Ingested CL 2024 Volume 4

[davidstap]

No description provided.

[github-actions]

Build successful. Some useful links:

• Complete site preview: https://preview.aclanthology.org/ingest_cl_2024_issue4
• Potential volumes of interest: 2024.cl-1, 2024.cl-2, 2024.cl-3, 2024.cl-4

This preview will be removed when the branch is merged.

[mjpost]

Thanks, @davidstap. I have a request, if you have time: can you migrate the ingestion script to use the acl_anthology python module instead? I think it shouldn't be too difficult:

• Documentation here
• Example scripts using it: create_hugo_data.py
• Likely @mbollmann could answer questions if there are any

[davidstap]

I looked at the documentation and example, but I'm having some difficulties migrating the script.

• I have changed the Anthology class from anthology to acl_anthology.
• Other functions that need to be changed: Paper and Volume from anthology need to be changed to their acl_anthology counterparts defined in collections.volume and collections.paper.
• Some helper functions need to be changed/removed: normalize from normalize_anth and make_simple_element, indent, compute_hash_from_file from anthology.utils.
• From the documentation and example script it is clear how to query Anthology data, but I am not sure how to change/create new data.
• @mbollmann if you have time I'd appreciate some pointers. Some insights on how to modify the main loop where data is stored in ingest_mitpress.py (lines 404-470) would especially be much appreciated.

[mbollmann]

@davidstap Thanks for starting to look into this! My hope with the new library is that manual construction of XML will be entirely unnecessary, but I haven't prioritized the necessary functionality for this yet. It’s mostly there, but not really finalized or tested. I can start with this particular ingestion script though to see what changes would be needed.

[davidstap]

Thanks @mbollmann that'd be great. Let me know if I can help.

[mbollmann]

Had a quick look now, and I would move the adaption of the ingestion script into a new issue/PR, as I don’t think it is that trivial. Currently, it’s fundamentally built around constructing XML nodes manually, which is not really something I’d like to support with the new library, as it defeats the whole point of encapsulating serialization logic within a library in the first place. The new library is designed from the ground up to have its classes closely mirror what’s in our data files, to make converting from/to XML easy, so I believe we should make proper use of that. I need to implement a couple more things for that to work smoothly, though.

@davidstap Can you send me the source file from MIT press that you used for this ingestion, so that I can test-run the script locally?

[davidstap]

Thanks @mbollmann, that makes sense and explains why it's not straightforward to make the adaption of the ingestion script. I've just shared the MIT press source file through e-mail.

@mjpost can you review the CL ingestion?