A page from an old manuscript featuring handwritten text in Devanagari script, likely in red and black ink. Below the text, an illustration depicts a seated man with a shaved head and a loincloth, holding a ritual object. To his left is a tiger-like animal with human-like facial features, lying down. To the right are decorative plants, suggesting a natural or symbolic setting. The page shows visible wear, with uneven edges, hinting at its age and historical significance.

Hindi Manuscript 884. n.d. Wellcome Collection.

Technology moves fast. I remember marveling at the transformative power I held in my first flip phone, just as I recall a decade later commenting on its obsolescence. Staying current takes an ever-accelerating cycle of adopting new technologies and moving on from previous innovations. All of which brings me to JSTOR’s Text Analyzer, which debuted in 2017 and was retired on August 30 of this year.

When it launched, Text Analyzer was a transformative innovation. Text Analyzer allowed researchers to search on JSTOR not by using keywords and advanced search, but by uploading or pointing to a document. Text Analyzer used natural language processing to read the document, figure out what it was about, and then use those terms to find other documents in JSTOR about the same subjects. Article after article described the practical impact of this approach on real-world research challenges. It helped junior researchers who could get stuck “keyword-thrashing” looking for the right set of terms that would bring back relevant literature. And it also helped senior researchers conduct literature reviews, by breaking down disciplinary silos.

After it was released, Text Analyzer continued to evolve and innovate. We released an API that allowed libraries and publishers to apply the same techniques to their own corpora. And we added a multilingual ability, in which you could upload documents in any of fifteen languages and the tool would help you find related materials in English. In 2018, Text Analyzer won the Society for Scholarly Publishing’s first ever People’s Choice Award.

Time, as it tends to do, marched on. Soon, the technologies that made Text Analyzer possible— natural language processing techniques like labeled topic modeling—were eclipsed by new technologies, such as the transformers that power large language models like ChatGPT. JSTOR itself has been at the vanguard, adopting these approaches to help researchers: in 2023 we launched the beta program for an interactive research tool that uses generative AI to help researchers conduct their research. If you used Text Analyzer primarily to conduct research— including discovering related materials—I encourage you to check out this new tool.

Some people who used Text Analyzer used it less as a jumping off point to other documents and more as a way to better understand their own materials. For example, a user might upload their dissertation to see which keywords and topics were discovered, helping them to see potential blindspots in their analysis. If you used Text Analyzer to better understand your own materials, I encourage you to explore Constellate. This service offers a wide range of tools and a comprehensive skill-building program to teach, learn, and conduct text analysis. Constellate provides even more advanced features for analyzing your own texts, visualizing data, and uncovering insights across large corpora.

Technology’s advancement is rapid, affecting us all. JSTOR will continue to evolve, using these advancements to help researchers do their important work. As they do, the tools we provide will change and grow. We’re glad Text Analyzer was such a useful tool while it was here, and now we’re excited that the capabilities it offered are available and even improved upon in other services. Constellate provides advanced text analysis features and educational resources, while JSTOR’s interactive research tool leverages AI and other advanced technologies to elevate research, teaching, and learning. These tools represent our commitment to evolve alongside technology to better serve the academic community.

Single author

About the author

Alex Humphreys is Vice President, Innovation, at ITHAKA, where he leads a team that scouts and develops the future of research and education through projects, partnerships and investments. Through JSTOR Labs, the Innovation team works with partners to create tools for researchers, teachers and students that are immediately useful—and a little bit magical.