Mining Technology Of Personalizing Service Requirement For The Traveler

With the development of the Internet technology, the way people acquire information become more multiple, and the amount people require for information increased fast, while traditional traffic information service provide not only low service level, but also single service pattern and indefinite about public travel requirement. Recently, research and applications about traffic big data and data mining technology offer the accomplishment of personal traffic information services, which can highly improve the quality of existed traffic information service and the travel experience of traveler, thus made the travel service being more personal. Therefore, establish personal traffic information service framework become one of the hot spot.Based on the above reason, this paper aimed at composing personal traffic information service and began the research from two sides, which targets at solving problems include travel requirements fuzzy and insufficient of hidden travel requirements in traffic information service. To begin with, this paper proposes a personal traffic information service requirement mining framework based on analysis the requirement of personal traffic information service and requirement mining technology, which consists of the construct of travel behavior model and mining of service requirement; Secondly, based on the research of traveler's behavior feature, this paper puts forward a stay point identify algorithm using stay point to describe travel behavior model, at the meantime,this algorithm also combined the feature of GPS information with travel behavior, besides, to eliminate the inaccuracy of model describe caused by wrong GPS information, this paper come up with a limited stay point cluster algorithm based on the idea of density cluster; then, considering the singleness and sparse of data in traveler behavior model, it is hard to reflect the problem in traveler's activity routine, thus, replaced single location trajectory using POI data set and service type trajectory, mining deep level travel information by frequent mode; last but not least, this paper analyses the limitation of existed service requirement recommend algorithm and find out that the travel similarity in travel model is associated with both location visit sequence and visit times, based on this, proposes a service requirement mining algorithm using trust index model.Result of experiment showed that, both the stay point identify algorithm based on requirement constrain and the limited stay point cluster algorithm can accomplish the construct of travel behavior mode, with a 95% stay point identify accuracy and a 90% error processed accuracy, besides, the intelligent language match algorithm based on condition limit appears a wider cover rate than traditional match algorithm; while the service requirement mining algorithm based on trust index model turns out a more accurate result than traditional method using cosine similarity. In one word, the algorithms proposed in this paper can accomplish the construct of the travel model and the mining of the service requirement.