Implement a Keywords/Tags generator based on contents

[Vu Nguyen]

Hi, I'm newbie at  PredictionIO, I want to implement a prediction engine that can be extract keywords/tags from articles text content.

I considered the Text Classification Engine template and have working on it for a week but there is still a lot of things that i have to research about to make this template return the result like i expected.

Let assume the data input and output like below:

example input event:

{

"eventTime": ... , 

"entityId": 1,

"properties": {"text": "Article's text content, "keywords1": ["keyword1", "keyword2", "keyword3"] },

"event": "article", 

"entityType": "content"

}

example query:

{

  "text": "article content"

}

query result:

{

    top_10_keyword: [

   "keyword1",

   "keyword2",

   "keyword3", ...

  ]

}

So, i would very appreciate if any one give me some advises or instruction to custom the Text Classification Engine template (how to edit DataSource, which algorithm should i go with, ...)

Thanks very much

[Kenneth Chan]

this may also help

http://templates.prediction.io/EmergentOrder/template-scala-topic-model-LDA

[Vu Nguyen]

Thank Kenneth Chan, i will take a look.
For anyone may interest with this, I'm following this guide to understand then try to implement a Keywords Generator with PredictionIO http://bdewilde.github.io/blog/2014/09/23/intro-to-automatic-keyphrase-extraction/