model_establish.txt

Techniques to use to establish baseline
--------------------------------------------------------------------------------
1. Word Frequency Analysis
    ~> TFIDF

2. Frequency based Single Document Keywork extraction

3. Content sensitive Single Document Keywork Extraction
    It matches and surpasses TF-IDF

Lexical Graph

Bayes Classifier with TF-IDF

--------------------------------------------------------------------------------
Method described by Xiao-yu Jiang, in A Keyword Extraction Method Based on 
Lexical Chains
@ http://download.springer.com/static/pdf/524/chp%253A10.1007%252F978-3-540-92137-0_40.pdf?auth66=1414260742_a37eb16bbe53dea943a4202402153af7&ext=.pdf

--> He has used it mainly for Chinese characters , where there is no boundary 
between two characters . 

Precision = |A ∩ B| / |A|
Recall = |A ∩ B| / |B|
where  |A| is the number of keywords extracted automatically,
|B| is number of keyworkds present in test data,
|A ∩ B| is the size of intersection between A and B 

--------------------------------------------------------------------------------