A fun example of structured data from the LATimes, which showed the popularity of dog names in LA County by postcode.

This is from one of the sessions at BeebCamp2, a BarCamp like event for BBC staff with some external folks like Suw, me, Charlie Beckett and others. Charlie has a great post on a discussion he led about user-generated content and what it adds to news, video games and also Twitter and Radio 4.

Eric Ulken, was the editor of interactive technology at the LATimes. He was one of the bridges between technology and the editorial

News organisations:

• We collect a lot of data but don’t use it (We always thought that was a shame. We had a computer-assisted reporting team at the LATimes, wouldn’t it be nice if we used that.)
• What online readers want from us is bigger than ‘news’ in the traditional sense
• We need to be an information soure.

They did a homicide map, which mapped all of the murders in LA in a year on a map and which illustrated a blog that reported all of the murders in LA County in a year.

The project was well received, and they decided to develop a data desk. It brought together the computer-assisted reporting unit, investigative reporters, the interactive technology team and the graphics team to bring together the data desk. They all sat together in the newsroom. A lot of synergies were created. The Times had 10 to 15 investigative reporters on different desks from different disciplines.

Ten bits of advice:

1. Find the believers.
2. Get buy-in from above
3. Set some priorities
4. Go off the reservation (We had a real problem with our IT department. They had their priorities and we had ours. We invested in a server system using Django.)
5. Templatize. Never do anything once. Do things you can reuse.
6. Do breaking news. There is data in breaking news. They did a database of the victims. They added information to the database as it became available. The database was up in 24 hours after the crash. They had built most of the pieces for previous applications. (There was a question about accuracy. Eric said the information was being gathered, but it wasn’t structured. The information was edited by a line manager.)
7. Develop new skills. They sent people out to workshops. They had hired a Django develop who was also a journalist. He taught Django to others in the office.
8. Cohabitate (marriage is optional). The investigative reporters and computer-assisted reporters still reported to the pre-existing managers, but by being together, they saw possibilities for collaboration without reworking the organisation.
9. Integrate.
10. Give back. They worked to give back to the newspaper.

They used Javascript to add this to other parts of the site. They created these two datasets from the train crash and the homicides, but they also have used publicly available data in their projects. He showed their California schools guide. Apart from the standard data analysis available from state and national educational agencies, they also created a diversity rank that showed the relative diversity of the schools. They did do some reporting on the data. In analysing the schools data, they found discrepancies in reporting about the performance of the schools.

In a slightly more humourous example, he showed dog names and breeds by postcodes.

UPDATE: Eric has added some more details in comments below, and you can follow Eric’s work and follow his thoughts on his site.