RDA-L

 View Only
last person joined: 2 days ago 

Open discussion of RDA, RDA Toolkit, and related topics
  • 1.  Geographic coordinates in place name authorities

    Posted Jul 28, 2021 11:48 AM
    In geographic place name authorities, NACO catalogers regularly record geographic coordinates of the place in field 034 of authority records.  For example:
    034 ## ǂd E0072651 ǂe E0072651 ǂf N0465653 ǂg N0465653 ǂ2 geonames
    034 ## ǂd E0072700 ǂe E0072700 ǂf N0465700 ǂg N0465700 ǂ2 wikiped
    151 ## ǂa Bern (Switzerland)
    There does not seem to be an element in RDA that corresponds to coordinates for a place in a place description.  All I see for the place entity are the following elements:
    coordinates of cartographic content of: A work that is a cartographic work that has an area of coverage that is described using a mathematical system to identify its boundaries or location.
       Domain: Place. Range: Work
    longitude and latitude of: A work that is a cartographic work that has an area of coverage that is identified using longitude of the westernmost and easternmost boundaries and latitude of the northernmost and southernmost boundaries.
       Domain: Place. Range: Work
    strings of coordinate pairs of: A work that is a cartographic work that has an area of coverage that is identified by a polygon using coordinates for each vertex.
       Domain: Place. Range: Work
    The inverse relationships relate works to places:
    coordinates of cartographic content: A place that is the area of coverage of a cartographic work that is described using a mathematical system to identify its boundaries or location.
       Domain: Work. Range: Place
    longitude and latitude: A place that is the area of coverage of a cartographic work that is identified using the longitude of the westernmost and easternmost boundaries and the latitude of the northernmost and southernmost boundaries.
       Domain: Work. Range: Place
    strings of coordinate pairs: A place that is the area of coverage of a cartographic work that is identified by a polygon using coordinates for each vertex.
       Domain: Work. Range: Place
    Am I missing something in RDA, or is there actually no element in RDA for recording the geographic coordinates of a place, rather than the coordinates of a place depicted on a cartographic work?  Is what we do in NACO on geographic authority records simply outside the scope of RDA?
    Adam Schiff
    Adam L. Schiff
    Principal Cataloger
    University of Washington Libraries
    (206) 543-8409


    ------------------------------
    Adam Schiff
    Principal Cataloger
    University of Washington Libraries
    He/Him/His
    ------------------------------


  • 2.  RE: Geographic coordinates in place name authorities

    Posted Jul 29, 2021 05:01 AM
    Adam

    The element for recording the coordinates or other boundary or point definition of a place is Place: location

    The element definition is "A delimitation of the physical territory of a place".

    As your example shows, there are different values that may be assigned to this element, reflecting the different methods of 'locating' a place. These methods include centre-point coordinates (there are several distinct methods for determining the centre of an irregular boundary) and boundary coordinates (including bounding box and irregular polygon) that are sets of point coordinates. As the LRM says "The level of precision used can vary according to the context".

    Different authority files (vocabulary encoding schemes) for Place are unlikely to record all the possible values, so the element is unrefined. It is assumed that different communities will add, as required, refinements of the element, and if an international consensus emerges, presumably RSC would be receptive to a proposal for refining the element.

    As you point out, NACO is interested in the names of places, which are not the same entities as places themselves.

    The place depicted on a cartographic work may be labelled 'Edinburgh', but the map itself, if presented as a rectangular area, will contain places that are not called Edinburgh and are located outside of the city boundary. The place depicted is the rectangular area of the map, which contains the irregular polygon that is the 'boundary' of the city. Also, the city boundary itself is not fixed: the political boundary (administrative area) is not quite the same as the social boundary, the planning boundary, or the tourist boundary. And it changes through time.

    RDA assumes that data processing can help to make sense of this. A machine can calculate if a point lies within an area, and an area within a larger area. A sheet map that includes a distinct place may not be labelled with the name of the place.

    See the Wikipedia entry on the Ordnance Survey National Grid. It includes an depiction of Grid square TF, with a caption: "The map shows The Wash and the North Sea, as well as places within the counties of Lincolnshire, Cambridgeshire and Norfolk". The place actually depicted on the map is a rectangular area named 'TF' in the grid system, and it contains a place named 'Boston'. If 'TF' and 'Boston' are treated as distinct places, and their coordinates are recorded for each 'location' using the same coordinate system, then a search for 'a map of Boston' can hit the OS map as well a street map of the town if the retrieval system can compute 'inside' and 'outside' of a specific 'location'.

    Or an agent who creates metadata could simply add relationships to all of the places that the cartographic work depicts/covers.

    The IFLA Library Reference Model scopes the Place entity as:

    "The entity place, as relevant in a bibliographic context, is a cultural construction, it is the human identification of a geographic area or extent of space. Places are usually identified through a physical object (a geographical feature or a man-made object), or due to their relevance with regards to a particular agent (geopolitical entities such as countries, cities), or as the location of an event. The place as an extent of space is distinct from any governing bodies that exercise jurisdiction in that territory. The government responsible for a territory is a collective agent. Places can be contemporary or historical, on Earth or extra-terrestrial. Imaginary, legendary or fictional places are not instances of the place entity.

    A place can have fuzzy boundaries. The boundaries of a place can change over time (such as a city that absorbs adjacent suburbs) without changing the identity of the place for bibliographic purposes.

    As it can be a moving frame of reference, the entity place is not necessarily identified by its geospatial coordinates alone."

    ------------------------------
    Gordon Dunsire
    ------------------------------



  • 3.  RE: Geographic coordinates in place name authorities

    Posted Jul 29, 2021 10:45 AM
    I agree with Gordon that Place: location is an appropriate element to use for coordinates for a Place. However, I think this is only a partial answer.  Geographic databases may use coordinates as identifiers.  The definition of identifier for place is "A nomen that is an appellation of place that consists of a code, number, or other string, usually independent of natural language and social naming conventions, used to identify a place."  The NGA GEONet Names Server (GNS) assigns a unique MGRS (Military Grid Reference System) coordinate as well as a unique set of longitude and latitude coordinates for each place in its database. If you search 34SGH3182904816 in GNS, you get 1 result--Piraeus, Greece.  Ditto for the DD coordinates 37.961389, 23.638889  Thus, GNS meets RDA's definition of a VES and the coordinates meet the definition and scope note of identifier: A recording method that is a machine-readable string that is assigned to an entity in order to differentiate the entity from other entities within a local domain, or a notation for a term from a controlled vocabulary that is assigned to an aspect of the entity. Includes an identifier or notation taken from a vocabulary encoding scheme." I don't know much about Ordnance Survey National Grid, but usually grid reference systems are designed to yield a unique string, so  perhaps these might also be appropriate to record for identifier for a place. Note that the longitude and latitude coordinates given in the description of a place in GNS are not boundaries--they are centerpoints or close to it (they have to avoid putting the centerpoint on a private residence).  I agree with Gordon that boundaries are fuzzy as LRM says, but I think providing coordinates for the boundaries better serves the RDA definition of the element location: "delimitation of the physical territory of a place." So I think if I wanted to use the element location to describe the place Piraeus, I would give 4 coordinates for the rough boundaries.  If I wanted to use the element identifier to provide an appellation for Piraeus that could serve as a unique label, I would use the MGRS coordinate or the DD coordinates string.  That is just my opinion of how to use the element Place: location to provide the most helpful information to the user. I'm not saying it is wrong to record centerpoint coordinates for Place:location--just not as useful if the idea is to give the user information about the boundaries (delimitation is determining boundaries for something) of the physical territory of a place.
     
    I realize it may seem weird to some to say that coordinates can be an identifier for a place, but, an identifier is just a label we can use to refer to an entity that is unique within a local domain.  The expectation is that a user would be able to search the identifier and retrieve a description of the entity with several elements recorded for it, and that complete description is what tells the user which instance of an entity we are talking about with that label.  Similarly, the identifier Q232462 means nothing until we look it up in Wikidata and read the description for the person named Ani Di Franco.

    Curiously, you cannot record Place: location as an identifier (the only method allowed is structured description) and you cannot record Place: identifier for place as a structured description.  So what's a cataloger to do if they want to record "34SGH3182904816" as both Place: location and Place: identifier for place for the place called Piraeus and cite GNS as the VES? Honestly, I don't know. I find the instructions for Place: location very strange. It does not follow any pattern of other elements that have the recording method structured description.  One pattern can be found in Place: category of place, in which the structured description instructions tell you to use a value from a VES and record the VES or if no term is appropriate, then record using unstructured description.  Another pattern can be found for super-elements (where sub-elements are combined in a string) and then you are told to apply a string encoding scheme and record the type of string encoding scheme used.  There are elements like aspect ratio that do not have any instructions about data provenance, but there is still more there than there is for place. This is the only "instruction" for recording Place: location as a structured description: For general guidance on structured descriptions, see Guidance: Recording methods. Recording a structured description."

    Kate







  • 4.  RE: Geographic coordinates in place name authorities

    Posted Jul 29, 2021 02:43 PM
    Kate,

    I agree that GNS can be used as a VES, but the coordinates are not used as identifiers for the concept of a named geographical area in GNS. To use the example you cited from GNS:

    The authorized access point for place is "Peiraiás" (Provenance: GeoNet Names Server approved name)
    The identifier for place is: -826292 (Provenance: GeoNet Names Server's Unique Feature Identifier)
    The location is: 37° 57' 41" N, 023° 38' 20" E or 37.961389, 23.638889 or 34SGH3182904816

    Part of the reason for the strangeness of the "location" element with regards to the instructions is because the value must be a "A delimitation of the physical territory of a place." A VES is a value vocabulary for concepts. In order for there to be an applicable VES for "location" it would have to be a vocabulary for the concept of a delimited space.

    I suppose, in theory, one could set up a VES for Earth coordinates, with definitions for those coordinates, but why do that when one can simply record coordinates as an attribute of a place instead?

    As for whether or not the value of "location" is structured or an identifier, I would argue that for coordinates, the value is a structured datatype, because it can always be structured based on your position on the globe with relation to the equator and the line running north-south through a point in Greenwich. Other methods for geopositioning could also be used for constructing a value.

    There could be an argument that an unstructured description of the location of a place is possible. For the place known as "My House", I could say its location is so and so many miles north and west of a certain landmark and is bounded by "A" street and "Z" street.

    Damian






  • 5.  RE: Geographic coordinates in place name authorities

    Posted Jul 29, 2021 10:46 AM
    Gordon,

    Thank you for the explanation.  I received a different answer that coordinates are a type of identifier, but it makes more sense to me that the appropriate RDA element is Place: location.

    Adam

    ------------------------------
    Adam Schiff
    Principal Cataloger
    University of Washington Libraries
    He/Him/His
    ------------------------------