Closed novice7 closed 3 years ago
我们首先是获得了话题实体(问题提及实体)周围两跳的三元组然后去运行PageRank。 当前包含部分(实体,属性,值)这样的三元组,我们没有做特别区分
Question 1: How to get Freebase triples before Pagerank. Answer: Firstly, we filter neighborhood triples around topic entities and then we run Pagerank to reserve entities.
Question 2: Is there (entity, attribute, value) triples here? Answer: Yes.
那想请问下freebase三元组数据您是从哪获得呢,因为现在freebase已经停止服务了
不好意思哈, entities.txt中确实包含值
Full Freebase Dump can be downloaded from here https://developers.google.com/freebase/.
In this paper, we use Freebase dump downloaded from Microsoft
You can use this command:
wget https://download.microsoft.com/download/A/E/4/AE428B7A-9EF9-446C-85CF-D8ED0C9B1F26/FastRDFStore-data.zip --no-check-certificate
.
After downloading it, use fb_en.txt which contains triples in English.
For subgraph extraction, you can also refer to https://github.com/OceanskySun/GraftNet.
Later, I may give a more detailed process about preprocessing datasets in this repo.
Maybe one or two weeks later.
Do you know the faster way to get the real entity name through the entity MID ? Is there the only way to get the real entity name by getting Freebase/Wikidata Mappings first from https://developers.google.com/freebase/, and then searching the wiki address ?
You can find the entity name through type.object.name attribute. For more detail about Freebase usage, I suggest that you can search it online.
您好! 请问下您在进行pagerank之前,是如何获得Freebase三元组的?
请问有方法获得(实体,关系, 值)的三元组吗, 目前是只有(实体,关系,实体)的三元组吗