Research On Semantic Trajectory Extraction Method For Indoor Space

With the rise of mobile internet,location-based research has attracted more and more attention.By using massive location data,we can analyze user's behavior,explain people's daily routine and explore the potential value of customers.With the development of GPS,we already have a stable and mature solution for outdoor positioning.However,due to the complicated indoor environment,there is no unified and stable technical means at present,and we can only acquire limited knowledge of indoor behavior.What's more,data show that people have 80 to 90%of the time spent in the indoor environment.As a result,many companies take part in indoor positioning market over the years,and that has led to the sporadic emergence of indoor positioning and related data analysis.In order to solve the problem of indoor location data,this paper studies the theory and method of extracting semantic trajectory,which can reduce the error,compress the raw trajectory,enhance the ability of expression and make further trajectory data mining possible.Specifically,the main research contents and innovations of this paper include:(1)Using the existing IFC geometric model to construct indoor space model,which expands the input source of the model.At the same time,a hybrid indexing method,which combines spatial index and offline distance matrix,is proposed to optimize the calculation of minimum indoor walking distance.(2)Based on the spatial-temporal characteristics of trajectories,an event extraction algorithm based on DBSCAN algorithm is proposed.And An interval scheduling algorithm is introduced to solve the problem of event collisions in clustering.(3)We Develop a suite of indoor trajectory data analysis tools,which can adjust the semantic trajectory extraction algorithm in a visual way.Finally,by performing lots of experiments on the real dataset and synthetic dataset,we verify the validity of the algorithm.