Personalised Ratings

This could be used for films, music and perhaps other subjects. Let's say films for starters. A user registers on a website, and is asked to browse a list of films and give a mark out of ten to any they have seen and have any feelings for. They can mark as few or as many as they like, but for an accurate a service as possible they should fill out as many as they can, and then update their list as they see more films.

The user will then be able to request a review of a film they haven't seen. The rating of the film is obviously based on the reviews of other users, but the marks are calculated using weightings. Marks from someone who seems to generally like the same kind of stuff as you you are given a higher weighting than those from people who generally disagree with you. This way, you get to find out what a like-minded person would make of the film. I haven't really thought very much about the statistical and algorithmic details of this, admittedly.

The intention is to empirically determine how much people like something without having to worry about details. You might not even not be able to really put into words what it is that you like about a particular film, and it is my belief that someone else who also feels this way about the film will probably have similar reasons for liking it. The system would require quite a lot of data to be of any use, but once it's up and running it could be quite useful. For instance, you're at the video shop and you want to know what someone "like you" would make of a film. You could SMS your user number and film title to the service and get back a personalised rating. Might hopefully reduce incidence of the phenomenon of renting out a film that got a glowing review and finding that you think it blows.

I am unaware of anything like this in existence, but suspect that it's out there somewhere. *Sits back and awaits cries of "This is so baked it's burnt the house down"*.