Evaluating residential location inference of twitter users at district level: focused on Seoul city

Investigación académica: Recolección de datos, tecnologización y pandemia

“…3 Kim et al, (2016), expusieron que las personas viven actualmente en la era que escriben a través de sus teléfonos móviles información de forma voluntaria a través de servicios de notificación simple (SNS), es decir, datos masivos en tiempo real, que son fácilmente compartidos y analizados. Agregan que, a diferencia de otros SNS, Twitter proporciona aplicación abierta de forma gratuita, lo que permite a investigadores, empresas y oficinas gubernamentales ubicar, recuperar información y analizar datos para diversos fines.…”

Fast mining of spatial frequent wordset from social database

“…For such purpose, words are sorted in order from a word with larger frequent count to a word with smaller frequent count so that words that appear frequently in the entire area become patient nodes and words that appear less frequently in the entire area become child nodes, and then words are converted to wids (3). The spatial coordinates of ssd are mapped in spatial partition unit on the hierarchical spatial grid grid and gid is allocated (5).A new wordset data is inserted by entering wids and gid of ssd on sfpTree (6). If the traversal for all data in the social database is finished, the construction of sfpTree is completed.…”

“…Next, header table FTcondTree of spatial word condition tree condTree and condTree is extracted from all spatial words (w, gid) in FT Li using spatialWordCondTree. In order to minimize the size of condTree created in each traversal, inverse traversal is carried out in order from a word with smaller count to a word with larger count (5)(6). The spatial word condition tree is non-spatial prefix tree that only extracts the tree part which satisfies a specific word w and a specific space gid from sfpTree which includes spatial information.…”

“…Since such social data is dynamic in comparison with web documents and most data includes spatial information including GPS locations, it has a characteristic of transaction data. However, such social data mostly consists of text, so only a simple attempt such as change in the frequent count has been made in the studies applying the previous data exploration algorithms [3,4,5,6,7].…”