Making a map from thematically multi- sourced data: the potential of - PowerPoint PPT Presentation

23 Septembre 2014 – 17th ICA Generalisation workshop Making a map from “thematically multi- sourced data”: the potential of making inter-layers spatial relations explicit Cécile Duchêne IGN, COGIT Laboratory cecile.duchene@ign.fr

« Thematically Multi-Sourced » maps One kind of map commonly produced: � Sources 2, …, n Thematic data Source 1 Base data => backdrop map 23.09.14 2

« Thematically Multi-Sourced » maps One kind of map commonly produced: � Sources 2, …, n Thematic data Source 1 Base data => backdrop map Rest of the presentation: � 1 base dataset + 1 thematic dataset (each from 1 source) � Each internally consistent (no redundancies, etc.) � 23.09.14 3

« Thematically Multi-Sourced » maps One kind of map commonly produced: � Sources 2, …, n Thematic data Source 1 Base data => backdrop map Rest of the presentation: � 1 base dataset + 1 thematic dataset (each from 1 source) � Each internally consistent (no redundancies, etc.) � 23.09.14 4

« Thematically Multi-Sourced » maps These maps : � have existed for a long time, produced by expert cartographers � (Das et al. 2012) � have exploded over the web (web mapping, by non cartographers) � often have a bad legibility => potentially bad decision making (Jaara et al. 2011; Das et al 2012; Gaffuri 2011; Balley et al. 2014; Sester et al. 2014) � …often due to a bad management of relationships btw thematic and background layer 23.09.14 5

« Thematically Multi-Sourced » maps 23.09.14 6

« Thematically Multi-Sourced » maps 23.09.14 7

« Thematically Multi-Sourced » maps 23.09.14 8

« Thematically Multi-Sourced » maps 23.09.14 9

MOTIVATION Conviction: thematically multi-sourced maps could be improved � by a better, explicit management of relations � Objective: analyse where we are now, and what could be done � Additional motivation: “Despite all advances in digital generalisation, no overall generalisation theory has been worked out, nor are there convincing solutions for digital generalisation of all relationships between cartographic objects” Ormeling (2011) 23.09.14 10

OUTLINE Introduction � � Relations in thematic maps: role and evolution through scale � Managing thematic-background relations Open research issues � 23.09.14 Intro Rels in TMS maps Managing rels Open issues 11

ROLES OF RELATIONS IN THEMATIC MAPS Relations at least as important as features : well known � (Papadias and Theodoridis 1997; Ruas and Mackaness 1997; Mackaness and Edwardes 2002; Touya et al. 2012; Mackaness et al. 2014) � Topographic maps: generalist. What relations are interpreted? � Thematic maps: � The theme of interest for the user is known � Backdrop data = spatial context for thematic data (Sester et al. 2014) � Relations with the backdrop map are interpreted: => spatial + semantic 23.09.14 Intro Rels in TMS maps Managing rels Open issues 12

RELATIONS EVOLUTION THROUGH SCALE? Some relations are (should be) invariant (Bobzien and Morgenstern 2003) : � adjacency/inclusion, reachability, relative position � Generalisation should caricature « almost present » relations (Duchêne et al., 2012) � Thematic-background relations should be abstracted when the LOD decreases (Jaara et al., 2013) 23.09.14 Intro Rels in TMS maps Managing rels Open issues 13

OUTLINE Introduction � � Relations in thematic maps: role and evolution through scale � Managing thematic-background relations Open research issues � 23.09.14 Intro Rels in TMS maps Managing rels Open issues 14

Managing relations… when? At two main stages: � Integration of thematic data and (one) background dataset LOD n … LOD 2 LOD 1 � Derive map at intended LOD Generalisation Replacement of background + « migration » (Jaara et al. 2011) 23.09.14 Intro Rels in TMS maps Managing rels Open issues 15

Managing… what kinds of relations? Existing VS expected relations � This route is locally very close to the road This route should be locally equal to the road � Relations holding at instances VS type level This accident is close to that junction Accidents are on roads � « Hosting » (Jaara et al. 2012) VS « Peer to peer » relations The road « hosts » the accident The accident happened west to the bridge 23.09.14 Intro Rels in TMS maps Managing rels Open issues 16

Managing relations… how? For integration and migration, two (complementary) approaches: � Define expected relations at types level � Analyse initial relations at instances level � Deduce expected relations at instances level � For generalisation: � Constrained-based approaches enable to define expected relations at instances level � A few studies specifically focused on relations management (Edwardes 2007; Gaffuri et al. 2008; Duchêne et al. 2003, 2012) 23.09.14 Intro Rels in TMS maps Managing rels Open issues 17

MODELLING RELATIONS: HOW? How to model relations between foreground and background � instances ? � (Too) few studies, to our knowledge � Model for relations and constraints on them (Touya et al. 2012, 2014; Jaara et al. 2013) � Object « hosting » another one thus influencing its behaviour (Picault et Mathieu 2011; Maudet et al. 2013; on-going) Extension of CityGML with a framework for relations (Bucher et al. � 2012) 23.09.14 Intro Rels in TMS maps Managing rels Open issues 18

OUTLINE Introduction � � Relations in thematic maps: role and evolution through scale � Managing thematic-background relations Open research issues � 23.09.14 Intro Rels in TMS maps Managing rels Open issues 19

Better know/describe relations Taxonomy of common thematic-background relations? � Spatial + some sematic attached that might guide generalisation? � Network section hosts … landmarks …events punctual objects… � Generic knowledge on how relations are « allowed » to be transformed through LODs? Does it vary a lot with the use case? � Typical relations attached to typical LODs? (cf. CityGML) Dependency on use case? � … Typical use cases?? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 20

Generalisation of background data � … how is it influenced by the presence of background data? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 21

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies 23.09.14 Intro Rels in TMS maps Managing rels Open issues 22

Could we set up advanced SDIs… LOD? LODn or … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …able to semi-automatically integrate external thematic data at the relevant LOD… 23.09.14 Intro Rels in TMS maps Managing rels Open issues 23

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …able to semi-automatically integrate external thematic data at the relevant LOD… 23.09.14 Intro Rels in TMS maps Managing rels Open issues 24

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …able to semi-automatically integrate external thematic data at the relevant LOD… 23.09.14 Intro Rels in TMS maps Managing rels Open issues 25

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …and to derive meaningful maps at coarser level(s)? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 26

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …and to derive meaningful maps at coarser level(s)? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 27

Could we set up advanced SDIs… LODn … Themes (high level) LOD2 Thematic-backdrop relations… Typical use LOD1 cases? Knowledge on thematic Backdrop data: MRDB structure data semantic: ontologies …ideally: on the fly? ...how to combine generalisation and migration for that? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 28

Thank you! Questions? 23.09.14 Intro Rels in TMS maps Managing rels Open issues 29