Metadata Tagging

Note - this will be a bit rough initially...

Geographic Tagging at MIT

There are several ways to talk about or record geographic locations at MIT. MIT Facilities data is managed using coordinates based on the Massachusetts State Plane coordinate reference system using NAD 27 feet. (What's the EPSG code for this?). Let's call these MITGeoCoords for now. MIT people use either Building/Room (e.g. N52-200), official names (e.g. MIT Museum), or unofficial names (e.g. Infinite Corridor - is that really unofficial?).

Buildings come and go, so it's possible that a reference to a location can imply a time range (i.e. the lifespan of the building) but it makes more sense to also include a time tag for which the geographic tag is valid.

Often the precision of the specified coordinates will not match the precision of the thing being tagged. Using a designation such as 7-100LA (Lobby 7) talks about a very large area. It includes the kiosk on the North side as well as the Boswells cafe seating on the South side. In this case, using MITGeoCoords could be more precise. However most people will not know the MITGeoCoords for a specific spot. The opposite problem can also be true. If you're talking about Lobby 7 and use an MITGeoCoord to pin down an exact spot, that might not be the actual spot you mean, or you might be talking about the entire lobby rather than a specific spot. Thus we also need to capture an "intended precision" indicator.

Tagging can also be done in a computer assisted way. We could build a web interface that presents a map of an area and ask people to click on the spot they wish to tag an item with. This can also supply a false sense of precision and there will have to be good feedback mechanisms. If someone were to click on a specific spot on the wall of Lobby 7, a slightly sloppy click could be interpreted to mean the men's room that lies just on the other side of the wall.

Another issue is whether a geographic tag means a specific spot, or whether it refers to an area. The area could be an entire room, part of a room, or could encompass multiple rooms (e.g. Building 9; Building E10, second floor; Killian Court; the space occupied by a lab)

Thus it seems that we need to store at least some of the following information:

1. "Geometry type" - point, linestring, polygon, 3d-volume
2. Actual MITGeoCoords for the type
3. Building/Room(s)
4. Official Name(s)
5. Unofficial Name(s)
6. "Date/Time type" - single time, time span
7. "Certainty" information - how specific is the tag meant to be in space and time?

Note that the items in 1-6 can to some degree be computed from each other, or at least be validated against each other.

Any tagging system will be only as good as the user's ability to use it, so we don't want to make it be overly complex. There should be as much software assistance as possible.

A possible approach would be to develop a web interface that shows a map, lets the user click to create points, lines, areas, etc. and then provides feedback about what is actually included in that click. For some older buildings (e.g. building 20) it should be possible to provide historical outlines.

Such a map interface would need data that is not currently available to general users, or even to general users in the MIT Community. We will have to work with the MIT Facilities department to either let us use data or to provide the services we need. The latter would allow Facilities to better control access to the data and would allow them to keep the data as current as possible.