Work

  • On Wednesday the team had a day-long hackathon into improving search relevancy. We split into sub-teams and just tried stuff out for the day, before gathering and presenting our findings. My team looked into using bigrams in relevancy scoring. For example, if you search for “David Cameron speech”, then a page containing “David Cameron” or “Cameron speech” would score higher than a page which contained those three words scattered across the entire document. We had some good results, so I think a fuller investigation is probably on the cards.

    Interestingly, some work on bigrams had been done before, but was removed because it didn’t work properly. Though fixing it didn’t take much effort, so I’m not sure why it wasn’t just fixed back then.

  • I took Thursday and Friday off.

Miscellaneous

  • I read The Dragon Reborn (by Robert Jordan), the third book in the Wheel of Time series.

  • On Tuesday I went to Cambridge for the “How To” book tour. The book is pretty good, though I’ve only read a bit of it so far.

  • On Thursday I went to York to give a talk about GOV.UK to HackSoc.