GB1900 Newsletter 4

            Hello all GB1900 volunteers.

            Welcome to the fourth GB1900 newsletter.

            The Latest Progress Maps

            These are now online in the usual place:

            However, they are a bit different in two ways.

            Firstly, rather than a "One Name Only" map, there is “Needs Confirming”. This includes pins which have only one name but also pins which have two names which differ, and it excludes pins which we have identified as being just numbers, or spot height “benchmarks”. The idea is to exclude pins which are basically a mistake, but more work is needed on this (see below).

            Secondly, we have had to change the initial zoom level to 11, so you will see only a relatively small area. This is unfortunate but unavoidable: the server at the National Library of Scotland this runs on has been struggling to cope with the current number of pins, and if you zoom out it becomes impossibly slow.

            These progress maps show a great deal of progress! In particular, not very long ago most of the north of Scotland was empty of pins but there are now few areas where nothing has been done.

            — In northern Scotland and to some extent in Wales, many of the actual place names include accented characters. These won’t be on your computer’s keyboard, but you can include them in your transcription by selecting them from the row of accented characters below the box you type the name into. These accents matter! They are fiddly to include, so if you don’t feel able to work with them please help in these areas by inputting all the usual strings like “Waterfall” and “Pier”, and leave the accented names to others.

            — Please ensure you transcribe the entire text string. Where a place name contains multiple words please include them in a single entry containing all those words rather than having individual pins for each word or just the first word. For example "Cnoc Coille Bhàrr" should be a single pin, not a pin for each word or just "Cnoc". If it's a very long name you may want to write or type it out somewhere else before putting it in the transcription box.

            — In other areas, which are mostly complete, please help us finish up by using the “Needs Confirming” map to find an area where there is a cluster of names to be confirmed. Confirm these, and keep an eye out for names that have been completely missed. Once you have finished that, if you have a little time left, please “pan” sideways and just keep going in one general direction, confirming isolated pins as you spot them. This doesn’t take long unless you hit another cluster, and just keep going until you have to go do something else, or hit the coast!

            Processing the Results

            The new progress maps are the results of our first attempt to systematically process your work, and come up with a single confirmed name for each point. For the version of the data behind the progress map, our results were:

            Agreed text set                                   2,246,295

            Numeric data or other mistakes              58,829

            One name only                                      249,618

            Two names but conflict                            35,809

            Names conflict but only by letter case       7,807

            Names conflict                                           8,776

            TOTAL PINS                                       2,607,134

            One conclusion is that the web site is under-estimating the number of confirmed names by quite a large amount, as it currently says “1,994,962 Names Confirmed”, but we are afraid we now have no capacity at all to change how the web site itself works — this processing is being done outside the GB1900 system, in our GBHDB system.

            More generally, we have more work to do on the last two categories, the 16,000-odd pins where we have three transcriptions which do not match. The ones where the only difference is between upper and lower case letters we can apply more standardisation rules, such as saying that names which end in “Farm” should always have the first letter of each word in capitals. Most of the others can be sorted out by looking at all three transcriptions, and we expect there will be only a couple of thousand which need someone to look again at the maps.

            However, obviously more important than those 16,000 “conflicting” pins are the 285,000 pins which “Need Confirming” — by you, please.

            Finalising GB1900

            Our last newsletter wrote about the technical problems that stop GB1900 being kept running indefinitely. We still can’t give a precise date it will close down, but it will be in December. We are going to some effort — and annoying our IT people — to keep it running until then, rather than closing down in September, but it is very clear by now that this will enable the system to be truly complete.

            Completion of course also needs your continued assistance — especially with confirmations, as discussed above.

            Even if there were not these technical issues, we think it is really important the project has a clear end, so that people can start using the results without worrying that there will be a very slightly better version of the raw data in another few months. Although the work we have done so far to process the data has been writing scripts which can be re-run with new versions of the data, sorting out the harder conflicts will be a more manual process we do not want to start until we know we have a final version.

            Beyond GB1900 — and our volunteer survey

            We have had many suggestions for how GB1900 can be improved, and requests that it be made available for use with other sets of maps. Our developer left in March, so we now have no ability at all to “tweak” the system. Anyway, because it is an elderly piece of software written by somebody else, we don’t think it is a good starting point for future projects.

            What we (the Portsmouth part of the team) are doing instead is seeking funding for a quite new system. The core of this proposal is to create new software for our other web site, A Vision of Britain through Time, which in some senses is even more elderly; for example, so that it could be used on phones and tablets. However, we are hoping we can add features enabling contributors to add to the system, and in particular to add their information on top of historical maps. Obviously, this would draw on our experience with GB1900.

            This will be quite a bit more expensive than just sprucing up Vision of Britain, and we hope you can help us make the case for a more contributory system. Partly to help with this, and partly as a piece of academic research into the crowd-sourcing process, we are starting work on a survey of our contributors. We want to explore how a project like ours, where people can work on particular places they have a personal connection with, differs from a “citizen science” project like Galaxy Zoo, where unless you are a recent immigrant from the Andromeda galaxy it is hard to have any personal connection. The first phase of this survey will be “depth interviews” with a selection of some of our more active volunteers, and we will be sending them personal invitations, but the second phase will be an online questionnaire we will ask you all to complete. Of course, participation in this survey will be entirely voluntary.

            With our continuing thanks for all your efforts.

            Best wishes,
            The GB1900 Project Team
            Updated: 09 Oct 2017 04:49 AM
            Help us to make this article better
            0 0