Menu

Help!!!!!!!!!!

LiQiang
2016-12-01
2016-12-20
  • LiQiang

    LiQiang - 2016-12-01

    Rules are automatically computed by the ruleminer on a daily basis as soon significant user patterns are identified. But why in my system no rules are automatically generated between two items.Thanks

     
  • Stephan Zavrel

    Stephan Zavrel - 2016-12-01

    There are a lot of possible explanations for why no rules are generated.First check the Admin Tool an go to the Administration section (link in the upper right corner). There go to plugin logs and check if the rulminer is actually executed. (There should be log entries). If you see entries there you can paste some log output here.

    If the Rulminer is running fine and you still see no recommendations it can be that there simply are no significant similarities in user behaviour to find any recommendations. You need at least 2 different users performing the same actions on at least the same 3 items for a third user to see recommendations:

    1. User A views Item 1
    2. User A views Item 2
    3. User A views Item 3
    4. User B views Item 1
    5. User B views Item 2
    6. User B views Item 3
    7. Calculate rules
    8. User C views Item 1 and gets Item2, Item3 as recommendations.

    That is the bare minimum for the user to see any recommendations. If the data you sent so far does not show at least this pattern anywhere, you will not get any results!

    Given your other thread it may also be a problem with the sessionids. If you set them wrong (always the same one or always a different one), rules will also not be generated!

    Hth
    Stephan

     
  • LiQiang

    LiQiang - 2016-12-02

    EASYREC_TEST
    Overview • Items & Rules • Statistics • Top Ranked • Hot Recommendations • Cluster Manager • Type Manager

    Statistics for "EASYREC_TEST"

    The following tables show the activities for the selected tenant.
    Note: The data presented here is automatically updated after every rule generator execution normally scheduled once a day, so it may be a little outdated.
    If you want to, you can refresh the statistics now.
    Overview
    number of clicks on recommendations (conversions) 0 (0) help
    number of total actions (this month) 127 (68) help
    number of maximum allowed actions per month unlimited help
    number of total items 43 help
    number of total users 7 help
    average actions per user 18.14 help
    recommendation coverage 70.87 % help

    Number of Users who made
    1 action 1
    2 actions 0
    3 - 10 actions 3
    11 - 100 actions 3
    101 and more actions 0

     
  • LiQiang

    LiQiang - 2016-12-02

    EASYREC_TEST
    Overview • Items & Rules • Statistics • Top Ranked • Hot Recommendations • Cluster Manager • Type Manager

    Statistics for "EASYREC_TEST"

    The following tables show the activities for the selected tenant.
    Note: The data presented here is automatically updated after every rule generator execution normally scheduled once a day, so it may be a little outdated.
    If you want to, you can refresh the statistics now.
    Overview
    number of clicks on recommendations (conversions) 0 (0) help
    number of total actions (this month) 127 (68) help
    number of maximum allowed actions per month unlimited help
    number of total items 43 help
    number of total users 7 help
    average actions per user 18.14 help
    recommendation coverage 70.87 % help

    Number of Users who made
    1 action 1
    2 actions 0
    3 - 10 actions 3
    11 - 100 actions 3
    101 and more actions 0

    14%
    43%
    43%
    
    1 action
    2 actions
    3-10
    11-100
    > 100 actions
    

    Rule Statistics

    is related (provided by SlopeOne) help
    endDate 2016-11-28T10:44:08.573+08:00
    startDate 2016-11-28T10:44:08.551+08:00
    numberOfActionsConsidered 0
    numberOfRulesCreated 0
    actionDuration 0
    deviationDuration 3
    noCreatedDeviations 0
    noModifiedDeviations 0
    noUsers 0
    nonPersonalizedDuration 0

