Research On Crowdsourcing-based Group Awareness Technology And System Implementation

In recent years, with the spreading of World Wide Web and smart mobile devices, location based services in people's lives have become increasingly popular. For some public and fixed objects location information, we could obtain the good and comprehensive query results by using the current mainstream search engines. However, for the objects that are in construction or newly constructed, traditional search engines may not be able to update their information in the first time and get their geographical locations. Meanwhile, for the movable object, the users may want to know their location information and the moving locus at a particular time, however this is a great challenge to a conventional search engine or a map.To solve the above problems, we use the concept of crowdsourcing, and encourage people to participate in solving problems by using the online socialnetwork. It can locate the untagged large physical facilities position and portray their moving locus. In our paper, firstly, we introduce the basic knowledge and research method of crowdsourcing, and design the system model combined the function of sensing large facility location. Then, we study and discuss the key technologies in our model.The users usual are more familiar with the surroundings.In the part of task management model, makes the target objects location and users residence into latitude and longitude. So we propose the mission recommendation mechanism based on distance. It lists tasks in accordance with the distance ascending and published time descending, afterward, the tasks will be personalized recommended to users.Because of the unpredictability of users behavior in the crowdsourcing system, it is critical to analyze and verify the annotation results. We come up with two verification strategies. The first one is mean method which uses the average value as the final result. Tthe second strategy, the initial result sets are firstly screened by K-means clustering algorithm so that the obviously abnormal data will be removed. Then the optimal point will be selected from the filtered result sets by using the voting mechanism. The second method uses the principles of human-machine collaboration.Finally, we construct a platform named SocSense based on group awareness. Through do experiments on the SocSense platform, we get the highly accuracy results and confirm the practicality and efficiency of the system.