[Talk-GB] Conflation of TfL data into OSM (was: Is TfL data allowed on OSM?)

Whittaker, Ed ed.whittaker at sweco.co.uk
Mon Jul 11 16:20:40 UTC 2022


Firstly, apologies for the slow response and for creating the issues discussed here. Unfortunately it corresponded with a period of leave for key staff and there is an ongoing issue with some members ability to post to the mailing list.

Secondly, we know that we have a lot of concerns to digest and improvements to make to our methodology so we have paused further conflation until we can address issues in a way that everyone is happy with. We've attempted to engage through such things as a talk at London Geomob, updates to the project wiki and postings to talk-gb-London though appreciate we need to do much more and have perhaps not targeted the right places.

We've tried to be careful during early conflation and believe we have imported some useful data, especially bike parking. However, we understand the frustration with some of the edits to cycle lanes as a result of the age of the TfL surveyed data. We'd be happy to review these edits and roll back where necessary (e.g., where a more recent tag has been overwritten/removed).

We've a number of tasks to address to get this conflation back on track,

** GitHub Issues Resolution
-------------------------------------
We are talking to CycleStreets regarding helping us to close these issues. We hope to see some movement on these very soon.

** Process for merging TfL and OSM Data
---------------------------------------------------------------------------
We know there is plenty of good data in the TfL CID dataset that is worth conflating though know we need to build a stronger set of process to make sure more recent OSM edits are not overwritten by older CID data. Our initial thoughts are to conflate data where data is not present (e.g. importing a bike stop where non is present in OSM). Where a feature already exists in OSM, we need to check the history of edits and not remove/overwrite tags that are newer than the CID survey date. We also need to make sure that conflated tags/features do not conflict with those already coded in OSM. Do people think this sounds like the start of a sensible strategy that we can expand upon in the wiki? We've considered other approaches (Bing Aerial imagery - Not detailed enough, Bing Streetscene - Too old, Mapillary - Potentially useful on main roads though likely too dated elsewhere)

** Wiki Updates
----------------------
Whilst we've expanded the wiki and tried to follow the guidelines, we know we need to include more process information. The main change will be expanded the process section with a detailed description on how we plan to  merge TfL CID data without impacting existing good OSM data.

** General Process Following
---------------------------------------
We've tried to follow the OSM conflation guidelines though need to do more. This includes better changeset comments and  username bio's referencing the conflation project wiki

I hope some of this has helped answer questions and show we are keen to engage and get this process right.

Ed Whittaker
Senior Transport Planner
Sweco UK Limited | Solihull 
Telephone 0121 711 6600
ed.whittaker at sweco.co.uk
www.sweco.co.uk

LinkedIn | Instagram
Reg. No.: 2888385 | Reg. Office: Leeds (Registered in England and Wales) 
Reg. Office Address: Sweco UK Limited, Grove House, Mansion Gate Drive, Leeds, LS7 4DN 
For more information on how Sweco processes your personal data, please read here.




More information about the Talk-GB mailing list