Exploring library catalogues as data – a lesson for ProgrammingHistorian

This repository is to support writing activities of a lesson provisionally titled as "Exploring library catalogues as data". These activities belong to ProgramminHistorian's ENABLAR programme

• ENABling Library and Archive participation in digital Research co-learning communities.

The repository will contain not only the text of the lesson and supplementary materials (images, tables etc.) but some data and scripts as well.

Planned topics of the tutorial

Methods/tools:

• MarcEdit (Argula)
• describing some important MARC data elements (Péter)
• programming MARC records with Python [PyMarc] (and showing alternatives) (Péter)
• small/local LLMs? (Michele)
• Gephi for visualisation (Argula)

Data analyses:

• extraction
  • place and personal name extraction and normalisation (with roles) (Péter)
    • enriching subject indexing/ information (Argula) – adding additional information about named entities in records (Argula) – external data to match with smaller datasets: FAST, VIAF, etc. (Argula)
    • create a map (Péter)
  • date extraction and normalisation (Péter)
    • create a timeline (Péter)
    • Consider using bibliographic data to visualise the history of early book editions from the time of their first printing (geomapping and timelines). (Arnoud)
  • subjects
    • analysing subject coverage in a collection (Argula)
• how to work across two datasets computationally (Argula)

Objectives, principles, general questions:

• Assist catalogers/metadata librarians in their day-to-day work and specialized projects (Doreen)
• How to help book historians in answering their questions? (Péter)
• Copy cataloging versus original cataloging depending on the topic of the project (Halie)
• Also being aware of local cataloging practices that may or may not impact the analysis or computational methods (Doreen)

Datasets:

• a smaller library catalogue (e.g. Estonian nat. bibl.), or a part of a larger one (Belgian, German, Czech etc.) (Péter)
• https://github.com/pkiraly/qa-catalogue#datasources
• https://about.muse.jhu.edu/muse/open-access-marc (Doreen)
• Could you pull based on a specific category in MARC or subject headings etc and then create around that. (Halie)

Open questions:

• Question for Peter (and anyone else who is interested in contributing): Is the lesson goal to show researchers how to use marc records to answer their research questions? (I.e. need to explain how to navigate a set of marc records?) Starting from sample research questions then find datasets and methods to work with them? (I am reminded of this link Peter showed me: https://bibliodata.substack.com/p/an-outline-of-an-imagined-training) Or is it to select a small set of marc records and build a model or something that can be used for i.e. lesson 2 or 6 or linked sticky notes? (Doreen)
• The focus is on training and fine-tuning a small LLM to make it an expert in dealing with bibliographic data. (Arnoud)

directory structure

• data: the input files containing unmodified library records
• data_output: output files of the analyses
• fig_output: output images of the analyses
• scripts: the scripts that contribute in data analysis
• figures: the figures displayed by the lesson itself

(This structure follows the suggestion of the Library Carpentry's Introduction to R lesson.)

Some Important Reminders:

• Tutorials should not exceed 8,000 words (including code).
• Keep your tone formal but accessible.
• Talk to your reader in the second person (you).
• Adopt a widely-used version of English (British, Canadian, Indian, South African etc).
• The piece of writing is a "tutorial" or a "lesson" and not an "article".
• Adopt open source principles
• Write for a global audience
• Write sustainably

The full guide: Writing and Formatting a New Lesson.