search

shahinmv / warehouse-matching

Authentication and login system with a smart search and query engine in the back-end tier. Simple front-end layer for showing the suitable warehouse options.
Other
0 stars 0 forks source link

RBM Recommending Option 1 #5

Open serkosi opened 2 years ago

serkosi commented 2 years ago

https://github.com/shahinmv/thesis-app/blob/4faeaa144cfe58b86ab147787ea397da7864c4ac/rbm.py#L1

After the user completes an action about Warehouse X, the SSRS might bring up a recommendation as below by selecting one of the given a/b options:

Other users who [a- gave a score, b- worked with] the [a- service, b- region] of Warehouse X, also [a- gave high score, b- worked with] those warehouses listed here.

shahinmv commented 2 years ago

This option, in some way can be used to learn from other users data and get the best results to show to the user. Lets say user inputs the services they require, RBM will learn from other users who also booked warehouse with same service, mainly will use the scores. Then the results will be displayed.

serkosi commented 2 years ago

image

I closed the other issue (RBM Recommending Option 2) and continue from this issue to our discussion.

In my opinion, we need to use user scores/clicks to identify the features that those warehouses have in common. The features are the ones which merchants like, not warehouses. In our understanding, those preferred features might be different things which are given as below. I listed them just as a starting point, we can modify them later on.

From the scores given in the sketch, people scored the couple, warehouse-3 and ware-house-4, similarly to warehouse-6. That shows there is a common feature between them. RBM will train the model by checking each hidden note to see if any of them reflects this pattern in the scores. Lets discuss those above in our coming meeting today with your ideas. Then we can discuss further on the testing part of the ML.