The Research On Context-aware Computing And Application For Mobile Users

Context-aware computing technology attracts extensive attention with the development and popularization of mobile devices especially smart phones, as well as the rise of the mobile internet and the Internet of Things. The combination of context-aware computing and mobile applications produces a new technology named mobile context-aware computing. The main task of mobile context-aware computing is making mobile smart devices obtain context information proactively, recognize and perceive context, and provide appropriate services to users according to the context information. Services based on mobile context-aware computing offer great convenience to users daily life and better satisfy the users’personalized needs, therefore they are increasingly favored by people.A hot research direction of mobile context-aware computing is activity recognition which is the basis of mobile context-aware services such as online social network awareness, individual health monitoring and personalized mobile search. However, the computation capability and storage capability of mobile devices are limited, so it has become a critical issue that how to establish an efficient and accurate classifier for mobile users’activities. To this end, this paper proposed an activity recognition algorithm based on Extreme Learning Machine (ELM). The algorithm extracts features from acceleration data collected by mobile phones, selects features according to multi-attribute including the mutual information, the sensitivity for classification and the computing cost of features, generates the feature subset which is suitable for the activity recognition on mobile phones, and then uses ELM to train a neural network with these features data in order to obtain a classifier for mobile uses’activities. The algorithm aims at improving the recognition accuracy and reducing computational overhead.Mobile search is one of the most popular mobile applications, which is favored by mobile users because of its advantages such as celerity, convenient and no limit on time and place. Mobile search considering user contexts is a promising application of mobile context-aware computing. It is able to improve the personalization of searches, simplify the interaction between users and mobile devices, as well as remedy disadvantages of mobile devices with limited input and display. The comprehensive awareness of user contexts and the reasonable inference of user intents are research emphases on the fusion of mobile search with mobile context-aware computing. Therefore, this paper proposed a query expansion algorithm based on user contexts. This algorithm uses mobile phones to acquire direct user contexts and combines with the activity recognition algorithm based on ELM to infer indirect user contexts. In this paper, the acquired direct user contexts include acceleration, location, date and temperature, etc. And the inferred indirect user contexts include users’activities and transportation modes, etc. The expert system deduces extended words to represent user intents from these two types of context information. These extended words are used in query expansion in order to provide the personalized mobile search to users and improve users search experience.The proposed algorithms are validated by experiments based on real-world datasets. Results demonstrated that the proposed activity recognition algorithm based on ELM can train the classification model with high accuracy. Acceleration features used in the classification is suitable for mobile phones because of their low computational complexities. The activity recognition algorithm can collect user context data and recognize activities in real time. Users’activity information recognized by the algorithm and other user context information sampled by mobile phones can be well integrated into query expansion in mobile search. The query expansion algorithm based on user contexts can enhance precision and meet users’personalized needs.