can't get the result of the paper

[Tangnameless]

I think the result posted in the paper is for real news labels.

I set:
1 - pos(fake news) input: [0, 1]
0 - neg(real news) input: [1, 0]

I got:
tp:388.0
tn: 2405.0
fp:188.0
fn: 408.0
on politi fact dateset

the number of fake news is 1/10 of real news, seems ok

if we calculate pre, recall, f1 on real news
we got:
pre: 0.854
recall: 0.9274
f1:0.889

It looks like the result in the paper

[thinking024]

Forgive me, but how did you generate case_image.npy, case_head.npy and other npy format files? I'm quite confused

[bufanx]

Have you got these files now? I really require them, thank you.

[Alibizzz]

Forgive me, but how did you generate case_image.npy, case_head.npy and other npy format files? I'm quite confused

hello, regarding the meaning of 'key_list' in your code and how to generate the 'case_keys.txt' file, have you already resolved this issue?

[Alibizzz]

I think the result posted in the paper is for real news labels.

I set: 1 - pos(fake news) input: [0, 1] 0 - neg(real news) input: [1, 0]

I got: tp:388.0 tn: 2405.0 fp:188.0 fn: 408.0 on politi fact dateset

the number of fake news is 1/10 of real news, seems ok

if we calculate pre, recall, f1 on real news we got: pre: 0.854 recall: 0.9274 f1:0.889

It looks like the result in the paper

Hello, what does 'key_list' mean and how to generate a 'case_keys.txt' file, thank you

[123x12]

我认为报纸上公布的结果是真正的新闻标签。

我设置： 1 - pos（假新闻）输入：[0, 1] 0 - neg（真实新闻）输入：[1, 0]

我得到： tp：388.0 tn：2405.0 fp：188.0 fn：408.0 关于 politi fact dateset

假新闻的数量是真实新闻的十分之一，看起来还可以

如果我们计算真实新闻的预测值、召回率和 f1 ，结果为： 预测值：0.854 召回率：0.9274 f1：0.889

看起来像论文中的结果

您好，我想请问一下，为什么我按照readme来生成的pf_embedding，然后在运行的时候出“发现样本数不一致的输入变量”的错误，烦请您指导一下。麻烦您啦。