marble‐processor‐sklearn - miguelfc/marble GitHub Wiki

Type

Processor.

Function

This processor assigns a polarity value to each message related to a specific topic.

To calculate this value, it uses the python sklearn and the algorithm developed by Marco Bonzanini, based on tf-idf weighting and different variations of Support Vector Machine. The options are used to specify the type of classifier to use.

Usage

Requirements

The corpora are downloaded from the internet at startup time, so you would need access to the internet from the sklearn processor instance.

Options

type: Defines the type of SVM classifier to use. Options available:

• svc_rbf
• svc_linear
• svc_liblinear

Example Recipe

{
  "name": "processor-sklearn",
  "options": {
    "type": "svc_rbf"
  }
}