Analysis Of Spatio-temporal Characteristics Of User Behavior Based On Social Data

The deep understanding of the temporal and spatial patterns of user behavior has enormous application value in many fields.Typical application scenarios include infor-mation dissemination,personalized recommendations,and business intelligence.As a discipline that studies the temporal and spatial characteristics of human behavior,hu-man dynamics has formed a mature analytical framework.Using these classical theories,on the one hand,the user behaviors can be effectively analyzed based on typical statis-tics and models;on the other hand,the analysis results can verify the completeness of these definitions.Exploring the places where disagreements occur will often produce new academics.See the ground.However,user data naturally possesses complex and multi-dimensional essential attributes,which makes it difficult to gain insight into user behavior using analysis based on prior knowledge.The human visual abilities and their own intuitive intuitions are particularly suitable for dealing with such issues.Based on this,this article has mainly done the following research work:(1)Statistical analysis of temporal and spatial characteristics of human dynamics.Using the human dynamics theory,the temporal and spatial characteristics of the user's check-in behavior are verified and analyzed.In terms of time characteristics,the time interval distribution of signing behaviors of group users is approximately power-law dis-tribution with power exponent of 2.64;the power exponent of group level increases with activity,and the power exponent of individuals at the individual level is positively corre-lated with activity;user sign-on behavior The bursts of memory and memory are more divergent.In addition,the signing behavior of users also shows obvious periodicity and volatility.In terms of spatial characteristics,the group jump distance distribution obeys a power law step with a power index of approximately 1.394;the user's radius of gy-ration converges with time and finally tends to be stable;the number of users visiting a unique location increases with time;the power law increases;The re-scaling results show that the population jump distance distribution of different radii groups has collapsed to a curve;the probability distribution of the user's return to historical trajectory points at the time of t has a periodicity and eventually becomes stable;these phenomena indicate the user The mode of travel behind sign-in behavior has the characteristics of boundedness,periodicity and regularity.(2)Visualization algorithm implementation.First clear the role of the visualization algorithm,based on algorithm requirements to develop algorithm design work.For spa-tial attributes,the Mercator projection algorithm is used to map geographic coordinates of latitude and longitude,and the registration behavior of the circle is coded.For the time attribute,the stacking graph algorithm is used to represent the time sequence changes of the sign-on behavior,and the comparison between multiple check-in types is performed;The calendar thermogram algorithm explores the characteristics of finer temporal gran-ularity;it uses an improved front-strand circle filling algorithm to present a location-user-global-check-in-single-entry hierarchy;for the exploration of location functions,a metaphor based on watches is designed.The location function exploration algorithm uses the dynamics of the dial to change the function of the location.(3)Visualization system implementation.First of all,abstract the design task of the system,proceed from the abstract module diagram of the system,describe in detail the functions of multiple views and interaction modules implemented by the visual algorithm mapping.Based on the system module architecture,follow the data pipeline process,and use the interaction module to connect Multiple views and rich interactive features were implemented to build a visual analysis system.Finally,starting from the overview of the system user interface,the effectiveness of the system was confirmed through five typical case studies.