Customized predictive algorithm

[hectoritr]

Each user has its own cache.
The dropdown should show the user customized model (Whatsapp)

[lopezjuanma96]

To Do List:

Client Side:

• add Dropdown for user specific models (we can add Whatsapp source, for example) @asimjawad OR in the same dropdown identify user specific models (couldbe same color as cache predictions)
• Identify cache predictions with a specific style. @asimjawad
• Add a parameter N to control how much of the prediction is based on cache, this can be, for example, every N model predictions shown one cache prediction is shown until cache is emptied. @asimjawad (see image Below)

Request Object:

• use UID from Firebase Authentication in userName key @asimjawad

Server Side:

• re-structure cache to have model-specific and general cache (maybe a cloud function can be runned weekly to find predictions present in all model-specific cache for one user and add them to the general one, that will appear for all caches) @lopezjuanma96
• use UID to get the user specific cache prediction @lopezjuanma96
• use UID to get the user specific model list (for Dropdown) @lopezjuanma96
• use UID to get the user specific model prediction @lopezjuanma96
• validate UID with Firebase Authentication on server? @lopezjuanma96

Response Object:

• Find what to do with words that appear both in the cache and model predictions @asimjawad @lopezjuanma96.

[lopezjuanma96]

An example of what we could do changing the parameter N for the influence of cache in the predictions