AASL Column, June 2024

Barret Havens and Barbara Opar, column editors

Column by Barret Havens, University Librarian and Associate Professor, Woodbury University

Using ChatGPT to Translate Abstract Architectural Concepts into Controlled Vocabulary Terms

The Artificial Intelligence chatbot ChatGPT, when fed appropriate instructions (also known as “prompts”), is an effective tool that can be used to overcome challenges with connecting faculty and students with architectural information resources. For example, most architecture librarians have encountered situations where one of our users expresses an information need related to an abstract architectural concept that is not adequately represented by synonymous terms within controlled vocabularies such as Library of Congress Subject Headings (LCSHs). Although architecture librarians can employ a variety of techniques to identify relevant LCSHs in cases like this in order to locate information within a library’s discovery system, ChatGPT can be used for this purpose quickly and effectively. Three scenarios demonstrating the effectiveness of using ChatGPT in this manner will be explored within this article.

Scenario 1: A user is looking for books that explore the topic of architectural massing

Problem: The author (an architecture librarian) and the user he was assisting struggled to find descriptors relevant to the concept of architectural massing within the controlled vocabulary of existing LCSHs. Phrase searching “architectural massing” did not provide adequate results.

The screenshot below demonstrates how a prompt was fed into ChatGPT to identify relevant, existing LCSHs on this topic:

Scenario 1 outcome:

The subject heading Architecture—Composition, proportion, etc., which was recommended by both ChatGPT and by Association of Architecture School Librarians (AASL) members during a listserv discussion about the difficulty of locating books on architectural massing, was used successfully in the author’s library discovery system to target relevant books on the topic, which satisfied the user’s need.

It is worth noting that further discussion about the lack of an adequate LCSH related to architectural massing took place within the Association of Architectural Librarians (AASL) Listserv during the fall, 2023 semester. This discussion resulted in the submission of a proposal to add architectural massing as an official LCSH. This proposal was accepted, and, as a result, architectural massing will be added as an LCSH.

Scenario 2: A user is looking for books that explore the topic of interstitial spaces

According to Abdel & Pintos, interstitial spaces are “(In-between) spaces, that act as buffers to, and link our private spaces to the public and functional buildings or landscapes. They are the hallways, waiting areas, elevators, staircases, entrances, and transitional zones that weave our built environment together.”

Problem: Neither interstitial spaces nor in-between spaces are existing LCSHs. Phrase searching either of those terms failed to identify adequate resources within the author’s library discovery system.

The screenshot below demonstrates how a prompt was fed into ChatGPT to identify relevant, existing LCSHs on this topic:

Scenario 2 outcome:

Although some of the 7 LCSHs identified by ChatGPT were deemed either too broad or only tangentially relevant to the concept of interstitial spaces, the LCSH Transitional spaces (Architecture) was used successfully in a subject search to identify resources that satisfied the user’s need.

Scenario 3: A user is looking for books that explore the topic of narrative architecture

Problem: Narrative architecture is not an existing LCSH. Although phrase searching “narrative architecture” did produce a list of results that included books that the user deemed relevant, both the user and the author felt the need to dive deeper into the topic by using other search terms that are synonymous with narrative architecture.

The screenshot below demonstrates how a prompt was fed into ChatGPT to identify relevant, existing LCSHs on this topic:

Scenario 3 outcome:

Although the LCSH Storytelling—Architectural aspects suggested by ChatGPT sounded highly promising as a synonym for the concept narrative architecture, it proved to be a rarely used LCSH and, consequently, did not return adequate results when plugged into the author’s library discovery system as a subject search. However, when storytelling was used as a keyword in combination with architecture (storytelling AND architecture), useful items were identified that were not discovered with the original search using the phrase “narrative architecture”. Therefore, ChatGPT’s suggestion to use storytelling as a synonymous term did prove useful after all. Furthermore, though it is a slightly broader translation of the concept narrative architecture, subject searching using the LCSH Architecture—symbolic aspects, which ChatGPT also suggested, yielded useful results (books related to symbols as a way of conveying meaning through architectural design, which is closely associated with narrative forms of architecture).

In summary, although there are a variety of techniques known to architecture librarians, and tools other than ChatGPT that could be used to identify relevant LCSHs, in all three scenarios described in this article, ChatGPT provided suggestions that were used to resolve the problem noted for each scenario quickly and efficiently. It is likely that in the not-so-distant future, A.I.-powered features will be incorporated into most library discovery systems, potentially obviating the need to use ChatGPT as a separate platform for the purpose of identifying relevant LCSHs. In the meantime, ChatGPT performs the function well.

Citation

Abdel, Hana, and Paula Pintos. “Interstitial Spaces and Public Life, the Overlooked Interventions That Weave Our Built Environment.” Archdaily, 8 July 2022, archdaily.com/984818/interstitial-spaces-and-public-life-the-overlooked-interventions-that-weave-our-built-environment . Accessed 24 June 2024.