-
-
Notifications
You must be signed in to change notification settings - Fork 2.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add warning about confabulations to "New entry from plain text" #11825
Comments
Follow-up: Use the openAI API to create BibTeX. Has much better results. @InAnYan Can you "unzip" this feature wish? - Which code to use in JabRef to interact with openAI etc. Prompt is like "Please convert the following bibliography entry to a well-formatted high-quality BibTeX entry". |
are there statistics about "much better results"? |
I tried with the example of our homepage - https://docs.jabref.org/collect/newentryfromplaintext In: O. Kopp, A. Armbruster, und O. Zimmermann, "Markdown Architectural Decision Records: Format and Tool Support", in 10th ZEUS Workshop, 2018. Out 1: @Article{Kopp_2018abc,
author = {Kopp, O and Armbruster, A and Zimmermann, O},
date = {2018},
journaltitle = {Communications},
title = {Le gigantisme architectural en Union soviétique},
doi = {10.3406/comm.1985.1625},
number = {1},
pages = {45-67},
volume = {42},
publisher = {PERSEE Program},
} Out 2: @inproceedings{kopp2018markdown,
author = {Oliver Kopp and Andreas Armbruster and Olaf Zimmermann},
title = {Markdown Architectural Decision Records: Format and Tool Support},
booktitle = {Proceedings of the 10th ZEUS Workshop},
year = {2018},
} Guess which output is from which tool! I know that both are wrong. - Anita will go crazy. Refined prompt:
@inproceedings{kopp2018markdown,
author = {O. Kopp and A. Armbruster and O. Zimmermann},
title = {Markdown Architectural Decision Records: Format and Tool Support},
booktitle = {Proceedings of the 10th ZEUS Workshop},
year = {2018}
} |
Thi Lo's right, I would like to see a paper for overview of plain citation parsers with comparisons LLM/GROBID |
Description of Problem:
"New entry from plain text" is based on the Grobid project, which uses machine learning techniques. Results are probability based, therefore not always 100% accurate.
We have had users that are confused by this and think the resulting data is or always should be true.
Solution:
Add a warning about confabulations or hallucinations in the GUI. It could be phrased similar to the warning that is shown to users that make use of the AI Chat or AI Summarization feature.
Additional context
The text was updated successfully, but these errors were encountered: