search

thunlp / MuGNN

Source code for ACL2019 paper "Multi-Channel Graph Neural Network for Entity Alignment".
60 stars 11 forks source link

Error when no rules are found #5

Closed fujiaz19 closed 3 years ago

fujiaz19 commented 4 years ago

Hi, I'm trying to run your code on a new dataset when after running the format_data.py, AMIE cannot find any rules for one of the graphs. Then, when I ran the train.py, it reported the following error: Screenshot from 2020-09-28 01-15-22

I wonder if you could kindly let me know how I should deal with such situation so that I could run it successfully? Thank you so much in advance!

acharkq commented 4 years ago

This error occurs in printing.

You can try to comment line366-367 and line374-375 in the `cross_graph_completion.py' file.

Let me know if that works.

fujiaz19 commented 3 years ago

Hi, thank you very much for your reply! The problem has been resolved now.

By the way, would you mind elaborating a bit more on if/how MuGNN uses attributes as I didn't find much information in the paper? Many thanks!!

acharkq commented 3 years ago

We didn't use any attributes in the MuGNN work.

You can check out my latest work Exploring and Evaluating Attributes, Values, and Structures for Entity Alignment on EMNLP 2020 for utilizing various types of attribute triples within a GNN framework.

fujiaz19 commented 3 years ago

I see, so can I ask what attr2id and ent2att files are for?

Thank you very much! I'll check that out!

acharkq commented 3 years ago

These two files are not used in the model.

acharkq commented 3 years ago

You are welcome!