Say you have a hot geolocation mobile app that you are building and you wanted to use whit.li in as close to a near real time manner. For example when I open my app I want to see how I compare with those within a limited geographical range. If I have 20 users that I need to compare myself to and rank based on compatibility what kind of performance could I expect?
I notice the current demo app loads the list of people you are compatible with very quickly (I'm assuming it's cached) but when I actually click on a person to see more it takes > 20 seconds for the detail view. I'm assuming the api is doing a comparison right at that moment of you and the person you want to see more about. If I had 20 people that I had to compare to me in real time (because thats who is close by at the moment) it seems like it could take a while. What do you think?
If you use the compare contexts (shopping, working, traveling etc) calls, yes, it could take a long time. If I were going to do what you have proposed, I would store the Whit.li Vector Set and the specific interests from each user in your app. Then when someone logs in, if just get their vector set and interests and then you could do a very quick analysis of that data to the stored data. That is very close to what we did with the 1-1 app and one of the reasons for making the vector set numeric - so that it would be small to store and easy to compare in an app.
Say you have a hot geolocation mobile app that you are building and you wanted to use whit.li in as close to a near real time manner. For example when I open my app I want to see how I compare with those within a limited geographical range. If I have 20 users that I need to compare myself to and rank based on compatibility what kind of performance could I expect?
I notice the current demo app loads the list of people you are compatible with very quickly (I'm assuming it's cached) but when I actually click on a person to see more it takes > 20 seconds for the detail view. I'm assuming the api is doing a comparison right at that moment of you and the person you want to see more about. If I had 20 people that I had to compare to me in real time (because thats who is close by at the moment) it seems like it could take a while. What do you think?
Message edited by andig 1 year ago
Tags
andig – 1 year ago
If you use the compare contexts (shopping, working, traveling etc) calls, yes, it could take a long time. If I were going to do what you have proposed, I would store the Whit.li Vector Set and the specific interests from each user in your app. Then when someone logs in, if just get their vector set and interests and then you could do a very quick analysis of that data to the stored data. That is very close to what we did with the 1-1 app and one of the reasons for making the vector set numeric - so that it would be small to store and easy to compare in an app.
Chris Clarke – 1 year ago
Appreciate your feedback