Improving and extending the Gazetteer

Introduction

About the Gazetteer

The Genuki gazetteer provides the ability to define the names of places that each of our place nodes covers and also co-ordinates enabling us to show where to find them on a map. So when a researcher finds a place name they can use our gazetteer to access the information we may hold about it. This is a modern variation on the information we provide  in the Gazetteer topic sections in our nodes which mainly holds extracts from historical gazetteers which provide a  description of where a place is in relation to other places in a county along with a description of that place.

We have developed the gazetteer over a number of years based upon a number of data sources, and the technology available to manage the data has also developed over that time. The UK+Ireland is a large geographic area and as maintenance of our gazetteer data has been devolved to various individuals the status and quality of the data that we hold varies by geographic region. This is something that needs resolving before we can start providing additional functionality based upon the gazetteer. So before starting work in this area you should become acquainted with the detailed description of the gazetteer and the functionality it can provide.

Key Priorities

There are three main activities for a maintainer, broadly in this order of importance.

• Correcting errors in existing entries
• Improving the accuracy of existing entries
• Expanding the range of entries

Do not assume that all the data in the gazetteer is correct as just a few may not be quite right, especially if work was done on them by a volunteer without local knowledge. Some have been found associated with the wrong place node, especially when there are multiple places with the same name in a county. Then there can be the odd one in the wrong county due to confusion caused by county boundaries and where county boundaries run through a parish.

Correcting errors

We do not have an automated process for identifying errors: ultimately any place can be in 

Listing gazetteer entries and starting points for revising them

In order to gain an idea of the data we currently hold, take a look at the overview of Gazetteer nodes by county. That identifies two sets of things that need attention, places with an approximate point location, and those without an associated place node. It does not currently highlight anything relating to gazetteer entries relating to areas such as parish boundaries. That is functionality that is being investigated for future development and is thus is currently excluded from this document. The list that this produces can be long, and when working using the list as a source it can result in slow refresh rates. So to avoid this, there is a filter at the top of the list which enables you to produce shorter lists by selecting just those entries starting with a particular letter.

To gain a more focused view of the gazetteer entries associated with a place node, view that node and click the Gazetteer tab which is just to the right of the Edit tab. That gives a list of those entries with basic information about each one and you can also see them marked on a map along with other nearby places that are in nearby Place nodes. You can also get to this screen by clicking the Related links in the previously described lists.

Approximate locations

England/Scotland/Wales

The approximate locations are the first things that need to be addressed as the tools available to help deal with entries not yet linked to a place node rely on having a precise location. This description excludes the techniques available to handle the Irish entries as that data was produced in a different way and needs a different approach to resolve that.

The map most likely to contain places needing a more precise location is the NLS map. So click the link to edit the gazetteer node, select the NLS map in the widget,click the icon to change a location and drag the marker to the new spot. 

NLS use osgb36 lat/lon so can be about 100 yards wrong, but that is close enough for gazetteer point location. If you want to be pedantic switch to a layer with one of the other maps and make fine adjustments. The location recorded is the centre of the bottom line of the square round the marker.

The approximate locations tend to be that of the kilometre square on the Ordnance Survey map on which it appears. So the place is likely to be to the NE of the approximate location. Then change the marker  to say it is an exact location and save the node.

Now some places have been added that were extracted from more detailed maps such as the OS 25 inch maps. These are much smaller named locations such as farms or isolated building in rural areas and as such do not appear on the NLS one inch maps. These can usually be found in the GB1900 which contains a more exact location. The lat/lon from this source source can be pasted into the Data field below the map in the widget where you need to click the Replace text. If you want to check that this is a correct value you should view the NLS 25 inch map for that area. We have managed to find a map layer, currently called NLS 1900, that lets use this map.

Ireland

OSM has townland boundaries on their maps and when you zoom well in the names appear alongside the boundaries. A technique that works well is to use the the list of entries within the parish.  From the list of entries requiring work, click the parish name which takes you to the parish node. The click the Gazetteer tab entry to the right of Edit and this gives a list of all entries in that parish, a map of all the nearby entries, and another map of just those entries linked to the parish.

