufficient storage space and processing performance to obtain stable analysis results. However, the lack of real-time context information often limits the provision of information according to the situation. In addition, user-centered analysis is mainly used to bias the contents of the situations or object information frequently used by users. Also, there is a lack of information about objects or situations that are not frequently used. In this study, "object" refer to things in the IoT environment that can collect information, such as sensory information, analyze information, and exchange information through communication, including people.
This study aims to solve the following problems. First, it solves the problems that occur in intelligent services that do not consider the existing context. The existing service lacks information about the context and surroundings, so there is a problem in that the provision of information according to the context is limited. This study solves the problem by proposing a method for collecting and analyzing information considering the dynamic situation of objects. By collecting and analyzing information centered on objects, it is possible to provide a wealth of information related to the context of objects as a service. Second, it solves the problem of increasing data space that occurs when collecting information from existing objects. Methods of reducing the amount of information collected and improving the quality by limiting the scope of information collected have been proven effective in the past. However, in order to limit the scope of information to be collected, it is necessary to understand the characteristics of the information well. In this study, explicit characteristic information of objects is used to limit the scope of information collection. In order to obtain explicit characteristic information of an object, an explicit interest analysis is performed using static information of an object. Finally, we study interest analysis methods in objects in order to enrich the information on objects and provide appropriate information to users. Most of the existing interest analysis methods were conducted using social network data and user behavior patterns (clicks, purchase history visits, etc) when using the site. However, in this paper, we propose a method that uses both context information, social network data, and web crawling data for analysis. In addition, the existing interest analysis require a lot of analysis time to be applied to an object with limited resources. However, in general, existing analysis methods has a problem in that analysis accuracy is lowered if sufficient analysis time is not provided. Therefore, in this study, an interest analysis technique that maintains accuracy while having an appropriate response speed to provide a service from an object is studied. Moreover, we design a framework to provide intelligent service to which the proposed analysis technique is applied.