giving n sentences, remove common phrases in each sentence, a phrase is defined by 3 or more consecutive words glassdoor: imo.im

glassdoor_removephrase.py

#!/usr/bin/python

def remove(buf):
    dict = {}
    wordslist = []
    for line in buf:
        words = line.split()
        wordslist.append(words)
        for i in xrange( len(words) - 2 ):
            phrase = ' '.join(words[i:i+3])
            try:
                dict[phrase] += 1
            except:
                dict[phrase] = 1
    ret = []
    for words in wordslist:
        flag = [True] * len(words)
        for i in xrange( len(words) - 2 ):
            phrase = ' '.join(words[i:i+3])
            if dict[phrase] == len(buf):
                flag[i] = flag[i+1] = flag[i+2] = False
        tmp = [words[i] for i in xrange( len(words) ) if flag[i]]
        ret.append(' '.join(tmp))
    return ret

def main():
    buf = ['i love this game', 'i dont love this game', 'he love this game', 'she love this game too']
    ret = remove(buf)
    print ret

if __name__ == '__main__':
    main()