hard queries... local queries … people are not looking for websites but actually businesses on a map

[實測]
Japanese toys in san francisco

Buy live lobster in kissimmee fl

Bee hive removal pasadena md

Vegan donuts near me

We could write rules for each of these, but it becomes unwieldy rather quickly.
--> We start by thinking about how to collect the data to make it a ML problem.

Search coffee near me 在圖中 出現了兩個選項
選項 1 : Bill's Dinner
選項 2 : anna's Gourmet Cafe
要選擇哪個呢 ?
距離較近的 Bill's Dinner 還是較遠的 anna's Gourmet Cafe

如果 anna's Gourmet Cafe 需要 經過橋梁，那就需要花更多時間啦 => prefer Bill's Dinner
那如果 Bill's Dinner 需要 10分鐘 才可以上coffee 且 沒有得來速(也就是說 你得坐那兒)，這時候 anna's Gourmet Cafe 感覺較妥當。

那到底 哪個會是對於消費者 最好的選擇呢?
別再猜啦，直接問問看使用者!
EX : 手機GPS 定位你在某個景點或是餐廳附近，他會跳出通知，詢問評價，如果有回饋，這時候就有資料啦。

So we look at a bunch of data and do a trade off … distance vs. quality of coffee …
還是有其他因素可以考慮進去: service time ，但在這個 example，只考慮 distance 和 quality of coffee。

As an AI-first company, we might start with heuristics, but we do so with the mindset
that we are going to throw away the heuristics just as soon as we have enough data
about user preferences ...

一開始的策略(方法)為 heuristics，但資料(user preferences)足夠多的時候，我們就可以拋棄 heuristics
example == labeled
input : distance to shop
label : does the user like the result or not?

[運作]

So we take an example of a shop 1 km away and user says great …I’ll go 1km
And we ask another user 3km away and they say “ I don’t even like gourmet
coffee.”
We aggregate a bunch of different examples eventually when we realize when
it’s far enough away, nobody wants to go …

then fit the model

Machine Learning is about collecting the appropriate data and then finding the right
balance of good learning, trusting the examples