Librarian of Things

Mita Williams. Her blog.

Libraries can provide alternatives to Ai with queryable databases


I’ve been working through some ideas about how libraries can position themselves as alternatives to large language models (llms) while also working through possible means by which we can augment and improve current llm practices and packages.

It is unfortunate that many people feel more comfortable asking llms their research questions rather than with librarians but here we are.

In response, this post works through one possibility: what if we created data sets that are designed to answer questions?

§ Chat XLS?

I cannot ask my library catalogue to give me a list of Canadian Law Reviews and Journals that we currently subscribe to, so I had to make this spreadsheet of current titles by hand.

While I was making this guide, I made Wikidata pages for those publications that were lacking a presence in Wikidata. Here’s an example of one the pages that I made for this project: TMU Law Review.

You might be asking yourself, why? Why do all this data entry to add this information into Wikidata? And that’s a fair question.

For a good summary of some of the benefits of Wikidata inclusion for those who manage, organize, or use scholarly publications, I’d recommend reading this 2024 article, LIS Journals’ Lack of Participation in Wikidata Item Creation.

For what its worth, I added these journals for a reason not captured in the above article; I added these titles because I wanted to demonstrate how libraries and library staff could contribute to Wikidata as a service for our mutual benefit and use.

§ Let’s ask questions about Canadian law reviews and journals

What questions should we be able to get good answers to while doing the work of contributing to or supporting Canadian legal scholarship? I came up with these:


§ How I re-used data from other Law Librarians to improve Wikidata records

Here’s another reason to convert our bibliographic information into linked, openly licensed data. Doing so allows for greater potential connection and re-combination with other openly linked data sets.

Recently, I learned of an open dataset from McGill librarian Ana Rogers and student reference librarian, Melissa Moreau:

Law Journal Open Access Policies

The data collected represents an environmental survey of academic law journal open access publishing policies. Journal selection is based on publishing practices among law faculty members from six research universities across Canada. Publication data is based on records from Web of Science (WoS) and OpenAlex, open access policy data was primarily collected manually from publisher websites. (2025-04-21)

If you will indulge me, I would like to show you how I augmented the Wikidata entries of these Law Journals so that they could include their copyright license information.

First, I downloaded the dataset from Borealis and uploaded it to my local copy of Open Refine.

I then reconciled the column of journal names against law review and scientific journals names in Wikidata, with additional information being pulled from the ISSN column to help confirm that the right titles were being matched.

Of the 133 journals that could be matched with Wikidata entries, only 14 had a clear designation of licensing information available to potential authors.

I then added these licenses to their respective Wikidata pages. (For complicated reasons, I did end up doing this manually, but there are ways to easily have OpenRefine do this work along with Quickstatements).

Now we can try asking this question again. Before I started this blog post, the query below returned three journal titles. Now let’s see what it returns…

We can make this list grow, together:

Fediverse Reactions

Mita Williams

Leave a Reply

Your email address will not be published. Required fields are marked *