    good rated together (provided by ARM) help
    endDate 2016-11-28T10:44:08.510+08:00
    startDate 2016-11-28T10:44:09+08:00
    numberOfActionsConsidered 0
    numberOfRulesCreated 0
    cutoffDate 2015-11-29T10:44:09+08:00
    duration 0
    iterations 1
    lastConf 0.0
    lastSupport 2
    metricType CONFIDENCE
    nrBaskets 0
    nrProducts 0
    sizeCountMap 0
    sizeL1 0
    sizeL2 0
    sizeRules 0

    viewed together (provided by ARM) help
    endDate 2016-12-02T15:00:00.289+08:00
    startDate 2016-12-02T15:00:00+08:00
    numberOfActionsConsidered 126
    numberOfRulesCreated 104
    cutoffDate 2015-12-03T15:00:00+08:00
    duration 0
    iterations 1
    lastConf 0.0
    lastSupport 2
    metricType CONFIDENCE
    nrBaskets 7
    nrProducts 43
    sizeCountMap 0
    sizeL1 23
    sizeL2 52
    sizeRules 104

    bought together (provided by ARM) help
    endDate 2016-11-28T10:44:08.532+08:00
    startDate 2016-11-28T10:44:09+08:00
    numberOfActionsConsidered 0
    numberOfRulesCreated 0
    cutoffDate 2015-11-29T10:44:09+08:00
    duration 0
    iterations 1
    lastConf 0.0
    lastSupport 2
    metricType CONFIDENCE
    nrBaskets 0
    nrProducts 0
    sizeCountMap 0
    sizeL1 0
    sizeL2 0
    sizeRules 0

    distribution of association values
    number of items with rules that have a association value:
    < 20 % 0
    20 - 40 % 5
    40 - 60 % 2
    60 - 80 % 4

    80 % 4

     

    Last edit: LiQiang 2016-12-02
  • LiQiang

    LiQiang - 2016-12-02

    I have done as you say(Many users view many different items with different sessionids).Rules are generated successfully.But still no recommendations are given for users......

     
  • LiQiang

    LiQiang - 2016-12-02

    Here are my sql file in my database. Please help me analyze the reasons when you have time .Thank you very much.

     
  • Florian Kleedorfer

    If the rules are generated (and your sql file shows they are), they are used to compute recommendations. However, a recommendation list gets filtered before it is sent to the client, based on previous actions of the particular user. The idea is to avoid showing the same items to the same user over and over again. It is possible that you now have created actions for all users on all items, and so all items are filtered out.

    To check if that is indeed the problem, try adding an item and view it only with a subset of the users, then run the generators and see if you get recommendations for a different user.

     
  • LiQiang

    LiQiang - 2016-12-02

    {yourServerURL}/api/1.1/otherusersalsoviewed?apikey=8ab9dc3ffcdac576d0f298043a60517a&tenantid=EASYREC_DEMO&itemid=42&userid=24EH1723322222A3&itemtype=ITEM&requesteditemtype=ITEM. When I send call like this. I receive only one item which is the same itemid with the one in the url in a xml file.No recommendation.why? thanks!

     
  • LiQiang

    LiQiang - 2016-12-02

    ’try adding an item and view it only with a subset of the users‘ means delete all the items I added before?

     
  • LiQiang

    LiQiang - 2016-12-02

    ’try adding an item and view it only with a subset of the users‘ means delete all the items I added before?

     
  • LiQiang

    LiQiang - 2016-12-02

    ’try adding an item and view it only with a subset of the users‘ means delete all the items I added before?

     
  • LiQiang

    LiQiang - 2016-12-05

    Thank you very much. Recommendation items for users is not null now.

     
  • LiQiang

    LiQiang - 2016-12-14

    Does the easyrec has functions like Statistics collection sites PV, UV, user behavior data, such as: the length of the user access(how long does a user view one page), user access depth (how many pages) and other data

     
  • Florian Kleedorfer

    Unfortunately, there are no statistics to get from easyrec except for what you see in the admin interface.

     
  • LiQiang

    LiQiang - 2016-12-20

    I have found some useful data about users are existing in the database ,but they have not show us in the admin interface.I will develop it later.

     

Log in to post a comment.

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.