In [1]:

    

from htmresearch.encoders.cio_encoder import CioEncoder


    

In [2]:

    

%matplotlib inline


    

In [3]:

    

import matplotlib.pyplot as plt


    

In [4]:

    

import matplotlib.image as mpimg


    

In [5]:

    

import numpy


    

In [6]:

    

encoder = CioEncoder(retina="en_synonymous")


    

In [7]:

    

def plotText(text):
    fingerprint = encoder.encode(text)
    img = numpy.zeros(128*128)
    img[list(fingerprint['fingerprint']['positions'])]= 1
    img = img.reshape((128,128))
    plt.imshow(img, cmap="spectral")
    print "Sparsity=",fingerprint['sparsity'],"bits=",len(fingerprint['fingerprint']['positions'])


    

In [8]:

    

def compareText(text1, text2):
    f1 = encoder.encode(text1)['fingerprint']['positions']
    f2 = encoder.encode(text2)['fingerprint']['positions']
    print "Bits for '"+text1+"' = ",len(f1)
    print "Bits for '"+text2+"' = ",len(f2)
    overlap = len(set(f1) & set(f2))
    print "Overlap=",overlap
    return overlap


    

In [9]:

    

plotText("James quit smoking but it was not an easy decision.")


    

    




Sparsity= 0.0362548828125 bits= 594






    

In [10]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "James stopped smoking but it was not easy decision")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'James stopped smoking but it was not easy decision' =  594
Overlap= 594






    
Out[10]:





594

In [11]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "James stopped smoking but it was not a simple decision")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'James stopped smoking but it was not a simple decision' =  602
Overlap= 501






    
Out[11]:





501

In [12]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "James stopped smoking but it was not a simple choice")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'James stopped smoking but it was not a simple choice' =  570
Overlap= 392






    
Out[12]:





392

In [13]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "Sue stopped smoking but it was not a simple choice")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'Sue stopped smoking but it was not a simple choice' =  559
Overlap= 294






    
Out[13]:





294

In [14]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "Sue stopped cigarettes but it was not a simple choice")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'Sue stopped cigarettes but it was not a simple choice' =  693
Overlap= 237






    
Out[14]:





237

In [15]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "We start with ten base sentences each with ten words")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'We start with ten base sentences each with ten words' =  298
Overlap= 65






    
Out[15]:





65

In [16]:

    

compareText("James quit smoking but it was not an easy decision.", 
            "Sue keeps smoking a lot of cigarettes")


    

    




Bits for 'James quit smoking but it was not an easy decision.' =  594
Bits for 'Sue keeps smoking a lot of cigarettes' =  549
Overlap= 263






    
Out[16]:





263

In [17]:

    

plotText("the ski lift of projectors requires random resorts vacation chairs kids")


    

    




Sparsity= 0.063720703125 bits= 1044






    

In [18]:

    

plotText("biology")


    

    




Sparsity= 0.010009765625 bits= 164






    

In [19]:

    

compareText("mountains vacation","skiing")


    

    




Bits for 'mountains vacation' =  318
Bits for 'skiing' =  164
Overlap= 24






    
Out[19]:





24

In [20]:

    

compareText("James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.", 
           "The tobacco industry has long tried to hide the effects of second hand smoke. However there is now a preponderance of evidence demonstrating its ill effects. Second hand smoke is clearly dangerous to anyone who breathes it in.")


    

    




Bits for 'James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.' =  1197
Bits for 'The tobacco industry has long tried to hide the effects of second hand smoke. However there is now a preponderance of evidence demonstrating its ill effects. Second hand smoke is clearly dangerous to anyone who breathes it in.' =  1259
Overlap= 495






    
Out[20]:





495

In [21]:

    

compareText("James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.", 
           "the dangers of cigars. a cigar is basically a cancer stick. Cigars are known to contain over 7000 chemicals, at least 250 of which are directly tied to lung cancer.")


    

    




Bits for 'James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.' =  1197
Bits for 'the dangers of cigars. a cigar is basically a cancer stick. Cigars are known to contain over 7000 chemicals, at least 250 of which are directly tied to lung cancer.' =  985
Overlap= 337






    
Out[21]:





337

In [22]:

    

compareText("James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.", 
           "people with health problems are turning to jogging as a solution. It raises heart rate and causes a rush of endorphins to the body. It also makes people more resilient to disease. It's the ultimate preventative medicine for health issues.")


    

    




Bits for 'James loved to puff on his cigarettes. However he recently read a bunch of articles describing their unhealthy effects. James decided to quit cigarettes completely. It clearly was not an easy decision, but he felt it was the right one.' =  1197
Bits for 'people with health problems are turning to jogging as a solution. It raises heart rate and causes a rush of endorphins to the body. It also makes people more resilient to disease. It's the ultimate preventative medicine for health issues.' =  1362
Overlap= 392






    
Out[22]:





392

In [23]:

    

compareText("James","tobacco")


    

    




Bits for 'James' =  164
Bits for 'tobacco' =  164
Overlap= 10






    
Out[23]:





10

In [24]:

    

plotText("advsh")


    

    




Sparsity= 0.00994873046875 bits= 163






    

In [ ]: