有关这篇论文的方向的问题
Opened this issue · 3 comments
JayJQK commented
感谢您能够分享您的论文,我读了之后有很多启发。
有一个问题我想搞清楚,计算图的相似度是不是重点在图的结构上的相似度?因为GED就是聚焦怎么样能够让两个图的结构变成一样。
第二个问题是我想将本结构化的文本表示成图的形式,然后计算文本图之间的相似度,这可能更加聚焦在图上的节点而不是主要聚焦在图的结构上,这种任务您的论文适用么。
万分感谢!
shengdoupi commented
你好,请问楼主尝试了吗?我也在做差不多的课题(利用gnn测文本相似度)。
dlNone commented
你好,请问楼主尝试了吗?我也在做差不多的课题(利用gnn测文本相似度)。
我也在做相关的课题,交流交流?QQ:284891031
yunshengb commented
Hi,
I think GED is mainly for structural matching, but can also handle node
features. As for text matching with graphs, one challenge is how to handle
node features which are text. In that case, GED can generalize to
continuous node features and you can represent node features with
pre-trained word/phrase embeddings. Another challenge is how to construct
the graph link structure for text. There are many existing approaches for
constructing graphs from text, and I am happy to discuss more here.
Thanks,
Yunsheng
…On Mon, Jul 20, 2020 at 1:43 AM dl_None ***@***.***> wrote:
你好,请问楼主尝试了吗?我也在做差不多的课题(利用gnn测文本相似度)。
我也在做相关的课题,交流交流?QQ:284891031
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#15 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACAAS2EQSVWWFO6GKGHQPK3R4P7SXANCNFSM4J7HS4YQ>
.