Week 12: 13-10-2017

Dator day

Data day #33 happened. The room was… bijou. The planning with the website team session was constructive. Much, much better than last time. Really much better.

Chris from business systems came along for the first time. Dan came to grudgingly accept it’s not like the old days anymore. Robert tried handwriting. Michael came dressed as his granddad.

We agreed we’ll:

Showing and telling

Tuesday, as ever, was show and tell day. Samu gave an overview of the fixed query API improvements (the bits of the API that drive the beta website). Raphael presented on machine learning before heading off for his GDS data science accelerator graduation. General applause, Raphael.

Wednesday was the all staff briefing. Dia talked through the Data and Search 101 presentation. Which was very clear and seemed to go down well.

One world, one web, one team

Matt, the backend lead for the website team, gave a talk about the architecture of the beta.parliament.uk Rails application. Most of the data and search team went along to find out how our data becomes web pages.

Dan gave Jamie and Jeanette moral support at the digital portfolio board where they presented work on the new website. Dan had a cameo with a live demo of current website search. Jeanette helpfully drove the tabs.

Dan went to one website stand up this week, and a backlog review session. Aidan definitely went to more than one stand up. Michael thinks he went to one but, much like the Sundays, can’t be sure.

Domain modelling

Anya, Silver and Michael spent Tuesday turning last week’s Written Questions and Answers domain model into a first pass data model. They spent some time with Sally Carter, the House of Lords gatekeeper for written answers, and the proposed model seemed to make sense. It’s quite different from how things work at the moment but would have the advantages of saving effort, reducing errors, having fewer documents and more linky data, and making a better, more linky website. Time will tell.

Anya and Michael met with Ben, Colin and Ed to prioritise which bits of the domain model they should be working on next. The results are in the Doing, Doing next and To do columns on Trello.

Data quality

Chris has been working on a Lords’ inflation tool (still don’t ask) for Ben. It started life as a massively over engineered Sharepoint / Access odyssey and ended up as a single page spreadsheet.

Michael spent some time with Angela running through the Red Book database and trying to document it. Red Book is the system used by the House of Commons to manage Select Committee data. From work so far it looks to be a similar shape to the Committee Office Database (a similar system used in the House of Lords) with evidence sessions belonging to zero or one inquiries and inquiries belonging to zero or one committees. More work is needed.

Data platform

Work to import and publish data about government roles (departments, positions and incumbencies) is now complete.

The Member photo import to the data platform has been deployed and is now running.

The search product team had their first retrospective and planning for sprint 5. Caroline helped out. The team went through the backlog and agreed on the prioritised items.

To anticipate possible new search features, work has been done to categorise Parliamentary material published on the web by inferring attributes from the huge variety of inconsistently structured URLs. The work is generally referred to as search hints. Raphael has continued the work initiated by Samu and has been adding regular expression rules to capture currently indexed Parliamentary sites. At time of writing there are over 1,000 rules. Exclamation mark here.

Raphael took Dia through search hints. Dia was happy and impressed.

Robert, Samu and Raphael also met to chat about the progress on search hints. All is on target. Robert met with Liz in preparation for analysis on search once the beta is officially included in the public index. Robert also met with Lopa to hear more about “webmaster” things and beta preparation.

Robert and Colin met to chat about Members and search. A pleasant time was had.

Measuring things

Matias started to explore the JIRA API with the goal of building reports from the data. So far it’s going good, except for the fact it has a maximum result set limitation. Work continues.

Sara and Liz met with Elise from the House of Commons Library to talk about some research she is doing using the European survey on political attitudes. She’s interested in finding groups of people based on their attitudes. They discussed approaches to analysis, gave her a few pointers that are hopefully helpful, and chatted about how she might test and verify findings using a different year’s survey results.

Aidan, Liz, Dia, Jamie, Trine and Lopa met to discuss tools and next steps in establishing the ability to run A/B tests for search.

Sara began setting up the analysis for A/B testing the display of document URLs in the search results page. This involves creating the measures we want to track and defining baselines to compare to the test results.

Machines that do learning

Back in September, Jianhan submitted evidence to the House of Lords Select Committee on Artificial Intelligence. All the written evidence is now published. Jianhan’s contribution is here.

Phil from the Indexing and Data Management Section of the House of Commons Library went with Raphael to the GDS accelerator graduation / wrap-up event. He’s planning to pass on notes to the data science people in the library. Raphael thought his presentation went okay. He’s writing a PDS blog post and there’ll be a GDS blog post in a few months to see where the projects ended up.

Angela from the website development team has been accepted on to the next data accelerator.


Robert and Dan had a chat about using the GDS digital academy as part of Julie’s capability work. Both training and recruitment were covered.

Aidan’s truly terrible week

Aidan had a truly terrible week. There’s no way to make this sound fun so:

Did anybody say blockchain?

Colin said blockchain. But jokingly. So that’s fine.

Things that caught our eye