def ma_overlap_zi(row):
question = cut(row["question"])
answer = cut(row["answer"])
di_question = []
di_answer = []
for w in question:
for i in range(len(w) ):
di_question.append(w[i])
for w in answer:
for i in range(len(w) ):
di_answer.append(w[i])
di_overlap = set(di_question).intersection(set(di_answer) )
di_weight_p = dict({})
for k in range(len(di_question) ):
if di_question[k] in di_overlap:
# print int(100*((k+1)/(len(question)+1)) )
di_weight_p[di_question[k] ] =((k+1)/len(di_question))**3.2# zi_weight[ int(100*((k+1)/(len(di_question)+1)) )]#((k+1)/len(di_question))**3.2
di_weight_all = 0.0
for k in di_overlap:
di_weight_all += di_weight_p[k]
return di_weight_all /(len(di_answer)+40)
评论列表
文章目录