SentimentExtractor Example | Teradata Vantage - SentimentExtractor Example: Dictionary Table and Model File - Teradata Vantage

Input

• Input table: sentiment_extract_input, as in SentimentExtractor Example: InputModelFile ('dictionary'), AnalysisType ('document')
• Dict: sentiment_word, as in SentimentExtractor Example: Dictionary Table Instead of Model File
• InputModelFile: custom_sentiment_lexicon.txt, which looks like this:

  excellent       1
  disappointed    -1
  junk    -4
  mistake -4
  flutter 1
  fond    1
  fondly  1
  inspiring       1
  enjoy   3
  happy   4
  love    3
  nice    2
  hate    -3
  complaints      -5
  good    4
  poor    -1

SQL Call

If the SQL call uses both InputModelFile (default or custom) and a dictionary table, then the dictionary table overrides the InputModelFile when they conflict.

SELECT * FROM SentimentExtractor (
  ON sentiment_extract_input PARTITION BY ANY
  ON sentiment_word AS Dict DIMENSION
  USING
  InputModelFile ('dictionary: custom_sentiment_lexicon.txt')
  TextColumn ('review') 
  AnalysisType ('document')
  Accumulate ('id', 'product')
) AS dt ORDER BY ID;

Output

 id product      out_polarity out_strength out_sentiment_words                                                            
 -- ------------ ------------ ------------ ------------------------------------------------------------------------------ 
  1 camera       POS                     2 excellent 2, excellent 2. In total, positive score:4 negative score:0         
  2 office suite POS                     2 terrific 2, terrific 2. In total, positive score:4 negative score:0           
  3 camera       POS                     2 nice 1, good 4. In total, positive score:5 negative score:0                   
  4 gps          POS                     2 outstanding 2, incredible 2. In total, positive score:4 negative score:0      
  5 gps          POS                     2 nice 1, big 0, good 4. In total, positive score:5 negative score:0            
  6 gps          NEG                     2 complaints -5. In total, positive score:0 negative score:-5                   
  7 gps          POS                     2 messed 2. In total, positive score:2 negative score:0                        
  8 camera       NEG                     2 hate -3, stuck -1, difficulty -1. In total, positive score:0 negative score:-5
  9 television   NEG                     2 junk -2, poor -1, big 0. In total, positive score:0 negative score:-3         
 10 camera       NEG                     2 not tolerate -1, constant 0. In total, positive score:0 negative score:-1