For a while now with bits of technology I’ve written for The Guardian both in curating conversations and collating and whitelisting people talking about a subject, I’ve been interested in sentiment and identity and who talks about what and in finding ways of seeing the data and letting people search it. 

As a small illustration for my talk at the start of Bonnier Hackday I rewrote part of the twitterfall app that I’d built for Activate. I then bolted onto it some faceting around people and also added a fulltext search using AppEngine’s secret weapon the SearchableModel.

I’ll try and document and clean up the code and then send it up to GitHub for people to play with.

This is basically what the app does:

• It hoovers everything with a relevant hashtag and keeps hoovering through the magic of cron and the task queue.
• It lets you explore who said the most and what they said.
• It also lets you search for a phrase within that hashtag.

It is a simple thing, the product of about 5 hours work, but has a lot of interesting learning for some bigger things which are being called Project 3 at ninety10group.