Visual Query Methods And Applications On Human Behavior Data

Human behavior data transforms the way we live and work. Studying human behavior data is of great value. Yet, mining the data is quite difficult because it has big volume, multiple types as well as rich sources. Visual query is a comprehensive and essential tool for exploratory data analysis, which has the inherent advantages to make sense of big human behavior data. Visual querying on human behavior data is storage-and-computing-intensive, yielding various challenges: the lack of mature visual query models, the lack of effective data organization methods, the difficulty of data query; the lack of flexible interfaces, the lack of innovative and intuitive visual designs.This thesis summarizes related work and introduces a visual query cognition model. By this model, visualization and data query can be integrated to allow for explorative analyzing human behavior data. This thesis describes how visual query can be leveraged to analyze human behavior data for multiple objectives, such as social behaviors analysis in social space, driving behaviors analysis on road segments, and travel behaviors analysis in the urban environment. A large number of experiments, cases and experts’feedbacks validate the usefulness of visual queries on studies human behavior data. Specifically, main works and contributions of this thesis are as follows:We propose a visual query method for semantic-aware social behaviors analysis. We extract hidden topics and named entities based on semantic correlations of social behaviors, we categorize topics and named entities to enrich semantics of social data, a hierarchical semantic graph model is presented to simplify the graph model of raw data, as well as multiple indices. Furthermore, A multi-category-based navigation interface is designed to facilitate discovering social behaviors. Split rings are employed to encode multiple semantics of social data. User studies and case studies from Twitter data show that split rings highlight salient features of social behaviors.-We propose a micro-pattern scheme for driving behaviors analysis. We present a intuitive visual query approach to analyze real driving behaviors and reasoning transport situations based on taxi trajectory data. A sketch-based interface is designed to support dynamic querying and reasoning within coordinated multiple views. In particular, we propose a novel road-based query model for experts to interactively conduct analysis tasks. This model is supported by a bi-directional hash structure, TripHash, which enables real-time responses to the data queries over a large amount of trajectory data. Case studies with a real taxi GPS trajectory dataset (>30GB) show that our system performs well for revealing transport situations in micro-patterns.-We present a visual query platform for analyzing population travel patterns from multi-source behavior data. The platform is based on an general solution to fill the gaps between various data. Four subdivision plans are discussed to store the huge quantity of data in a distributed database. Moreover, an adaptive interface is designed to support visual querying on multi-source behavior data. Two interactive tools are designed to input spatio-temporal constraints, and two visualizations are applied to show population static and dynamic distributions. An optimized MapReduce procedure are proposed to enable real-time responses to the data queries. Case studies reveal population travel behaviors in the urban environment (e.g. citizens gathering and scattering patterns) by integrating taxi trajectory data and cell phone call records.