annotation inconformity problems

[alexnotes]

There exist some false annotation problems. The annotation in weiboNER_2nd_conll.* files is consistent with weiboNER.conll.*. It seems like that there exist tens of falsely annotated problems. Can you fix it?

Besides, Can you provide the word level annotated data? It should look like that:

口腔	O
溃疡	O
加上	O
这	O
玩意	O
～	O
酸酸	O
甜甜	O
好	O
滋味	O
。	O

What's more, can you provide the annotated data with conll 2003 format? The second column is POS column. Like this:

美国 NS -X- B-LOC
总统 N -X- O
特使 N -X- O
格尔巴德 NR -X- B-PER
表示 V -X- O
, W -X- O
他 R -X- O
同 P -X- O
克罗地亚 NS -X- B-LOC
总统 N -X- O
图季曼 NR -X- B-PER
的 U -X- O
会谈 VN -X- O
是 V -X- O
“ W -X- O
积极 A -X- O
和 C -X- O
有益 A -X- O
的 U -X- O
” W -X- O
, W -X- O

[alexnotes]

In annotated data, there exist unrecognized characters. For example, line 1875 in weiboNER_2nd_conll.dev file.

[VioletPeng]

We don't have these annotations. You can generate word segmentation and POS using any software you like.

[alexnotes]

For the other two questions, can you fix it?

[VioletPeng]

Hi @kisslotus,

Thank you for pointing out the possible missing annotations in the data. We collect the data using crowdsourcing, although we have 3 people annotate the same sentence, local mistakes are still inevitable. We encourage you to make corrections to the dataset and send pull request. Thanks!