Killingbird said

joelshitshow said
Recommended videos would be nice, based on the videos I’ve already seen. It could be based on pro or topic. Of course you’d probably have to write a proprietary algorithm for that, yeah?

Very cool idea. although I admit I have no idea how to do it!

There’re a lot of different ways that can be done depending on exactly what data is stored in the database.

• Manually wire videos. This is the worst solutions and basically relies on manually grouping videos that relate to each other.
• Choose videos on related metadata, usually tags, rating, pro, length, stakes and comments.
• Use all of the above plus track other users viewing habits (that is, correlate to what other users that have seen that video viewed next). This last one will only work if you have that information in your database (although you can start collecting it). It would be similar to that amazon feature of “other users who looked at this item also looked at…”

Implementation should assign weight to each of those values (i.e. a tag coincidence weights 10, a pro coincidence weight 20, length weights 30 – 1 per minute difference) and then order by that weight (and maybe randomize a little bit). Should be relatively simple to build a VIEW into the database that presents that information although depending on the structure of the database it may require to write the algorithm in code.