howardhsu/de-cnn

Code for ACL 2018 paper "Double Embeddings and CNN-based Sequence Labeling for Aspect Extraction"

de-cnn's Issues

laji

How is the data organized in npz files?

Hello, thank you for sharing the code.
It seemed that train data was processed and saved in the npz file. Would you please explain that how the data is organized in .npz file? I found that the shape of train_X is [2895,83], valid_X is [150,83]. What's the exact meaning of these information? Many thanks!

Cross domain usage of model

Can we use the same model for cross domains, i.e., other than restaurant and laptop?

eval.jar and A.jar

hello,
Sorry, I am not code expert. Maybe it's a simple question. I am wondering, where do I find the eval.jar and A.jar. I have downloaded the other files and placed in the folder but I don't know how to create or where to find the jar file. Can you please guide me? thanks

Using words in the test data to build the embedding matrix

As per my understanding, we are not supposed to take any hint (or any information) from the testing data for the model construction. However, in your implementation, when the word_idx.json is computed, the words in the testing set are also considered. (As per my understanding)

Can you please clarify on this matter.

FileNotFoundError: [Errno 2] No such file or directory: 'java': 'java'

when i run evaluation.py there are some errors as follow :

Error: Could not find or load main class Main.Aspects

Caused by: java.lang.ClassNotFoundException: Main.Aspects

subprocess.CalledProcessError: Command '['java', '-cp', 'script/eval.jar', 'Main.Aspects', 'data/official_data/pred.xml', 'data/official_data/Laptops_Test_Gold.xml']' returned non-zero exit status 1.

i do not know what's wrong here

F1 score replication

I have run the same code but could achieve the F1 score of 71.9 with the same hyperparameters, instead of 74.37 ( mentioned in the paper ). Can you please suggest what might have gone wrong?
Thanks in advance!

Amazon Review Dataset Laptop for Train Embedding

Excuse me sir, I want to ask about Amazon Laptop Dataset that you mention it on your paper when trained domain embedding. How do you get the laptop review only from the amazon review dataset? Maybe you can give explanation or the code. Thanks in advance.

Yelp Dataset for Train Embeddings

Hello sir, I want to ask about the yelp dataset challenge 4 for train the embeddings

1. In the paper you said "We only use reviews from restaurant categories that the second dataset is selected from Cuisines.wht". Are you filter the restaurant category first then filter based on the category in file Cuisines.wht or do something else?
2. And are you filter the location of the restaurant too? Ex : restaurant in new york only or etc.
   Thank you sir

mask value in crf

Hi,
Thanks for the code.
I wanted to what is the x_mask and how is it calculated when passed to crf ??

issue with gen.vec file

How to get a gen.vec file from GloVe: http://nlp.stanford.edu/data/glove.840B.300d.zip?

Missing .jar files

Hi,

thanks for your great work!

Do you happen to know, where one can download both required .jar files?
I'm unable to download them from the official pages, making it impossible to evaluate my models :(

Am I doing something wrong or looking in the wrong place?

Thanks in advance!

Another Language

First of all thank you so much for your effort in this code. I want to ask you if we can use this code with another language such as the Arabic language.

Thanks a lot,

CRF开关问题

您好，感谢开源！我试了你程序中的CRF，但是CRF的loss值非常大，请问您程序中的crf层是怎么使用的呢？谢谢

如何原始xml文件得到restaurant.npz文件

感谢您的代码。
请问，您是如何从原始xml文件得到restaurant.npz文件的呢？restaurant.npz内部格式及顺序是什么？

再次感谢！

statistical significant

Hi, thanks for sharing. It is mentioned in the paper that the result is statistical significant at the level of 0.05. Can you explain how to get the number 0.05?

原始数据能否共享下

您好，是否可以提供下原始数据呢，在文件data/official_data/的全部数据，谢谢您，我在整理下载数据的时候总是断。
data/official_data/Laptops_Test_Data_PhaseA.xml
data/official_data/Laptops_Test_Gold.xml
script/eval.jar
data/official_data/EN_REST_SB1_TEST.xml.A
data/official_data/EN_REST_SB1_TEST.xml.gold
script/A.jar

数据使用

你好，首先感谢代码开源；
我在阅读代码时，没有发现在训练阶段哪个地方使用了原始的xml格式的数据，好像是.npz的数据，这里跟readme说明是否有出入，谢谢！