Crowdsourcing

[gcobb]

One of the things we have always complained about with the Tivo service has been that it has been hard to get problems with the data fixed. So, if Tribune mistakenly assign two different programmes with similar names to the same series, or don't realise that F1 and Formula One are the same thing we can't get it changed. More seriously, if the schedule changes we can't get the times updated.

The OzTivo setup seems to allow for users to make corrections -- crowdsourcing the corrections. At least it did in the early days -- I haven't been able to track down whether that is still in use.

Has anyone been thinking about how we might make crowdsourcing work with our service? In particular, I am interested in what are the processes and data flows involved. For example, questions like "are there any manual checks or processes involved in the normal, regular data flow?", "how would someone look at data and make corrections?", "are only some types of corrections possible (in other words -- some data once set cannot be changed)?", "how do corrected slices get regenerated?", "what delay would people see before updated data appeared on their Tivo?", "are there any problems if some corrections arrive too late for some users?", etc.

In fact, maybe documenting (creating a picture?) how we expect the data flows to work (in the normal case, and then in the case of corrections) in the UK setup would be useful for all on the team -- I would certainly find it useful. Maybe one of the non-programmers on the "How can I help?" thread would be interested in this if no one is doing it already?

Graham

[healeydave]

Hi Graham,

The Oz system does still use crown sourcing for adding and refining the scraped data, in fact I think they rely quite heavily on user input.
I may be wrong, but from what I have seen from the uk xmltv data, I am hoping that will provide us a good base, comparable to our previous tribune experience.
Once we have got to the point of testing the database population and downloading functionality, we will have a better idea of comparisons.
Either way, I'm hoping we be able to use the same system where-by users can logon to the 5wktivoguide database and make alterations to improve the scraped data where necessary.