2005-01-05, 08:26 PM
I am running through some thoughts in my head about building a correlation database like tivo that offer show suggestions based on your recording habits.
I wanted to start a discussion and run some thoughts and questions to my peers.
--What we have--
We have a list of shows a user recorded or plans on recording.
I know for zap2it users we have a database that lists the categories associated with a show.
So we can then associated shows like user X watches Stargate, Stargate Atlantis. We can use categories to build these associations.
Battlestar Galatica is a new show that the user is not recording. It is on the same channel and in the same categories as Stargate. We can then automatically record it.
We can also have logic that determines what actors are in the shows you like and then any show that might have those actors in it.
Now the list of things we would need
1. a database that the different users can connect to and upload their lists of recordings, and future recordings too.
I imaging an exe that would run nightly (or weekly) sending a list of shows you watch. Downloading possible associations.
Filtering out the shows you have said you do not want ever recorded again.
2. Some database structure that would show relationship between shows. (I am not sure how this would be done.)
I am not a db guy. Would we want a table of shows. That would then link to a foreign key to a table for the shows associated with this one.
How many users would have to have both shows before the correlation kicks in? 20%? Maybe the user would specify this.
Strong Correlation 60%
Moderate Correlation 40%
Week Correlation 20%
Any thoughts on how we would do this? Should the correlation be in generes only.
I know that zap2it has actor information and genere information. I would like other users to be able to use this as well. So actors and genere would be optional checks we could do.
Jorm
I wanted to start a discussion and run some thoughts and questions to my peers.
--What we have--
We have a list of shows a user recorded or plans on recording.
I know for zap2it users we have a database that lists the categories associated with a show.
So we can then associated shows like user X watches Stargate, Stargate Atlantis. We can use categories to build these associations.
Battlestar Galatica is a new show that the user is not recording. It is on the same channel and in the same categories as Stargate. We can then automatically record it.
We can also have logic that determines what actors are in the shows you like and then any show that might have those actors in it.
Now the list of things we would need
1. a database that the different users can connect to and upload their lists of recordings, and future recordings too.
I imaging an exe that would run nightly (or weekly) sending a list of shows you watch. Downloading possible associations.
Filtering out the shows you have said you do not want ever recorded again.
2. Some database structure that would show relationship between shows. (I am not sure how this would be done.)
I am not a db guy. Would we want a table of shows. That would then link to a foreign key to a table for the shows associated with this one.
How many users would have to have both shows before the correlation kicks in? 20%? Maybe the user would specify this.
Strong Correlation 60%
Moderate Correlation 40%
Week Correlation 20%
Any thoughts on how we would do this? Should the correlation be in generes only.
I know that zap2it has actor information and genere information. I would like other users to be able to use this as well. So actors and genere would be optional checks we could do.
Jorm
WinXP Home
3 X MVPs
Hauppauge 250 MCE
Hauppauge 150
P-4 2.4 GHz / 768 megs Ram
610 Gigs Of Media Storage
3 X MVPs
Hauppauge 250 MCE
Hauppauge 150
P-4 2.4 GHz / 768 megs Ram
610 Gigs Of Media Storage