Apache PredictionIO Template Ecommerce Recommender

Clone this repo:
  1. a7b8efb 0.12.0-incubating release by shimamoto · 3 months ago develop
  2. b755c31 Python 3 support by Shinsuke Sugaya · 8 months ago
  3. 6ef7c9b Fix warnings by Naoki Takezoe · 8 months ago
  4. 1d0d71d 0.11.0-incubating release by Donald Szeto · 9 months ago v0.11.0-incubating
  5. 93abf81 Bump to PIO 0.11.0-incubating and enhance sbt configuration by Naoki Takezoe · 9 months ago

E-Commerce Recommendation Template

Documentation

Please refer to http://predictionio.incubator.apache.org/templates/ecommercerecommendation/quickstart/

Versions

v0.12.0-incubating

  • Bump version number to track PredictionIO version
  • Sets default build targets according to PredictionIO
  • Fix warnings

v0.11.0-incubating

  • Bump version number to track PredictionIO version
  • Rename Scala package name
  • Update SBT version

v0.5.0

  • Update for Apache PredictionIO 0.10.0-incubating

v0.4.0

  • Change from ALSAlgorithm.scala to ECommAlgorithm.scala

    • return popular bought items when no information is found for the user.
    • add “similarEvents” parameter for configuration what user-to-item events are used for finding similar items
    • re-structure the Algorithm code for easier customization and testing
  • add some unit tests for testing code that may be customized

v0.3.1

  • use INVALID_APP_NAME as default appName in engine.json

v0.3.0

  • update for PredictionIO 0.9.2, including:

    • use new PEventStore and LEventStore API
    • use appName in DataSource and Algorithm parameters

v0.2.0

  • update build.sbt and template.json for PredictionIO 0.9.2

v0.1.1

  • update for PredictionIO 0.9.0

v0.1.0

  • initial version

Development Notes

import sample data

$ python data/import_eventserver.py --access_key <your_access_key>

query

normal:

$ curl -H "Content-Type: application/json" \
-d '{
  "user" : "u1",
  "num" : 10 }' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}
$ curl -H "Content-Type: application/json" \
-d '{
  "user" : "u1",
  "num": 10,
  "categories" : ["c4", "c3"]
}' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}
curl -H "Content-Type: application/json" \
-d '{
  "user" : "u1",
  "num": 10,
  "whiteList": ["i21", "i26", "i40"]
}' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}
curl -H "Content-Type: application/json" \
-d '{
  "user" : "u1",
  "num": 10,
  "blackList": ["i21", "i26", "i40"]
}' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}

unknown user:

curl -H "Content-Type: application/json" \
-d '{
  "user" : "unk1",
  "num": 10}' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}

handle new user

new user:

curl -H "Content-Type: application/json" \
-d '{
  "user" : "x1",
  "num": 10}' \
http://localhost:8000/queries.json \
-w %{time_connect}:%{time_starttransfer}:%{time_total}

import some view events and try to get recommendation for x1 again.

accessKey=<YOUR_ACCESS_KEY>
curl -i -X POST http://localhost:7070/events.json?accessKey=$accessKey \
-H "Content-Type: application/json" \
-d '{
  "event" : "view",
  "entityType" : "user"
  "entityId" : "x1",
  "targetEntityType" : "item",
  "targetEntityId" : "i2",
  "eventTime" : "2015-02-17T02:11:21.934Z"
}'

curl -i -X POST http://localhost:7070/events.json?accessKey=$accessKey \
-H "Content-Type: application/json" \
-d '{
  "event" : "view",
  "entityType" : "user"
  "entityId" : "x1",
  "targetEntityType" : "item",
  "targetEntityId" : "i3",
  "eventTime" : "2015-02-17T02:12:21.934Z"
}'

handle unavailable items

Set the following items as unavailable (need to specify complete list each time when this list is changed):

curl -i -X POST http://localhost:7070/events.json?accessKey=$accessKey \
-H "Content-Type: application/json" \
-d '{
  "event" : "$set",
  "entityType" : "constraint"
  "entityId" : "unavailableItems",
  "properties" : {
    "items": ["i43", "i20", "i37", "i3", "i4", "i5"],
  }
  "eventTime" : "2015-02-17T02:11:21.934Z"
}'

Set empty list when no more items unavailable:

curl -i -X POST http://localhost:7070/events.json?accessKey=$accessKey \
-H "Content-Type: application/json" \
-d '{
  "event" : "$set",
  "entityType" : "constraint"
  "entityId" : "unavailableItems",
  "properties" : {
    "items": [],
  }
  "eventTime" : "2015-02-18T02:11:21.934Z"
}'