search

SALT-NLP / Structure-Aware-BART

Source codes for "Structure-Aware Abstractive Conversation Summarization via Discourse and Action Graphs"
MIT License
64 stars 8 forks source link

ADSC dataset #6

Closed chijames closed 3 years ago

chijames commented 3 years ago

Hi,

I tried several runs of the S-BART w. Discourse&Action (last row of Table 2) on the SAMSum dataset, and the results are sometimes higher/lower than the BART baseline, which is probably caused by random seeds. Nevertheless, the performance gain on the ADSC dataset seems to be more substantial according to Table 3. I was wondering if it's possible to also release the code to test the model on that dataset?

Thanks.

jiaaoc commented 3 years ago

You could directly use your trained models to evaluate on ADSC.

jiaaoc commented 3 years ago

And the same way to extract the discourse graphs/action graphs

chijames commented 3 years ago

Is it possible that you still have the processed files and can upload them to Google drive?

jiaaoc commented 3 years ago

I might not keep those files, let me check.

jiaaoc commented 3 years ago

Sorry, i did not keep those files. But they should be easy to be generated following what you did on SAMSum.

chijames commented 3 years ago

ok, thanks.

chijames commented 3 years ago

Sorry to reopen this, but there are 5 summaries per dialogue in the ADSC dataset, may I know which one is used for evaluation?

jiaaoc commented 3 years ago

I think I used the first one.