UvA information scientist builds parliamentary search engine for De Correspondent

31 January 2017

Maarten Marx, information scientist at the University of Amsterdam, has built a search engine for online news forum De Correspondent that makes it possible to search through motions made by parties in the Dutch parliament (Tweede Kamer). This allows voters to check which election promises were actually upheld by political parties.

The search engine watstemthetparlement.nl (in Dutch) is an inventory of all motions, amendments and laws passed since 1995 and gives an insight into the promises made by political parties in the past cabinet term and how much of these were actually realised. The project came out of a joint project proposal submitted to Amsterdam Data Science by Maarten Marx and De Correspondent. The search engine was built with Elasticsearch. ‘Building the search engine wasn’t difficult, but gathering data proved to be quite a challenge,’ says Marx. Political scientists from Leiden University were able to help by providing data.

Challenges

In the run-up to the elections, Marx and his student-assistent Paul Schrijver will further develop the search engine. The aim is to make all parliamentary votes available from 1946 onwards. Currently, only records from the digital age, from 1995 onwards, are available. Older records need to be scanned and digitalised. One of the challenges here is that text extraction is complicated by outmoded language use. In addition, it proves difficult to stay up-to-date without having to make adjustments in between, because of the way in which parliament publishes its information. Also, the correct formula to rank votes in order of ‘unpredictability’ has not yet been found.

Active role for readers

To make sense of all the information in the search engine, De Correspondent has asked its readers for help. Findings by readers scouring the database will be used by De Correspondent for articles.

Published by  Faculty of Science