Look at the map to see where the markers for approximate locations appears and zoom in to find the name of a nearby townland boundary. Then go to the list at the top to edit that entry as you now know where it is. You can then work outwards from this point slowly filling in all the locations within the parish. There may already be some existing entries that are not in the centre of the townland, so these may benefit from some adjustment. The entry for the parish itself should either be the centre of a townland bearing the same name, or if there is actually a town with that name then at that point. Otherwise if it is just an area, check that the parish location is in the centre of the area once you've set all the townland locations.

As you get near the end of dealing with the list there may be some townlands you can't easily find. One technique that can help is to look at the final map and zoom in tiil the townland names appear. Then pan round the map looking for holes where an entry has been missed and also going round the edges. Another technique is to do a Google search as there is a townlands site which usually contains them and displays a map for each. Zoom out from that map and compare it with ours to see exactly where it should go.

The bottom map is also useful for spotting errors where a different townland with the same name has been associated with the parish. For those find the right spot on the map and change the location.

Then there some entries that ar not marked as townlands associated with parishes, and these will have a red icon on the centre map. If these are duplicate entries for townland entries that we already have, then delete that red one. Others can be place names and frequently bridges and cross roads. Set an exact location and associate them with the parish place node using the parish that the townland area is in.

Previous techniques used OSI maps but this is now considerably harder as OSI have improved their mapping website so that you can no longer select maps by location, you have to search by name within their site.

Associating entries with place nodes

In the genuki-1 gazetteer we had a field containing the url of the genuki-parish page that covered this place. This was a little restrictive as if the place page url changed the gazetteer entry also needed to be changed. When we changed to Drupal we also changed this to a much more efficient entity reference. This is in a Drupal field we have called Place Unit. So to add this missing information we need to fill in this field. Click Add item and you are shown a search screen from which you can choose place nodes. Add the name of the current county in the filter box at the top of the form and some characters from the place title to see a list from which to select the place node. It was not possible to specify the current county if the filter by default, but it does remember the last one used.

Where it currently finds more than one entry with that name there is nothing to visually help decide which is the right one. So try the first one and save the node and look at the changed node page and look at the Place Unit field.  Wave the mouse over the link to see the url as that can be used to confirm whether it is the one you want. It also shows you the county and place code in that place unit which can also help show which one it is. If you have selected an invalid place, then go back into edit and correct the value or clear it.

Now we do first need to know which is the place node this should be. The Vision of Britain site is the one that can help with that as it can tell us the administrative regions containing that location. This does mean that we need to sort out approximate locations first.

Correcting problems with gazetteer entries

- Wrong location

The simplest to fix. Just drag and drop the marker to the right place on the map and Save.

- Associated with wrong parish

This needs a change to the Place unit field. Remove the tick in the box against the wrong parish name. Then click Add item as described in the previous section to add the right one. Then Save the node.

- In the wrong county

• Go to the Section field and change it to the correct county.
• The Place code will need changing as the first three characters are the Chapman code for the county. So put the appropriate ones in. If there is already a place with that code in the new county additional changes may be required as the code must be unique. If not a warning message will appear when you try to save the node and you will have to fix that.
• The final field to change is the Url alias. Again this contains the Chapman code and a similar change to that in the place code must be made. Again the value must be unique, and if not you will discover that when you try to save it.
• Save the node and watch for any Errors being reported at this stage. You will need to correct them before it will save.

- Deleting Place nodes

If you decide that Place node is no longer required, there are quite a few things to consider and things to adjust in order to ensure nothing is lost.

• Church nodes are associated with a Place nodes as they both use the same Place code. So find the code used by the replacement Place node, and change all the Church nodes associated with the Place node being deleted to have the replacement value. The churches will then be listed under that node in the future.
• Change the url of any Plain nodes associated with the Place being deleted, and add links to them from the replacement Place node.
• If any Topic nodes were associated with the one being deleted, then their Place codes will need changing as this is how they are linked to the Place node.
• Then move any remaining content from the one being deleted and put it somewhere else.
• Then there is just the Gazetteer node to deal with. Change the Place unit to point to the replacement Place node.
• Finally unset the Primary flag in the gazetteer entry. Multiple gazetteer nodes can be associated with a Place node with the Place unit field. Only one of these should have the Prime flag set usually the one with the same name as the Place node.

Adding parish and other boundaries

This is still at the stage of investigating what is possible and techniques to add boundary data. You can gain some idea of what can be achieved by looking at Lytham (you will currently need to use scroll-wheel zooming). That of course is not an example of how the exact way it could be presented to a user, it is just a tool to help with development.