[MGNLPER-7] Research how to store a network model (trained user preferences) Created: 19/Feb/18  Updated: 06/Jul/18  Resolved: 06/Jul/18

Status: Closed
Project: Periscope
Component/s: None
Affects Version/s: None
Fix Version/s: 1.0

Type: New Feature Priority: Neutral
Reporter: Ilgun Ilgun Assignee: Cedric Reichenbach
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Relates
relates to MGNLPER-33 Store a network model Closed
Template:
Acceptance criteria:
Empty
Date of First Response:
Epic Link: Periscope back-end MVP
Sprint: Basel 150
Story Points: 5

 Description   

Currently, we have no way to store trained neural network within magnolia or file system. However, it is essentially to store it somewhere to that users' preferences will not be lost when instance has restarted.

User story:

As a user, I want Magnolia to provide increasingly relevant search results to me. I typically search for the same things. Remember what I searched in the past. Remember it across sessions.

Acceptance criteria:

  • Search results improve over time.
  • Search results are user specific.
  • Search results are persistent.
  • Decide where to store the trained model per user.
  • Decide how to store the trained model.
  • Performance criteria: the user should not notice that we load the model. Find a convenient time to load the model so the user doesn't perceive slowness when they try to search.


 Comments   
Comment by Cedric Reichenbach [ 06/Jul/18 ]

Solution:
UseĀ ModelSerializer#writeModel to serialize our network to an OutputStream, then pipe data into binary JCR node (using JcrUtils#putFile).

As soon as we have per-user result ranking, we may store it in the profiles workspace, under the corresponding user node.

Generated at Mon Feb 12 10:27:48 CET 2024 using Jira 9.4.2#940002-sha1:46d1a51de284217efdcb32434eab47a99af2938b.