WTF (What Type of Feature)? Classifying Features for Advanced Data - - PowerPoint PPT Presentation
WTF (What Type of Feature)? Classifying Features for Advanced Data Linking, Searching and Analysis Capabilities International Cartographic Conference, Dresden Laura Kostanski | Pier Giorgio Zaccheddu| Linda Merrin | Rob Atkinson August 2013
GOVERNMENT AND COMMERCIAL SERVICES THEME
Laura Kostanski| Pier‐Giorgio Zaccheddu| Linda Merrin | Rob Atkinson August 2013
Mildura, Victoria, Australia December 2012, Victoria Police warned motorists to not rely on maps produced by iOS6 “the official gazetteer – which is the authoritative reference for the names and locations of 384,104 places, objects and towns in the continent – contains an entry at the precise place to which Apple was directing hapless drivers until making a hurried correction on Monday” (Arthur, 2012) The Gazetteer of Australia contains over 380,000 place names, including 33 for Mildura.
A search of the Gazetteer of Australia (www.ga.gov.au/place‐names) identifies 33 records which include the name ‘Mildura’, of these there are ‐ Mildura Rural City (with the same coordinates as where the tourists were unfortunately directed) ‐ Mildura (the actual township). The former is identified as POPL, while the latter is identified as LOCB. ‐ POPL is described by the Committee for Geographic Names Australasia (CGNA) as a ‘mapped populated place’ ‐ LOCB is described as a ‘Locality (bounded), Town, Village, Populated place, Local government town, Town site’.
From the original dataset, Victorian State Gazetteer ‐ Rural City of Mildura is a LGA‐ government administrative region covering 22,214 square kilometres‐ ‐ Mildura is a township, which the misdirected tourists were intent on visiting. A centroid coordinate of the expansive LGA boundary is positioned in the exact location to which hapless tourists were directed by iOS 6.
Gazetteer entries consist of three minimum components‐ 1. An identifier (names) [N] 2. Feature type [F] 3. Footprint (point, line or polygon) [T] An identifier (N) can often be utilised multiple times for different location instances, and it is more often than not the feature type (T) which is identified to disambiguate one reference from another. However, current practices can hamper the discoverability and reliability of data for end users.
In the world of gazetteers, a feature can be described as a real world entity and a feature type is ascribed for categorization purposes (usually from a pre‐ determined typing scheme or ontology) (Janowicz, 2008). Finney and Watts (2011) have identified three use‐case scenarios common to the discovery of feature‐level information‐
‐ data providers, when developing their exchange schema to check for conformity with community‐agreed standards on feature type definitions; ‐ dynamically by a machine client, attempting to resolve and interpret an online reference to some repository content, where the reference was located in an exchanged dataset; and, ‐ by a machine client requesting, or posting respository content to fulfill some component of its
‐ It is in the typing and categorization of gazetteer data that information can be correlated and linked to make it more readily accessible and understood by both machines (for processing queries) and humans (for application and use). ‐ Machines are incapable of interpreting and understanding geospatial data in a way similar to human capabilities (Wiegand & Garcia, 2007). ‐ One of the main metadata requirements therefore is the adequate description of geospatial data feature types, so that similar details can be discovered and linked across multiple heterogeneous systems.
Based on the available literature, it can be assumed that the general issues affecting the development of truly interoperable gazetteer services (both thesauri and ontologies) include‐
Lack of clear definition of types (Zhao, Zhang, Wei, & Peng, 2008) ‐ Ambiguous referencing of types to categories (Jung, Sun, & Yuan, 2013) ‐ Missing metadata from original sources (Athanasis, Kalabokidis, Vaitis, & Soulakellis, 2009) ‐ Problems asserting ‘same‐as’
definitions across gazetteers (Wiegand & Garcia, 2007)
catalogues.
conceptualization used to achieve a shared and common understanding of a particular domain of interest’ (Gruber, 1993).
so that it might readily be distinguished from all others, while also being recognized with various relationship definitions for other types (such as the is‐ a hierarchical relationship).
for query and reasoning services required or applied by users (Janowicz, 1132).
Alexandria Digital Library Geonames.org
INSPIRE Australian Gazetteer
Presentation title | Presenter name | Page 11
administrative areas hydrographic features land parcels manmade features physiographic features Regions
Presentation title | Presenter name | Page 12
Presentation title | Presenter name | Page 13
/concept/#id /concept/BT /concept/descriptor /concept/non‐descriptor /concept/NT /concept/RT /concept/UF 9 administrative areas school districts 18 agricultural schools 350 institutional sites educational facilities academies 350 institutional sites educational facilities agricultural schools 350 institutional sites educational facilities campuses 350 institutional sites educational facilities colleges 350 institutional sites educational facilities military schools 350 institutional sites educational facilities schools 350 institutional sites educational facilities seminaries 350 institutional sites educational facilities training centers 350 institutional sites educational facilities universities 350 institutional sites educational facilities library buildings 350 institutional sites educational facilities research facilities 679 military schools 1005 administrative areas school districts 1006 Schools
Presentation title | Presenter name | Page 14
Category CODE Description A country, state, region,... H stream, lake, ... L parks, area, ... P city, village, ... R road, railroad S spot, building, farm T mountain, hill, rock,... U Undersea V forest, heath, ...
