Digitisation of biodiversity information has high priority for the GBIF Nodes. Most of the specimens in herbaria and natural history museums are not yet available in any digital database. The Volunteer Portal provide a simple tool for volunteer particiaptory registration of information from specimen labels.
Need for a gazetteer in the background. Principle idea is that this would allow for transcribing and georeferencing records simultaneously. Ideally, this should be linked to geographic objects (polygons, lines etc) when names describe objects which are linked to this type of information - example; administrative areas and waterbodies. However, just point information is a useful starting-point.
Out of the top of my head two alternatives:
Using national sources, such as e.g. karverket's stednamnebase for Norway. Either by download and local cache of data or by API services?
Using the geonames api. This does not come with link to geographic objects though, just points. May be not to difficult to match.
When starting to work with data-sources (literature) with more than one records of information per page (as a contrast to museum specimens labels) there is a need to annotate the records that has been transcribed (and georeferenced) to the document. This in order to view what have been transcribed before. Probably advantageous if this can be colour coded with records that are a) only transcribed, b) approved (either by curator of project, or verified by multiple matches - depending upon model for aprovment chosen), c) or conflicting (i.e. different users have entered contrasting information).
As a consequence of #1, there is a advantageous to have the transcription and georeferencing interface in one window (or in different tab). I.e. being able to visually verify, or possible, modify the georeferencing as transcribing.
As a second point, also related to visualize (and store) geographic objects mentioned in #1, showing geographic objects could be nice.