Extraction, integration and analysis of crowdsourced points of interest from multiple web sources

Abstract

The amount of user-generated geospatial content on the Web is constantly increasing, making it a valuable source of information for enabling, enriching and enhancing geospatial applications and services. However, this content is highly heterogeneous and diverse, varying significantly in quality and accuracy. Extracting, integrating, and mining these crowdsourced geospatial data from the Web is far from trivial. Among the main challenges are to retrieve data from multiple sources, each one providing its own access methods and restrictions, to deal with different schemas and taxonomies, and to find matching entities across multiple sources. In this paper, we present our work for retrieving and integrating crowdsourced Points of Interest (POIs) from popular Web sources. We retrieve POIs from different Web sources and we describe the steps taken for mapping the source categories to a common schema, detecting duplicate POIs, and eventually clustering them to identify hotspots. We present the results of this process applied to six major Web sources to retrieve and integrate POIs located in the metropolitan areas of three European capital cities.

POI Information Enhancement Using Crowdsourcing Vehicle Trace Data and Social Media Data: A Case Study of Gas Station

“…However, it is difficult to achieve this goal only using a single data (trajectory data or social media data). For example, the POI data extracted from social media data [16] (e.g., Yellow Pages, Dianping) have drawbacks such as low accuracy, redundancy, duplicates, and outdated [1,3,6], while GPS trace data lacks attributesemantic information. Therefore, multisourced VGI data should be fused to enhance POI information.…”

“…For the one to many, the center point of the polygon that was constructed by the points as the new POI point, and one of the most complete attribute data was written into the database. The zero to one case was divided into two situations, one was that the D-POI has been dismantled or changed to other types of POI, but the web data have not been updated [6]. Another was that the T-POI has not been extracted, but the D-POI exists in the real world.…”

“…Keeping point of interest (POI) data up-to-date, correct, and complete in a short updating cycle and low cost way is extremely important [1][2][3]. However, POI data such as gas stations and parking lots are mainly traditionally obtained from field surveys, remote sensing, and manual annotation [4][5][6]. These methods are costly, have a long update cycle, and are time-consuming, leading to limiting POI data services [5,6].…”

“…However, POI data such as gas stations and parking lots are mainly traditionally obtained from field surveys, remote sensing, and manual annotation [4][5][6]. These methods are costly, have a long update cycle, and are time-consuming, leading to limiting POI data services [5,6]. Consequently, using a low cost and fast method to enhance and update POI data automatically is very significant.…”

“…One is through collaborative mapping programs [5,6], such as OpenStreetMap, which are constructed manually from crowdsourcing. However, these sources are limited (e.g., focused on a single domain, limited spatial coverage) and are updated slowly due to insufficient user participation [5,6]. Another option is to extract POI data from global positioning system (GPS) trajectory data [7][8][9][10][11].…”