Presentation title | Presenter name | Page 15
Category CODE Feature CODE Feature Type Description S CTRM medical center a complex
health care buildings including two
more
the following: hospital, medical school, clinic, pharmacy, doctor's offices, etc. S CTRR religious center a facility where more than
religious activity is carried
e.g., retreat, school, monastery, worship S MSSN mission a place characterized by dwellings, school, church, hospital and
by a religious group for the purpose
providing charitable services and to propagate religion S MSSNQ abandoned mission S NOV novitiate a religious house or school where novices are trained S SCH school building(s) where instruction in
more branches
knowledge takes place S SCHA agricultural school a school with a curriculum focused on agriculture S SCHC college the grounds and buildings of an institution of higher learning S SCHL language school Language Schools & Institutions S SCHM military school a school at which military science forms the core of the curriculum S SCHN maritime school a school at which maritime sciences form the core of the curriculum S SCHT technical school post‐secondary school with a specifically technical or vocational curriculum S UNIP university prep school University Preparation Schools & Institutions S UNIV university An institution for higher learning with teaching and research facilities constituting a graduate school and professional schools that award master's degrees and doctorates and an undergraduate division that awards bachelor's degrees.
Presentation title | Presenter name | Page 16
Category Type(s) Administrative AF Airfields BATH CHAN LDGE OCEN SEA Bathymetric BAY BGHT COVE GULF BaysandGulfs BORE RH SOAK SPRG TANK BoresTanksandWaterpoints BCST BLDG COMM CP FARM HMSD RSTA RUIN SCHL SITE YD BuiltStructures CAPE ISTH PEN PT SPIT Caves CAVE CoastalFeatures BANK BCH BRKW ENTR ESTY LH NAVB SHOL SND STR WRCK DamsandLocks DAM LOCK ForestsandAgriculture FRNG MONU TOWR TREE HillsandMountains FRST GRDN PLAN IslandsandReefs HILL MT PASS PEAK RDGE RNGE SLP Landmarks ARCH BRK IS REEF MineandFuelsites GASF MINE QUAR OtherLandforms CLAY CLIF DSRT DUNE PL PLN ROCK SPAN ParksandReserves CEM RESV PointsCapesandPeninsulas ANCH DOCK HBR PIER PORT PortsandDocks LOCB LOCU SUB URBN RoadsandTrails BRDG FORD GATE HWY ROAD RTRK STOK TRK TUNN TrigStations TRIG Unknown CRTR DEPR GORG VAL ValleysandDepressions INTL LAGN LAKE RES SWP WTRH WaterBodies BEND CNAL DRN GLCR RCH STRM WRFL WaterCourses CNTY CONT DI PRSH STAT
from 17 National Mapping and Cadastral Agencies across Europe to create a unique european‐centric gazetteer service and data set [insert PGZ ref]. The updated EGN service provides features as (http://www.eurogeographics.org/eurogeonames)
purposes of this European gazetteer service, as the existing pre‐defined European catalogues, or nationalised thesauri had been considered to be unsatisfactory for the purposes of a harmonised data model across the jurisdictions (Zaccheddu & Afflerbach, 2009; Zaccheddu & Sievers, 2005).
Presentation title | Presenter name | Page 17
and the minority language group ones for each participating
Sorbian and (Western) Frisian.
than the data content itself, i.e. the mapping of the content was done mostly
English (and mostly displayed as such on the portal) and translated from English into the defined EGN languages, e.g. in Germany into German, West (Frisian) and (Upper) Sorbian.
Presentation title | Presenter name | Page 18
http://id.unsdis/id/catchment/567 http://id.unsdis/id/catchment/567 URL: spatial data access URL: spatial data access URL: observation data archive access URL: observation data archive access URL: live data access URL: live data access representations Basic properties provenance URL: virtual data product URL: virtual data product
Identifier Architecture
Sensor Web (image OGC 2006) Spatial Data Infrastructure
WFS WCS
Spatial databases Services
Linked Data Web
Observation Archive (Data Warehouse)
Transactions Data Marts
Services
Application Reports
Computational Models
classifications to which finer‐level feature types are mapped ‐ Beard (2011)
a multilingual, multi‐script format.
for classifying detailed feature types in use in different source gazetteers.
capabilities emerge to allow standardisation of a finer level of detail.
Presentation title | Presenter name | Page 20
Presentation title | Presenter name | Page 21
1. Support search narrowing 2. to be fine‐grained enough to allow for disambiguation
names 3. providing users with access to the original data sources to allow for validation and direct querying of the original feature type categories and descriptions
Presentation title | Presenter name | Page 22
work with UNGEGN on definition of feature categories
Methods for registering and subscribing existing gazetteer and ontologies to SIRF
Definitions of similarities between sets and instances (ie same‐as
similar‐to)
for determining similarities across heterogenous systems and the strength of these similarities (ie what are the minimum amount of similar connections required?)
Presentation title | Presenter name | Page 24
For more information Rob.atkinson@csiro.au
GOVERNMENT AND COMMERCIAL SERVICES THEME