This is an experimental project to provide analysis and summary of public comments on Bristol City Council Planning applications.
The analysis starts with extracting the lines on the planning application documents page which provide a summary of all documents submitted. Public Comments are categorised as a supporter or objector on the basis of the occurance of 'SUPPORT' or 'OBJECT' in an entry. The date of submission is extracted along with the address.
Mapping the location of submitters requires the address to be geo-located. The address latitude/longitude is found using the Google geo-location service. The address used is modified slightly to improve the quality of mapping, for example by appending 'UK' and removing parts of the address if these are seen to confuse the service. No formal analysis of the success rate has been done. On the sample application, out of 770 submissions, 15 contained no address and 28 (less than 4%) initially failed to be located. A brief examination of the map shows perhaps 5% of locations were mis-located. Manual editing of failing locations has reduced both missing and erroneous locations.
Analysis of the submissions over time is enhanced by the use of curated events which are deemed to be relevant to encouraging the public to submit either support or objection.
Analysis of opinion by distance from the centre of interest uses the location of each address, grouped in rings of 500m width, up to 12.50km and over.
The submission documents are available as PDF documents and these are scanned to extract the comment text and 'customer' details for analysis and inclusion in the map. Unfortunately, the text extraction from PDF is imperfect and some spaces are missing which adversely affects textural analysis. So far only text length has been used, and comments included in info windows on the map.
A Kit Wallace production. Issues and code (soon) on Github. Contact firstname.lastname@example.org.