Investigate ways to integrate third-party services

[brampitoyo]

@antlam wrote:

[…] one idea that came to mind is integration with services like Foursquare, Swarm, Yelp, and even Facebook check-ins. Not sure what this might look like in code, but allowing users to give us this specific information could help us greatly with our own recommendations.

[brampitoyo]

There are a few signals from third-party services that we may be able to use:

Tags

If you search or check in at places with specific tags, Prox can assume that you like the same type of place, and will prioritise those places when the app does a search

By the same token, if our goal is to surprise you, the app can purposefully avoid showing that type of place in order to show you new types and categories

Check-in time

If you search or check in at specific time, Prox will know when you’d like to go somewhere

• For example, some users like to exercise (signal: go to a park, check in at the gym) first thing in the morning, while others like to do it in the evening.
• Some users like to eat lunch out, while others like to eat dinner out, and maybe others bring their lunch and go to the park.

Prox can be sensitive to this. If you like to run in the evening, Prox will show you all the parks nearby in the evening, not in the morning.

Check-in frequency from other users

We can aggregate check in information from multiple services to determine whether a place is currently busy or has been unseasonably empty, and thus find ”hidden gems“.

• For example, if a restaurant is highly rated but really busy when we do a search. We’ll hide it. Waiting in line isn’t our idea of giving you a great day.
• On the other hand, if a restaurant is usually full of check ins but happen to be empty at the time, we’ll show it high on our list. This is a place you probably want to go to.

Check-in place order

People don’t move between places randomly. They always have intentions, and Prox can utilise this information.

• Most third party services have a mapping function. If you use it to navigate between places, it knows your movement pattern. Let’s say that, after visiting place A, 50% of people go to place B and another 50% go to place C.
• For example, if you’re moving between many bars on the same street, or even multiple places tagged #bars in the same city over the period of an evening, we know that you’re probably bar hopping.

Focusing again on finding “hidden gems”, if our user is visiting place A, we can then search for places in the same categories as B and C, then recommend that to visit instead of the more popular paths.

Alternatively, we can straight up recommend place B and C and say “Most people go here after visiting place A. So why not try it?”