Skip to content

Files

Latest commit

jaroslav-kucharjaroslav-kuchar
broken link fix
Jun 14, 2015
e7cd7db · Jun 14, 2015

History

History
156 lines (130 loc) · 8.24 KB

tutorial-external.md

File metadata and controls

156 lines (130 loc) · 8.24 KB

InBeat - external recommender system

InBeat offers by itself only association rules as the preference learning/recommendation algorithm. However, InBeat versatile export functionality allows for easy connection to other recommender toolboxes, such as MyMediaLite, which is specifically covered in this tutorial.

What this tutorial covers

Export formats

InBeat support three formats for exports: JSON, CSV and MySQL DDL script. The following table shows example data returned by the /export/interests call, which returns aggregated data. Example result in a table format is listed below.

accountId objectId userId sessionId type_Root type_Food type_Electronics type_Televisions type_Radios entity_Television entity_Onion entity_Salt interest
INBEAT-TUTORIAL http://example.com/objects/object1 http://example.com/users/user3 1433937669616 1 0 1 1 0 1 0 0 1
INBEAT-TUTORIAL http://example.com/objects/object2 http://example.com/users/user2 1433937683920 1 1 0 0 0 0 1 1 -1
INBEAT-TUTORIAL http://example.com/objects/object1 http://example.com/users/user2 1433937666223 1 0 1 1 0 1 0 0 0.5
INBEAT-TUTORIAL http://example.com/objects/object1 http://example.com/users/user1 1433937683920 1 0 1 1 0 1 0 0 1

Use the web admin console http://localhost:8880/admin/#/gain-export or cURL

Example of JSON

# Get export
curl --user "INBEAT-TUTORIAL:INBEAT-TUTORIAL" --header "Accept: application/json" "http://localhost:8880/gain/api/INBEAT-TUTORIAL/export/interests" > export.json
[
    {
        "accountId": "INBEAT-TUTORIAL",
        "objectId": "http://example.com/objects/object1",
        "parentObjectId": "",
        "sessionId": "1433937669616",
        "userId": "http://example.com/users/user3",
        "interest": 1,
        "type_Root": 1,
        "type_Electronics": 1,
        "type_Televisions": 1,
        "entity_Television": 1,
        "last": "1433937683920"
    },
    {
        "accountId": "INBEAT-TUTORIAL",
        "objectId": "http://example.com/objects/object2",
        "parentObjectId": "",
        "sessionId": "1433937683920",
        "userId": "http://example.com/users/user2",
        "interest": -1,
        "type_Root": 1,
        "type_Food": 1,
        "entity_Onion": 1,
        "entity_Salt": 1,
        "last": "1433937694713"
    },
    {
        "accountId": "INBEAT-TUTORIAL",
        "objectId": "http://example.com/objects/object1",
        "parentObjectId": "",
        "sessionId": "1433937666223",
        "userId": "http://example.com/users/user2",
        "interest": 0.5,
        "type_Root": 1,
        "type_Electronics": 1,
        "type_Televisions": 1,
        "entity_Television": 1,
        "last": "1433937666242"
    },
    {
        "accountId": "INBEAT-TUTORIAL",
        "objectId": "http://example.com/objects/object1",
        "parentObjectId": "",
        "sessionId": "1433937683920",
        "userId": "http://example.com/users/user1",
        "interest": 1,
        "type_Root": 1,
        "type_Electronics": 1,
        "type_Televisions": 1,
        "entity_Television": 1,
        "last": "1433937680581"
    }
]

Example of CSV

# Get export
curl --user "INBEAT-TUTORIAL:INBEAT-TUTORIAL" --header "Accept: text/csv" "http://localhost:8880/gain/api/INBEAT-TUTORIAL/export/interests" > export.json
accountId;objectId;parentObjectId;sessionId;userId;interest;type_Root;type_Electronics;type_Televisions;entity_Television;last;type_Food;entity_Onion;entity_Salt
INBEAT-TUTORIAL;http://example.com/objects/object1;0;1433937669616;http://example.com/users/user3;1;1;1;1;1;1433937683920;0;0;0
INBEAT-TUTORIAL;http://example.com/objects/object2;0;1433937666223;http://example.com/users/user2;-1;1;0;0;0;1433937694713;1;1;1
INBEAT-TUTORIAL;http://example.com/objects/object1;0;1433937666223;http://example.com/users/user2;0.5;1;1;1;1;1433937666242;0;0;0
INBEAT-TUTORIAL;http://example.com/objects/object1;0;1433937660867;http://example.com/users/user1;1;1;1;1;1;1433937680581;0;0;0

