{"id":591,"date":"2010-12-10T10:32:11","date_gmt":"2010-12-10T10:32:11","guid":{"rendered":"http:\/\/blog.soton.ac.uk\/webteam\/?p=591"},"modified":"2010-12-10T10:33:01","modified_gmt":"2010-12-10T10:33:01","slug":"studying-the-mps","status":"publish","type":"post","link":"https:\/\/blog.soton.ac.uk\/webteam\/2010\/12\/10\/studying-the-mps\/","title":{"rendered":"Studying the MPs"},"content":{"rendered":"

Last night Tony Hirst was trying to work out the birth-places and universities of the current UK MPs<\/a>.<\/p>\n

Here’s what I managed to produce while hacking with a glass of wine & TV on:<\/p>\n

\"\"<\/p>\n

I did some seat of the pants data munging so I thought it’s worth explaining the process I used:<\/p>\n

List of MPs<\/h3>\n

For the graph of MP’s birth-decade<\/a> I did earlier in the year, I used a dbpedia relationship which gave me a nice little subject ‘member of 2005 UK Parliament’. This time around I can’t find anything so easy for the 2010 Parliament.<\/p>\n

Solution, I used http:\/\/en.wikipedia.org\/wiki\/MPs_elected_in_the_UK_general_election,_2010<\/a> and a dirty little script…<\/p>\n

my @tables = split(\/<table\/, join(”,<>));
foreach my $tr ( split( \/<tr\/, $tables[3] ) )<\/address>\n
{<\/address>\n
my @td =\u00a0 split( \/<td\/, $tr );<\/address>\n
if( $td[5] =~ m!”\/wiki\/([^”]+)”! ) {\u00a0\u00a0\u00a0 print “$1\\n”; }
}<\/address>\n
(sorry for the godawful formatting, having hassles pasting code into wordpress since our upgrade.)<\/address>\n

This then got munged in a text editor to create a .ttl file (much nicer way to express RDF than XML, esp. when doing hacky scripts)<\/p>\n

This gives me this: data<\/a>, [Browse<\/a>].<\/p>\n

Later I did something almost identical to produce a file adding affiliations as a party label and as an icon red\/blue\/yellow\/other.<\/p>\n

This gives me this: data<\/a>, [Browse<\/a>].<\/p>\n

In retrospect I could have done this in one go, but it was late. Note the raw data of this file is just of the N-Triples format which is really easy to create and easy to import as RDF.<\/p>\n

Making the Map Data<\/h3>\n

I then wrote a quick PHP script using my own Graphite Library to turn this data into geocoded RDF. eg. each resource as rdfs:label, geo:lat, geo:long and also an icon predicate I made up for the day.<\/p>\n

View code: http:\/\/graphite.ecs.soton.ac.uk\/experiments\/parlibirth\/mpmunge.php<\/a><\/p>\n

View output: http:\/\/graphite.ecs.soton.ac.uk\/experiments\/parlibirth\/mpmunge.php<\/a><\/p>\n

As it’s a one-shot, I’ve just hard wired the relationship as “http:\/\/dbpedia.org\/ontology\/birthPlace” and then I just grab the results using curl. eg.<\/p>\n

curl http:\/\/graphite.ecs.soton.ac.uk\/experiments\/parlibirth\/mpmunge.php<\/a> > born.ttl<\/pre>\n
…and then modify to “almaMater” and repeat.<\/p>\n