Automated Cleansing of POI Databases

Abstract

In the context of geographic information systems (GIS), points of interest (POIs) are descriptions that denote geographical locations which might be of interest for some user purposes. Examples are public transport facilities, historical buildings, hotels and restaurants, recreation areas, hospitals etc. Because information gathering with respect to POIs is usually resource consuming, the user community is often involved in this task. In general, POI data originate from different sources (or users) and are therefore vulnerable to imperfections which might have a negative impact on data quality. Different POIs referring to, or describing the same physical geographical location might exist. Such POIs are said to be coreferent POIs. Coreferent POIs must be avoided as they could harm the data(base) quality and integrity. In this chapter, a novel soft computing technique for the (semi-)automated cleansing of POI databases is proposed. The proposed technique consists of two consecutive main steps: the detection of collections of coreferent POIs and the fusion, for each collection, of all coreferent POIs into a single consistent POI that represents all the POIs in the collection. The technique is based on fuzzy set theory, whereas possibility theory is used to cope with the uncertainties in the data. It can be used as a component of (semi-)automated data quality improvement strategies for databases and other information sources.

Urban human activity density spatiotemporal variations and the relationship with geographical factors: An exploratory Baidu heatmaps‐based analysis of Wuhan, China

“…The first is location‐based HAD data at different time points and different days from obtained Baidu heatmap based on LBS, followed by POI data obtained from Baidu API using crawling technology. The POI data are zero‐dimensional data (Tré et al, ), which does not directly reflect the intensity or the continuous distribution of data covering the entire study area. Therefore, we need to estimate the spatial distribution density of different variables using the location information of the POI data.…”

“…However, it is still necessary to further reveal the characteristic differences of Baidu’s thermal data between the active part and inactive part. Third, since the POI data has zero‐dimensional features (Tré, Britsom, Matthé, & Bronselaer, ), how to achieve spatial matching of POI data with Baidu’s thermal data and the necessary basis for geographic regression is still the problem that needs to be solved. Fourth, due to the limitations of the above‐mentioned problems, there is still a lack of study analyzing the impact of POI‐based variables on Baidu’s thermal data‐based real‐time urban HAD through the geographically regression model.…”

Measurement of Street Network Structure in Strip Cities: A Case Study of Lanzhou, China

“…Using the Google Earth map tool, remote sensing satellite images of Lanzhou City were extracted at a spatial resolution of 10 m. Consequently, spatial information and street networks for all sections were constructed. In addition, the 2019 POI data (containing seven attributes, including name, type, address, longitude, latitude, contact number, and administrative district) of the main urban area of Lanzhou City were obtained from Amap and divided into 12 major categories, including medical, leisure, life, company, transportation, automobile, finance, restaurant, scenic spot, shopping, science, education, and culture, and residential [49]. Each major category consisted of several sub-categories.…”

Fine-tuned Predictive Model for Verifying POI Data

“…POI data classification was done using various techniques including bounding box methods, land use identification and gazetteer information [14][15][16][17]. During the quality assessment for POI data, the researcher faced the limitation in obtaining the reference data [18]. We didn't find many works done in POI verification or accuracy measurement of the crowdsourced geotagged POI data.…”