Recommendations for Crowdsourcing Services Based on Mobile Scenarios and User Trajectory Awareness

Introduction

With the rapid development of wireless communication technology and mobile smart terminals, based on location service, mobile networks anytime, anywhere can be obtained with its unique features such as mobility, practicability, and portability (Jensen, C. S., et al. 2001). These services and information content are widely used in many fields. The so-called location-based service refers to the cooperation of mobile terminals, and wireless or satellite communication networks are used to determine the actual geographic location of mobile users, location-related information services are provided to users such as map navigation, logistics tracking, traffic monitoring, mobile crowdsourcing task recommendation. However, mobile internet services and information delivery are greatly affected by contextual information and mobile social networks. How to find services that users are interested in from the vast ocean of mobile information and improve user personalized service experience has become an urgent problem for mobile recommendation systems.

Compared with traditional Internet users, the biggest feature of users in mobile communication networks is the random change of user location over time. There is the change of location, it is possible to recommend services based on different locations of mobile users. The core of mobile crowdsourcing service is crowdsourcing task recommendation, which aims to push the spatio-temporal tasks to a set of workers (Tong, Y. X., et al. 2016; Kazemi, L., et al. 2013; Kazemi, L., & Shahabi, C. 2012), and workers complete the same task independently or in a cooperative manner (for example, taking pictures / shooting video or check-in to a specified location), while the constraints of time, location and other tasks are meet (Amador, S., et al. 2014; To, H., et al. 2015). In the mobile crowdsourcing application scenario, crowdsourcing task recommendation faces two important challenges:

The first challenge is the uncertainty of the mobile user's travel trajectory and its intent. Under the crowdsourcing mode, the task is performed by a non-specific group of people on the internet. The acceptance and execution of tasks follow the voluntary principle, which cannot be forced by users based on their own interests or intentions. When task recommendations are made, each potential user's trajectory changes, behavioral intentions, and the impact of various mobile scenarios on them must be fully considered, it tries to match the mobile task with the user's will as much as possible, the success rate of task recommendation and their satisfaction are improved with service recommendation.

The second challenge lies in the dynamic nature of the mobile crowdsourcing service scenario. For example, O2O(Online To Offline) services, such as taxis and takeaways, are constantly emerging from task publishers or potential performers, or they exit at any time, and their positions and trajectories are also dynamically changing. The real-time effectiveness of the space-time task recommendation algorithm and the ability to adapt to dynamic scenarios all raise higher requirements.

The current strategy for recommending tasks to users is mainly based on the current location of the mobile user (task assignment time), a spatio-temporal task is pushed for the user or a task execution route is planed for him, and it is less attention to the user's own trajectory and its location change trend. It may be because the task recommended to him deviates from its trajectory direction and behavioral intention, and he refuses to accept this task, which leads to a low success rate of spatio-temporal task recommendation.

In this paper, because a lot of meaningful information can be provided in the historical trajectory data of workers, a user's movement trajectory pattern, behavior habits, preferences are analyzed for certain places. Based on this analysis, the user's mobile location area is predicted, the tasks are pushied in the area after the prediction, it will increase the probability that the user will accept the task, while the extra travel costs, time and other costs are reduced.

The main contributions in this article are as follows: