Modelling of Trajectory Moving Data with use of Social Media: A Review

Abstract

From last decades, there has been more research on mobile phone network data. Usually, mobile phone user mobility can be track from GIS of device or online social media information with up gradation of location enabled services. Tracking using GIS system is much costlier, time consuming and requires more man power while social media used by smart phones and location enabled devices can easily track the location and time, which can be further utilized by researcher in spatiotemporal domain along with other application related attribute or information. User mobility pattern and prediction are recent interesting area in data and pattern mining. There are many applications for trajectory data modelling such as transportation, tourist place recommendation, land use application, spread of diseases tracking etc. As information in social media data is not dense and difficult to analysis, this article proposes group methodology of user textual data with respect to location and time of user post. In this, user clustering and moving pattern design are made alternatively with each other so that efficiency of method will increases in each stage of method. User clustering is based on Hidden Markov Models (HMMs) which uses definite number of unrecognizable latent states where each record keep up a correspondence to each latent state which has a probabilistic distribution that governs the generation of records as well as text augmentation for proper distribution of textual data with respect to location and time, which is also used to reduce text complexity and not dense data [2]. This method is used to increase prediction accuracy of trajectory moving data.