Skip to main content
 首页 » 编程设计

python之从三个列表中查找相关实体

2024年11月24日11haluo1

我有三个包含以下数据的列表:

Entities:  ['Ashraf', 'Afghanistan', 'Afghanistan', 'Kabul'] 
Relations:  ['Born', 'President', 'employee', 'Capital', 'Located', 'Lecturer', 'University'] 
sentence_list: ['Ashraf','Born', 'in', 'Kabul', '.' 'Ashraf', 'is', 'the', 'president', 'of', 'Afghanistan', '.', ...] 

sentence_list 是一个句子列表。在每个句子中,我想检查 Entities 和 Relations 中的任何单词,特定单词的组合是否应该添加到另一个列表中。例如,第一句中的 (Ashraf, born, Kabul)。

我做了什么:

第一个不完整的解决方案:

# read file 
with open('../data/parse.txt', 'r') as myfile: 
    json_data = json.load(myfile) 
 
for i in range(len(json_data)): # the dataset was in json format 
     if json_data[i]['word'] in relation(json_data)[0]: # I extract the relations 
         print(json_data[i]['word']) 
     if json_data[i]['word'] in entities(json_data)[0]: 
         print(json[i]['word']) 

输出:(Ashraf, Born, Ashraf),我想要的地方是(Ashraf, Born, Kabul)

下一个不完整的解决方案:我将 json_data 存储到一个列表中,然后我这样做了:

json_data2 = [] 
for i in range(len(json_data)): 
    json2_data.append(json_data[i]['word']) 
print(json_data2) 
 
 
''' 
Now I tried if I can find any element of `Entities` list and `Relations` list 
in each sentence of `sentence_list`. And then it should store matched  
entities and relations based on sentence to a list. ''' 
 
for line in json_data2: 
    for rel in relation(obj): 
        for ent in entities(obj): 
            match = re.findall(rel,  line['word']) 
            if match: 
                print('word matched relations: %s ==> word: %s' % (rel,  line['address'])) 
            match2 = re.findall(ent, line['word']) 
            if match2: 
                print('word matched entities: %s ==> word: %s' % (ent,  line['address'])) 

不幸的是,不起作用?

请您参考如下方法:

您可以使用以下 list comprehension :

to_match = set(Entities+Relations) 
l = [{j for j in to_match if j in i}  
        for i in ' '.join(sentence_list).split('.')[:-1]] 

输出

[{'Ashraf', 'Born', 'Kabul'}, {'Afghanistan', 'Ashraf'}] 

请注意,我正在返回一个 sets 列表以避免重复值,因为例如在 Entities 中 Afghanistan 出现了两次。

有用的读物​​: