dataset

[yassmine-lam]

Hi,

could u give us a description about the dataset format?

thank u

[yangheng95]

For the datasets, each row contains a token or word, an ATE tag and a polarity tag. If a token or word does not belong to an aspect, the polarity is -1, else: 0 for negative， 1 for neutral， 2 for positive.

However, each sample only label polarity for one aspect, other's polarity is -1 the same as non aspect words.

e.g.

I O -1
charge O -1
it O -1
at O -1
night O -1
and O -1
skip O -1
taking O -1
the O -1
cord B-ASP 1
with O -1
me O -1
because O -1
of O -1
the O -1
good O -1
battery B-ASP -1
life I-ASP -1
. O -1

[yassmine-lam]

ok thank u

[parasd086]

What's the difference between I-ASP and B-ASP?

[seominseok48349278]

What's the difference between I-ASP and B-ASP?

yangheng95/ABSADatasets#24

[liufenghupan]

您好，我正在构造一个自己的中文数据集，但是看了您的训练数据有一些问题想要请教，我看到您的每一个评论对象当中只标注一个属性，其他属性情感极性标位-1，但是您的所有的评论对象当中都只有一个B-ASP，如果一个评论对象当中有很多属性，每一个属性的标记不是都应该用B-ASP开始吗？还有就是我发现您的中文数据集中的极性只有0和2，为什么中文数据集比英文数据集少了“中性”的情感分类呢？感谢您的工作，希望能得到您的回复，谢谢。

[yangheng95]

#30 (comment)

因为原始中文数据集只有两个类别。

[liufenghupan]

#30 (comment)

因为原始中文数据集只有两个类别。

非常感谢！

[AsuSociety]

For the datasets, each row contains a token or word, an ATE tag and a polarity tag. If a token or word does not belong to an aspect, the polarity is -1, else: 0 for negative， 1 for neutral， 2 for positive.

However, each sample only label polarity for one aspect, other's polarity is -1 the same as non aspect words.

e.g.

I O -1 charge O -1 it O -1 at O -1 night O -1 and O -1 skip O -1 taking O -1 the O -1 cord B-ASP 1 with O -1 me O -1 because O -1 of O -1 the O -1 good O -1 battery B-ASP -1 life I-ASP -1 . O -1

Hi, I'm trying to prepare a data set for a project I'm doing and I couldn't understand the reference to the ATE tag, when does it get 0/1/2?
Compared to the example you gave here, my data set is divided into complete sentences that return positive/negative (in polarity).

Thanks in advance.