Analyze-Mallet-results

Analyze-Mallet-results is an offline web application for visualization and analyze outputs of Mallet package.

Required software

• Python 2.7 interpreter
• cherrypy, with jinja2 templating
• matplotlib
• NumPy
• Pandas
• Pypi (python-bidi)
• Seaborn
• Sklearn
• Spyre library

Before using Mallet

Analyze-Mallet-results work with results of topic-model on corpus of filse that divided in to different categories, and the analyze based on topic-model results and the diffrent categories. the input data need to be in hierarchial path such as file of the properties "green" and "sweet" is in path like \path-to-input-files\green\sweet, or alternativly \path-to-input-files\sweet\green ׂׂ(Hierarchy does not matter).

Analyze-Mallet-results used output of Mallet package. The files Analyze-Mallet-results used is composition.txt and keys.text. these names is results of run the topic-model with the command

    $ bin \mallet train-topics your-own-parameters –optimaize-interval some-int –output-keys keys.txt –output-doc-topics composition.txt 

In addition, Analyze-Mallet-results use the files:

• size_of_files.txt files. this file is the results of the terminal command

    $ du –ch path\to\your\infput-for-Mallet-files > size_of_files.txt  

• labels_dic.txt Analyze-Mallet-results assume that the input for Mallet files are divided into different categories.