Meet Cleo – the AI Egyptology Research Platform

Digital technology continues to make a foray into the world of Egyptology with scholars becoming more and more willing to adopt new technologies in the classroom. The Nile Scribes previously explored some great digital tools for use in the classroom. Since then we discovered a new platform launched this year that helps with researching the vast repertoire of Egyptian material culture: Cleo. For this week’s blog we spoke with Heleen Wilbrink, the creator behind Cleo, about the possibilities of using this new Artificial Intelligence (AI) research platform to study Egyptian material culture.

Nile Scribes: You recently presented Cleo at the CIPEG conference held at Swansea and the ASOR conference in Denver. What can you tell us about your project?

Heleen Wilbrink: Cleo is a virtual research environment for people studying ancient Egypt. You can search online over 30,000 ancient Egyptian objects by text, image, and location, in either English or Dutch. The objects come from the Brooklyn Museum, National Museum of Antiquities [NS: in Leiden, Netherlands], The Metropolitan Museum of Art, and The Walters Art Museum. Cleo is now available in beta at cleo.aincient.org. We plan to add many more collections and other sources in the coming years.

The welcome screen of Cleo (Photo: cleo.aincient.org)
The welcome screen of Cleo (Photo: cleo.aincient.org)

NS: What led you to come up with Cleo?

HW: I was heavily frustrated when I was searching for sources for my master’s thesis on the Deir el-Medina stelae at Leiden University. It felt like looking for a needle in a haystack. This was over 10 years ago, so many museum collections were not yet online. In the meantime, more and more museum collections have become available, but scholars (including students and Egyptophiles) are still struggling to find sources. Collections are not searchable all at once and do not follow the same standards. They use different languages, are often not user friendly, and do not offer the opportunities of image-based searches. The Global Egyptian Museum has proven that combining multiple collections on one platform is valuable. However, that project has not been updated for over a decade. That is why we developed Cleo.

NS: How did you develop Cleo?

HW: I often get asked about this. I started fantasising about connecting online resources for Egyptology about five years ago. After completing my master’s in Egyptology, I studied marketing and worked at a bank with big data, Artificial Intelligence (AI), IT, and sponsorship for many years. I thought: ‘It should be possible to apply the latest technology to the study of Egyptology by forming partnerships between academia and businesses’. I presented a first concept of this connected platform at the International Congress of Egyptologists in Florence. It was well received and this gave me the confidence to pitch it to Google. Together with Google, one of their partners, and the National Museum of Antiquities in Leiden, we created a prototype. The prototype led to an innovation grant of the Dutch SIDN fund and sponsorship in kind by Google Cloud. So I quit my day job and founded the social enterprise Aincient – working on my vision full time. With the help of many others we created the online museum or virtual research environment called Cleo. Many other people helped along the way and are still involved. They range from brainstormers, developers, validators of the ideas, and testers, to museum collections and university staff and students, to AI specialists and fellow entrepreneurs. This led to a successful launch of Cleo in September of this year.

Users can search by text or by image (Photo: cleo.aincient.org)
Users can search by text or by image (Photo: cleo.aincient.org)

NS: How does Cleo facilitate research into ancient Egypt?

HW: Cleo helps you find more sources for your research in less time so you can make new discoveries. You can search available materials from multiple international collections on Cleo in English and Dutch, and on any device you like (for example mobile or laptop). You can conduct a text search or upload an image of a particular object and find similar objects, which can be filtered by period or location. A personal collection of objects can be created for later reference or you can download the texts and images immediately.

Users can further refine their search by finding similar objects using the AI search function (Photo: cleo.aincient.org)
Users can further refine their search by finding similar objects using the AI search function (Photo: cleo.aincient.org)

This is how it works: after creating your Cleo account, you can start with a text search or upload an image. Then, you can refine with filters and explore your results. The metadata (descriptions) of all museum collections are standardised using Thot Thesauri, Pleiades, and a lot of correction work. The result is that you can use harmonised filters such as ‘Material’ or ‘Periods’. Find similar objects using our unique AI search by selecting objects of your interest. This feature is still in beta, which means it is not perfect. But it can offer you good suggestions for your research that you might miss in a text search, possibly because the descriptions of the museum are incomplete or inaccurate.

From within a list of scarabs, users can then employ the AI search function to find similar objects (Photo: cleo.aincient.org)
From within a list of scarabs, users can then employ the AI search function to find similar objects (Photo: cleo.aincient.org)

NS: With the continuing advances in digital technology, what potential do you foresee for Cleo?

HW: Our vision for the future includes several important elements. We want to not only add many more collections to Cleo – ideally from over 850 collections worldwide, but also to make Cleo available in more languages, like Arabic. The AI search function will also require improved functionality. We hope to improve the quality and introduce a new feature: searching for similar details of an object instead of the entire object. For example, I might be interested in all examples of a god in a specific posture on different types of material. Of course, we want to add other kinds of resources such as literature and relevant information from Trismegistos and other text databases.

The next step will be answering more complex questions about ancient Egypt, for example about the developments in art history over time and regions. What is needed is the addition of other kinds of resources such as literature and data sets on ancient Egyptians which can be pulled from online resources such as Trismegistos. A simple use of the literature would be that for a specific object you would automatically find all references in the literature and be able to read the texts of these references. Another use would be to search for names mentioned on a specific object to find other attestations we have for them, thanks to an integration with Trismegistos. A more challenging step would be to text-mine the integrated resources and find new patters that went unnoticed so far.

Cleo incorporates metadata and images of Egyptian objects from museum collections around the world (Photo: cleo.aincient.org)
Cleo incorporates metadata and images of Egyptian objects from museum collections around the world (Photo: cleo.aincient.org)

NS: Egyptologists are notorious for embracing new forms of technology only slowly. How do you see Egyptologists changing our discipline by incorporating new technologies in the future?

HW: Digital natives raised with the internet, email, social platforms, and now Artificial Intelligence, are experiencing the direct benefits of technology in their personal and professional lives. You are interviewing me for an online blog that will be shared amongst your readers via an email newsletter, which is perfectly normal nowadays. I see a trend amongst younger Egyptologists of adopting new technologies while studying ancient Egypt. For example, they search Google for sources, through online museum collections and the Online Egyptological Bibliography. They wish for Egyptology to incorporate the same seamless online solutions which they experience in their daily lives.

I do see differences in the adoption of aspects of Cleo among generations. There are, for example, mixed reactions to the automated translations offered by Cleo. For the younger generation, this feature is seen as an advantage because it allows you to search in your preferred language. The Dutch collection can now be searched by a native English-speaker as well. People are accustomed to this feature on Facebook and other review sites. However, people from the older generations will argue that the quality of the translation is not good enough, although the original text and a link to the online original is always just one click away.

Fortunately, change comes from older generations of Egyptologists as well. At the end of the twentieth century and the beginning of this one, great work has been done by the Groupe Informatique et Égyptologie of the IAE. Multiple platforms have been created for the study of ancient Egypt such as the Thesaurus Linguae Aegyptiae, Trismegistos, the UCLA Encyclopedia of Egyptology and Digital Giza. In the end, the best advice I received on creating momentum in digital Egyptology was “Dance with the dancers”. Work together with the people who are willing to collaborate. I hope that you, dear reader, feel like dancing.

Leave a comment below to tell us what you think of Cleo!

(Cover Photo: Metropolitan Museum of Art)

Leave a Reply