Weekly Report - March 11, 2024 - March 15, 2024

report.md

Weekly Report - March 11, 2024 - March 15, 2024

Tasks Worked On

• Student Newspapers
  • Continued import of Student Newspapers -- we're almost done with Batch 01
  • All that remains there is Batch 01/batch_OandW_03
  • Struggling a bit with Solr here. I can't remember if we've always had this problem to some extent, but at points gsearch or solr is dropping out (could be gsearch) and indexing stops working.
  • Haven't explored logs to determine why
  • For now, simply addressing by restarting services and manually rekicking things off
• Started Searching for a Student Worker
  • Posting is here
• Met with Rachel to update remaining billing things.
• Met with Louis to discuss Tyler Martindale's withdraw app.
  • Long story short, they need it ASAP.
  • Before I do anything else, I'm going to work on this to point it at staging.
  • I plan to work on this this weekend so that I can meet with Louis on Monday to confirm things are working as they should so they can start training with it next week.
• Finished migration of PDF collections and started import videos.
  • Going to watch this closely over the weekend to see how this goes.
• Reviewed Crossref resolution report
  • IJNS is still not following the DOI pattern I provided.
  • Updated related IJNS articles to use the minted DOIs.
• IIIF Annotations TSG
  • Met with stakeholders and cochairs to finalize initial docs for editorial committee
• Batch Remove and Rebuild
  • Met with Kirk to demo and discuss what's happening
  • Tried to review the fix but not working for me.
• M3 Profile
  • Updated profile to make machine dates facetable and allow Pdfs to have call numbers
• ArchivesSpace
  • Modified Archiveswurm to be able to fetch trees to make creating CSVs from resources and their descendants possible.

Outside of Work

• Experimented with Claude 3 for metadata generation and computer vision things (article identification)
  • If you want to see this in action see this
  • The code is ugly, but you can get an idea of what it can do.
  • Give it a manifest for a student newspaper, your api key, and an output file and you get something like below
  • Initial take aways
    • Super cheap
    • Needs refinement
    • max_tokens limit seems to be 4096 so need to process the issue in advance and split into chunks to ensure the full this is processed
    • keywords needs to be rethought and needs a bit of code to remediate the returned value to a controlled vocab we use

# Newspaper Issue [beacon:5927](https://digital.lib.utk.edu/assemble/manifest/beacon/5927)

## Cost:

Costs to run: 

Input cost: 0.001908, Output cost: 0.00030375

## Articles and Ads Found:

* IRA issues objections on open house policy (Page 1)
* McGovern blasts Thieu government (Page 2)
* Book studies pacifism (Page 3)
* Spaceship Earth Series to beginSaturday night (Page 3)
* Students to air transportation views (Page 4)
* Vols clash with Auburn Sat. on TV (Page 6)
* Second-Rated Notre Dame Invades Purdue Territory (Page 7)
* Department heads report excellence (Page 8)
* Petting games (Page 11)
* Booker Applauds New Voters (Page 11)


## Keywords:

* university newspaper
* university of tennessee
* politics
* education
* culture
* sports
* campus life
* current events
* community
* advertisement

Thinking About Next Week

• I have a guest lecture in SIS on Tuesday night and need to work on my presentation.
• Out Wedenesday - Friday
• Questions for Phone Interviews
• Work with Louis on Tyler's thing
• Meeting with UTIA about a project where they need help with DOI minting.
• Flesh out documentation / policies for DOIs