Example of MySQL DDL export result

# Get export
curl --user "INBEAT-TUTORIAL:INBEAT-TUTORIAL" --header "Accept: text/plain" "http://localhost:8880/gain/api/INBEAT-TUTORIAL/export/interests" > export.json
DROP TABLE IF EXISTS `INBEAT-TUTORIAL-undefined`; 
CREATE TABLE `INBEAT-TUTORIAL-undefined` (Id INT NOT NULL AUTO_INCREMENT PRIMARY KEY,`accountId` TEXT , `userId` TEXT, `sessionId` TEXT, `objectId` TEXT, `parentObjectId` TEXT, `interest` FLOAT default 0, `type_Root` FLOAT default 0, `type_Electronics` FLOAT default 0, `type_Televisions` FLOAT default 0, `entity_Television` FLOAT default 0, `last` FLOAT default 0, `type_Food` FLOAT default 0, `entity_Onion` FLOAT default 0, `entity_Salt` FLOAT default 0);
INSERT INTO `INBEAT-TUTORIAL-undefined`(Id, `accountId`, `objectId`, `parentObjectId`, `sessionId`, `userId`, `interest`, `type_Root`, `type_Electronics`, `type_Televisions`, `entity_Television`, `last` ) VALUES (0,"INBEAT-TUTORIAL", "http://example.com/objects/object1", "", "1433937669616", "http://example.com/users/user3", "1", "1", "1", "1", "1", "1433937683920");
INSERT INTO `INBEAT-TUTORIAL-undefined`(Id, `accountId`, `objectId`, `parentObjectId`, `sessionId`, `userId`, `interest`, `type_Root`, `type_Food`, `entity_Onion`, `entity_Salt`, `last` ) VALUES (0,"INBEAT-TUTORIAL", "http://example.com/objects/object2", "", "1433937666223", "http://example.com/users/user2", "-1", "1", "1", "1", "1", "1433937694713");
INSERT INTO `INBEAT-TUTORIAL-undefined`(Id, `accountId`, `objectId`, `parentObjectId`, `sessionId`, `userId`, `interest`, `type_Root`, `type_Electronics`, `type_Televisions`, `entity_Television`, `last` ) VALUES (0,"INBEAT-TUTORIAL", "http://example.com/objects/object1", "", "1433937666223", "http://example.com/users/user2", "0.5", "1", "1", "1", "1", "1433937666242");
INSERT INTO `INBEAT-TUTORIAL-undefined`(Id, `accountId`, `objectId`, `parentObjectId`, `sessionId`, `userId`, `interest`, `type_Root`, `type_Electronics`, `type_Televisions`, `entity_Television`, `last` ) VALUES (0,"INBEAT-TUTORIAL", "http://example.com/objects/object1", "", "1433937660867", "http://example.com/users/user1", "1", "1", "1", "1", "1", "1433937680581");

Example of usage in MyMediaLite

This section shows how InBeat can be used in conjunction with MyMediaLite.

For the item prediction task, the CSV output of InBeat should be transformed to the supported format (e.g positive only feedback file). The minimal requirement is to provide two columns: userId and objectId.

For transformation can be performed with the following simple python script

import pandas
data = pandas.read_csv("inbeat.csv", sep=";")
data.to_csv("mymedialite.csv",columns=["userId","objectId"],index=False,header=False)

The output of transformation is as follows:

http://example.com/users/user3,http://example.com/objects/object1
http://example.com/users/user2,http://example.com/objects/object2
http://example.com/users/user2,http://example.com/objects/object1
http://example.com/users/user1,http://example.com/objects/object1

Example of item recommendation using UserKNN algorithm in MyMediaLite:

./item_recommendation --training-file=mymedialite.csv --recommender=UserKNN --prediction-file=prediction.csv

The output is an Item recommendation file

http://example.com/users/user3	[http://example.com/objects/object2:0.4142136]
http://example.com/users/user2	[]
http://example.com/users/user1	[http://example.com/objects/object2:0.4142136]

Explanation:

  • for ...user3 and ...user1, ...object2 is recommended, because other users interacted with this object.
  • there is no recommendation for ...user2, since the user already interacted with all available objects.