Abstract
Points of interest (POIs) such as stores, gas stations, and parking lots are particularly important for maps. Using gas station as a case study, this paper proposed a novel approach to enhance POI information using low-frequency vehicle trajectory data and social media data. First, the proposed method extracted spatial information of the gas station from sparse vehicle trace data in two steps. The first step proposed the velocity sequence linear clustering algorithm to extract refueling stop tracks from the individual trace line after modeling the vehicle refueling stop behavior using movement features. The second step used the Delaunay triangulation to extract the spatial information of gas stations from the collective refueling stop tracks. Second, attribute information and dimension sentiment semantic information of the gas station were extracted from social media data using the text mining method and tripartite graph model. Third, the gas station information was enhanced by fusing the extracted spatial data and semantic data using a matching method. Experiments were conducted using the 15-day vehicle trajectories of 12,000 taxis and social media data from the Dazhongdianping in Beijing, China, and the results showed that the proposed method could extract the spatial information, attribute information, and review information of gas stations simultaneously. Compared with ground truth data, the automatically enhanced gas station was proved to be of higher quality in terms of the correctness, completeness, and real-time.
Highlights
Points of interest are an indispensable component in basic geographic information and play an important role in a variety of fields, including Location Based Service (LBS), scientific research, and commercial applications [1,2,3]
The geotagged User-Generated Content (UGC) data have opened a new way of acquiring Points of interest (POIs) semantic information but cannot be obtained from global positioning system (GPS) traces [12,13]
Social media data are sparse in space and time compared with GPS trace [16], which makes it difficult to sense the dynamic change of spatial information and activity behavior in the POI place [17]
Summary
Points of interest are an indispensable component in basic geographic information and play an important role in a variety of fields, including Location Based Service (LBS), scientific research, and commercial applications [1,2,3]. 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]. One is through collaborative mapping programs [5,6], such as OpenStreetMap, which are constructed manually from crowdsourcing 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]. Compared with collaborative mapping programs, this method is real-time, low-cost, updates quickly, and can monitor the temporal and spatial change of activity patterns in the POI place. Social media data are sparse in space and time compared with GPS trace [16], which makes it difficult to sense the dynamic change of spatial information (e.g., the area of POI, subsidy facility of POI, infrastructure, etc.) and activity behavior in the POI place [17]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
