Research And Application Of Personal Trajectory Pattern Mining Algorithm Based On Lifelog Data

In recent years,with the progress of modern information technology,A variety of location-based technology services also becomes developed and popular.More and more mobile terminal products carry location-aware device already coming into view,and has been widely used in people's daily life.By using of these products,people can easily get the position information of moving objects.It has become the hot issue in academia that how to gain more valuable information by analyzing a large number of mobile trajectory data from moving objects.However,most available studies are focusing on the vehicle or large people's trajectory mining,and few researches about individuals.Because of personal location data with a great degree of uncertainty,it seriously affected the existing trajectory pattern algorithm.However,mining individual's trajectory pattern can help providing more personalized service,and it is important for personal daily life management or some prediction work.To solve these problems above,we proposed a trajectory mining algorithm based on personal Lifelog data.Firstly,it's about data collection and filtering.For the data collected by the APP Moves from the phone,mostly to extract GPS data.for the photos collected by Autographer camera,mostly to get the main concept features.Secondly,cluster the original GPS data,and make the GPS points which are closed each other into one cluster.Finally,using the frequent pattern mining algorithm to mine the trajectory pattern.And using local sensitive hashing algorithm to add semantic description to the places in a trajectory pattern,by extracting features from photos to train classifiers,and further adding semantics to the whole trajectory pattern.In the clustering stage,by trying different clustering algorithms to achieve the desired effect.In the process of frequent pattern mining,we use two different algorithms for mining personal trajectory pattern under different needs of individuals.We do the experiment and analysis based on real data provided by users to verify the algorithms proposed in this thesis,and finally we complete the design and implementation of individual trajectory pattern mining prototype system.The results show that both the personal trajectory pattern mining algorithm and the semantic information adding algorithm is effective,It can mine personal trajectory model quickly and effectively,and the semantic description also makes the trajectory pattern more intuitive.The results also can be used to predict the future routes of users and helped to manage personal life.In addition,the algorithm has a good scalability and can be used in different scenarios.Therefore,the research work is very significant,and has a certain role in the development and application of personal pattern mining.