Machine Learning as a Service: making sentiment predictions in realtime with ZMQ and NLTK

15 min read Original article ↗

Transcript

  2. ABOUT ME

  3. DISSERTATION Let's make something cool!

  4. SOCIAL MEDIA + MACHINE LEARNING + API

  5. SENTIMENT ANALYSIS AS A SERVICE A STEP-BY-STEP GUIDE

  8. SUPERVISED MACHINE LEARNING SPAM OR HAM

  9. SUPERVISED MACHINE LEARNING SPAM OR HAM

  10. SUPERVISED MACHINE LEARNING SPAM OR HAM

  11. SUPERVISED MACHINE LEARNING SPAM OR HAM

  12. SUPERVISED MACHINE LEARNING SPAM OR HAM

  13. SUPERVISED MACHINE LEARNING SPAM OR HAM

  14. SUPERVISED MACHINE LEARNING SPAM OR HAM

  15. SUPERVISED MACHINE LEARNING SPAM OR HAM

  17. NATURAL LANGUAGE PROCESSING SOME NLTK FEATURES Tokentization Stopword Removal >

    > > p h r a s e = " I w i s h t o b u y s p e c i f i e d p r o d u c t s o r s e r v i c e " > > > p h r a s e = n l p . t o k e n i z e ( p h r a s e ) > > > p h r a s e [ ' I ' , ' w i s h ' , ' t o ' , ' b u y ' , ' s p e c i f i e d ' , ' p r o d u c t s ' , ' o r ' , ' s e r v i c e ' ] > > > p h r a s e = n l p . r e m o v e _ s t o p w o r d s ( t o k e n i z e d _ p h r a s e ) > > > p h r a s e [ ' I ' , ' w i s h ' , ' b u y ' , ' s p e c i f i e d ' , ' p r o d u c t s ' , ' s e r v i c e ' ]

  18. SENTIMENT ANALYSIS

  20. BACK TO BUILDING OUR API .. FINALLY!

  21. CLASSIFIER 3 STEPS

  23. FEATURE EXTRACTION How are we going to find features from

    a phrase? "Bag of Words" representation m y _ p h r a s e = " T o d a y w a s s u c h a r a i n y a n d h o r r i b l e d a y " I n [ 1 2 ] : f r o m n l t k i m p o r t w o r d _ t o k e n i z e I n [ 1 3 ] : w o r d _ t o k e n i z e ( m y _ p h r a s e ) O u t [ 1 3 ] : [ ' T o d a y ' , ' w a s ' , ' s u c h ' , ' a ' , ' r a i n y ' , ' a n d ' , ' h o r r i b l e ' , ' d a y ' ]

  24. FEATURE EXTRACTION CREATE A PIPELINE OF FEATURE EXTRACTORS F O

    R M A T T E R = f o r m a t t i n g . F o r m a t t e r P i p e l i n e ( f o r m a t t i n g . m a k e _ l o w e r c a s e , f o r m a t t i n g . s t r i p _ u r l s , f o r m a t t i n g . s t r i p _ h a s h t a g s , f o r m a t t i n g . s t r i p _ n a m e s , f o r m a t t i n g . r e m o v e _ r e p e t i t o n s , f o r m a t t i n g . r e p l a c e _ h t m l _ e n t i t i e s , f o r m a t t i n g . s t r i p _ n o n c h a r s , f u n c t o o l s . p a r t i a l ( f o r m a t t i n g . r e m o v e _ n o i s e , s t o p w o r d s = s t o p w o r d s . w o r d s ( ' e n g l i s h ' ) + [ ' r t ' ] ) , f u n c t o o l s . p a r t i a l ( f o r m a t t i n g . s t e m _ w o r d s , s t e m m e r = n l t k . s t e m . p o r t e r . P o r t e r S t e m m e r ( ) ) )

  25. FEATURE EXTRACTION PASS THE REPRESENTATION DOWN THE PIPELINE I n

    [ 1 1 ] : f e a t u r e _ e x t r a c t o r . e x t r a c t ( " T o d a y w a s s u c h a r a i n y a n d h o r r i b l e d a y " ) O u t [ 1 1 ] : { ' d a y ' : T r u e , ' h o r r i b l ' : T r u e , ' r a i n i ' : T r u e , ' t o d a y ' : T r u e } The result is a dictionary of variable length, containing keys as features and values as always True

