Please enable JavaScript.

Coggle requires JavaScript to display documents.

Data Structure Search (Theme Modeling (Process (Data Preparation…

- - - - Preliminary cleaning
        
        Remove Formatting and Line Breaks
        
        Remove not Textual Information
        
        Fix Typos
        
        Merge too Short Texts
      - Form a vocabulary
        
        Normalize Words - stemming
        
        Extract Terms
        
        Remove stop words and too rare ones
    - - Basic Assumptions
        
        Term order in the document is not important
        
        Each pair (d,w) belongs to at least one theme t from T
        
        Hypothesis of conditional independence: p(w|t,d)=p(w,t)
        
        Word Order is not important
      - Sparseness Assumptions
        
        Document belongs to a little number of themes
        
        Theme usually contains small number of terms
    - - A Document d is a mix of distributions p(w|t) with weights p(t|d): p(w|t) = SUM(p(w|t)p(t|d))_t_from_T