rer_build_history.py 文件源码

def build_history(data_list, supported_tags_phones,supported_tags):
    history_list = [] # list of all histories
    sents = []
    count = 0
    expected = []

    for data in data_list: # data is the inputs entered by a given student
        data1 = data['data']

        #data1 is for every sentence entered by user
        for rec in data1:
            updates = rec['updates']
            sent = rec['sentence']  
            relatedTags=[]
            relations=[]
            if "rels" in rec.keys():
                relatedEntities = rec['rels']   
                expected.append(relatedEntities)            
                for i in relatedEntities:
                    relations.append(i.keys())
                    for j in i[i.keys()[0]]:
                        relatedTags.append(j)
            words = []
            posTaggedSent = postagger(sent)
            #chunkPhrases = chunker(sent)


            if len(updates) == len(posTaggedSent):
                for i in range(len(updates)):               
                    words.append({"word":updates[i]['word'],"pos":posTaggedSent[i],"tag":updates[i]['tag']})
                    #------------------------------------------------------------------------------------------------
                    # NOTE: below code is a temporary hack to build the MAxEnt for just 2 tags - we will change this later
                    if (updates[i]['tag'] not in supported_tags_phones):
                        if updates[i]['tag'] == "Model":
                            updates[i]['tag'] = "Version"
                        else:
                            updates[i]['tag'] = "Other"                
                    #------------------------------------------------------------------------------------------------

            sents.append(words)
            history={}
            history['sentence'] = words
            history['i'] = count+1
            #history['phrases'] = chunkPhrases
            history['relatedTags'] = relatedTags
            if len(relations) > 0:
                history_list.append((history,relations[0][0],))
            else:
                history_list.append((history,"None",))
            count += 1


    return (history_list,sents,expected)