  27. DIMENSIONALITY REDUCTION CHI-SQUARE TEST

  28. DIMENSIONALITY REDUCTION CHI-SQUARE TEST

  29. DIMENSIONALITY REDUCTION CHI-SQUARE TEST

  30. DIMENSIONALITY REDUCTION CHI-SQUARE TEST

  31. DIMENSIONALITY REDUCTION CHI-SQUARE TEST

  32. NLTK gives us BigramAssocMeasures.chi_sq DIMENSIONALITY REDUCTION CHI-SQUARE TEST # C

    a l c u l a t e t h e n u m b e r o f w o r d s f o r e a c h c l a s s p o s _ w o r d _ c o u n t = l a b e l _ w o r d _ f d [ ' p o s ' ] . N ( ) n e g _ w o r d _ c o u n t = l a b e l _ w o r d _ f d [ ' n e g ' ] . N ( ) t o t a l _ w o r d _ c o u n t = p o s _ w o r d _ c o u n t + n e g _ w o r d _ c o u n t # F o r e a c h w o r d a n d i t ' s t o t a l o c c u r a n c e f o r w o r d , f r e q i n w o r d _ f d . i t e r i t e m s ( ) : # C a l c u l a t e a s c o r e f o r t h e p o s i t i v e c l a s s p o s _ s c o r e = B i g r a m A s s o c M e a s u r e s . c h i _ s q ( l a b e l _ w o r d _ f d [ ' p o s ' ] [ w o r d ] , ( f r e q , p o s _ w o r d _ c o u n t ) , t o t a l _ w o r d _ c o u n t ) # C a l c u l a t e a s c o r e f o r t h e n e g a t i v e c l a s s n e g _ s c o r e = B i g r a m A s s o c M e a s u r e s . c h i _ s q ( l a b e l _ w o r d _ f d [ ' n e g ' ] [ w o r d ] , ( f r e q , n e g _ w o r d _ c o u n t ) , t o t a l _ w o r d _ c o u n t ) # T h e s u m o f t h e t w o w i l l g i v e y o u i t ' s t o t a l s c o r e w o r d _ s c o r e s [ w o r d ] = p o s _ s c o r e + n e g _ s c o r e

  33. TRAINING Now that we can extract features from text, we

    can train a classifier. The simplest and most flexible learning algorithm for text classification is Naive Bayes P ( l a b e l | f e a t u r e s ) = P ( l a b e l ) * P ( f e a t u r e s | l a b e l ) / P ( f e a t u r e s ) Simple to compute = fast Assumes feature indipendence = easy to update Supports multiclass = scalable

  34. TRAINING NLTK provides built-in components 1. Train the classifier 2.

    Serialize classifier for later use 3. Train once, use as much as you want > > > f r o m n l t k . c l a s s i f y i m p o r t N a i v e B a y e s C l a s s i f i e r > > > n b _ c l a s s i f i e r = N a i v e B a y e s C l a s s i f i e r . t r a i n ( t r a i n _ f e a t s ) . . . w a i t a l o t o f t i m e > > > n b _ c l a s s i f i e r . l a b e l s ( ) [ ' n e g ' , ' p o s ' ] > > > s e r i a l i z e r . d u m p ( n b _ c l a s s i f i e r , f i l e _ h a n d l e )

  35. USING THE CLASSIFIER # L o a d t h

    e c l a s s i f i e r f r o m t h e s e r i a l i z e d f i l e c l a s s i f i e r = p i c k l e . l o a d s ( c l a s s i f i e r _ f i l e . r e a d ( ) ) # P i c k a n e w p h r a s e n e w _ p h r a s e = " A t P y c o n I t a l y ! L o v e t h e f o o d a n d t h i s s p e a k e r i s s o a m a z i n g " # 1 ) P r e p r o c e s s i n g f e a t u r e _ v e c t o r = f e a t u r e _ e x t r a c t o r . e x t r a c t ( p h r a s e ) # 2 ) D i m e n s i o n a l i t y r e d u c t i o n , b e s t _ f e a t u r e s i s o u r s e t o f b e s t w o r d s r e d u c e d _ f e a t u r e _ v e c t o r = r e d u c e _ f e a t u r e s ( f e a t u r e _ v e c t o r , b e s t _ f e a t u r e s ) # 3 ) C l a s s i f y ! p r i n t s e l f . c l a s s i f i e r . c l a s s i f y ( r e d u c e d _ f e a t u r e _ v e c t o r ) > > > " p o s "

  37. BUILDING A CLASSIFICATION API SCALING TOWARDS INFINITY AND BEYOND

  39. BUILDING A CLASSIFICATION API ZEROMQ . . . s o

    c k e t = c o n t e x t . s o c k e t ( z m q . R E P ) . . . w h i l e T r u e : m e s s a g e = s o c k e t . r e c v ( ) p h r a s e = j s o n . l o a d s ( m e s s a g e ) [ " t e x t " ] # 1 ) F e a t u r e e x t r a c t i o n f e a t u r e _ v e c t o r = f e a t u r e _ e x t r a c t o r . e x t r a c t ( p h r a s e ) # 2 ) D i m e n s i o n a l i t y r e d u c t i o n , b e s t _ f e a t u r e s i s o u r s e t o f b e s t w o r d s r e d u c e d _ f e a t u r e _ v e c t o r = r e d u c e _ f e a t u r e s ( f e a t u r e _ v e c t o r , b e s t _ f e a t u r e s ) # 3 ) C l a s s i f y ! r e s u l t = c l a s s i f i e r . c l a s s i f y ( r e d u c e d _ f e a t u r e _ v e c t o r ) s o c k e t . s e n d ( j s o n . d u m p s ( r e s u l t ) )

  40. DEMO

  42. FIN QUESTIONS