Centre de documentation
scientifique

Knowledge extraction from crowdsourced data for the enrichment of road networks / Gregor Jossé in Geoinformatica, vol 21 n° 4 (October - December 2017)  

Public [article]  inGeoinformatica > vol 21 n° 4 (October - December 2017) . - pp 763 - 795 Titre : Knowledge extraction from crowdsourced data for the enrichment of road networks Type de document : Article/Communication Auteurs : Gregor Jossé, Auteur ; Klaus Arthur Schmid, Auteur ; Andreas Züfle, Auteur ; et al., Auteur Année de publication : 2017 Article en page(s) : pp 763 - 795 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Bases de données localisées [Termes IGN] contenu généré par les utilisateurs [Termes IGN] densification [Termes IGN] données hétérogènes [Termes IGN] données localisées des bénévoles [Termes IGN] extraction de données [Termes IGN] géopositionnement [Termes IGN] navigation [Termes IGN] production participative [Termes IGN] réseau routier [Termes IGN] utilisateur Résumé : (Auteur) In current navigation systems quantitative metrics such as distance, time and energy are used to determine optimal paths. Yet, a “best path”, as judged by users, might take qualitative features into account, for instance the scenery or the touristic attractiveness of a path. Machines are unable to quantify such “soft” properties. Crowdsourced data provides us with a means to record user choices and opinions. In this work, we survey heterogeneous sources of spatial, spatio-temporal and textual crowdsourced data as a proxy for qualitative information of users in movement. We (i) explore the process of extracting qualitative information from uncertain crowdsourced data sets employing different techniques, (ii) investigate the enrichment of road networks with the extracted information by adjusting its properties and by building a meta-network, and (iii) show how to use the enriched networks for routing purposes. An extensive experimental evaluation of our proposed methods on real-world data sets shows that qualitative properties as captured by crowdsourced data can indeed be used to improve the quality of routing suggestions while not sacrificing their quantitative aspects. Numéro de notice : A2017-603 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE Nature : Article DOI : 10.1007/s10707-017-0306-1 En ligne : https://doi.org/10.1007/s10707-017-0306-1 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=86911 [article]

Uncertain Voronoi cell computation based on space decomposition / Klaus Arthur Schmid in Geoinformatica, vol 21 n° 4 (October - December 2017)  

Public [article]  inGeoinformatica > vol 21 n° 4 (October - December 2017) . - pp 763 -795 Titre : Uncertain Voronoi cell computation based on space decomposition Type de document : Article/Communication Auteurs : Klaus Arthur Schmid, Auteur ; Andreas Züfle, Auteur ; Tobias Emrich, Auteur ; et al., Auteur Année de publication : 2017 Article en page(s) : pp 763 -795 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Analyse spatiale [Termes IGN] décomposition spatiale [Termes IGN] diagramme de Voronoï [Termes IGN] incertitude des données [Termes IGN] plus proche voisin, algorithme du [Termes IGN] traitement parallèle Résumé : (Auteur) To facilitate (k)-Nearest Neighbor queries, the concept of Voronoi decomposition is widely used. In this work, we propose solutions to extend the concept of Voronoi-cells to uncertain data. Due to data uncertainty, the location, the shape and the extent of a Voronoi cell are random variables. To facilitate reliable query processing despite the presence of uncertainty, we employ the concept of possible-Voronoi cells and introduce the novel concept of guaranteed-Voronoi cells: The possible-Voronoi cell of an object U consists of all points in space that have a non-zero probability of having U as their nearest-neighbor; and the guaranteed-Voronoi cell, which consists of all points in space which must have U as their nearest-neighbor. Since exact computation of both types of Voronoi cells is computationally hard, we propose approximate solutions. Therefore, we employ hierarchical access methods for both data and object space. Our proposed algorithm descends both index structures simultaneously, constantly trying to prune branches in both trees by employing the concept of spatial domination. To support (k)-Nearest Neighbor queries having k > 1, this work further pioneers solutions towards the computation of higher-order possible and higher-order guaranteed Voronoi cells, which consist of all points in space which may (respectively must) have U as one of their k-nearest neighbors. For this purpose, we develop three algorithms to explore our index structures and show that the approach that descends both index structures in parallel yields the fastest query processing times. Our experiments show that we are able to approximate uncertain Voronoi cells of any order much more effectively than the state-of-the-art while improving run-time performance. Since our approach is the first to compute guaranteed-Voronoi cells and higher order (possible and guaranteed) Voronoi cells, we extend the existing state-of-the-art solutions to these concepts, in order to allow a fair experimental evaluation. Numéro de notice : A2017-604 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE Nature : Article DOI : 10.1007/s10707-017-0293-2 En ligne : https://doi.org/10.1007/s10707-017-0293-2 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=86913 [article]