Das deutsche Python-Forum

Code: Alles auswählen

def causal_mining (causal_relationship, encoding='utf-8'):
    main_filename = causal_relationship + '.txt'
    with codecs.open(main_filename, 'r') as main_file:
        sentences = sent_tokenize(main_file.read())

    print len(sentences)
    grammar_hypothesis = r'''Causal Relationship:{<RB><VBG><.*>+<MD><.*>+}
                                                 {<VBG|DT|IN|JJ|NN.*>+<.*>+<MD><.*>+}
                                                 {<NN.*><MD><.*>+}'''
                             #Sentence: {<.*>+}'''
    
    hp = nltk.RegexpParser(grammar_hypothesis) 

    cause = r'''Cause:{<VBG|DT|RB|JJ|NN.*>+<.*>+<MD>}
                      {<NN.*><MD>}
                      }<MD>{'''
    
    cp = nltk.RegexpParser(cause)
    
    impact = r'''Impact:{<MD><.*>+}
                        }<MD>{'''
    
    ip = nltk.RegexpParser(impact)
    
    writer = csv.writer(open("lern.csv", "wb"))
    
    for sentence in sentences:        
        sent = nltk.pos_tag(word_tokenize(sentence))
        tree = hp.parse(sent)
        string = ''.join(str(i) for i in tree)

        causal = re.compile('Causal Relationship')
        sign = re.compile('cause|affect|affected|enhance|higher|better|more|negatively|sooner|increase|perceived|Perceived|will|may')
        c = causal.match(string, 1, 20)
        s = sign.search(sentence)
        if c and s:
            y = ' Yes'
            print sentence + y
            for chunk in tree.subtrees():
                if chunk.node== 'Causal Relationship':
                    cau = cp.parse(sent)
                    for cause in cau.subtrees():
                        if cause.node== 'Cause': print cause
                        imp = ip.parse(sent)
                    for impact in imp.subtrees():
                        if impact.node== 'Impact': print impact

            positive = re.compile('cause|positively|affect|affected|enhance|higher|better|more|increase')
            negative = re.compile ('negatively|sooner')
            unclear = re.compile ('perceived|Perceived')
            p = positive.search(sentence)
            n = negative.search(sentence)
            u = unclear.search(sentence)
            
            if p:
                print 'positive'
            if n:
                print 'negative'
            if u:
                print 'unclear'

            a = [sentence, y]
            print a
            print '\n'

            #writer = csv.DictWriter(open("lernset.csv", "wb"),["Sentence", "Causal Mining"])
            #writer.writerow({"Sentence" : "Sentence", "Causal Mining" : "Causal Mining"})
            #daten = ({ 'Sentence' : sentence, 'Causal Mining' : y})
            
        else:
            n = ' No'
            print sentence + n
            print '\n'
        
    writer.writerow(['Sentence', 'Causal Mining'])
    daten = (a)
    writer.writerows(daten)

    #reader = csv.reader(open("lernset.csv", "rb"))
    #for row in reader:
        #print row