Tech could someday let people even in dry climates
get clean water straight from the atmosphere›››
Bedouins are experiencing major shifts, transforming from a nomadic people into a population comfortable with modern technology and city living. With this shift, however, comes the danger of forgetting the old ways entirely.
If the people are no longer practicing, the heritage needs preserving in other ways.
The UAE National Archives and Library has been collecting records about life in the region since 1968. Today, more than 5 million historical documents are stored at the National Archives, including 800 audio-visual recordings of Emirati storytellers and elders, preserving dialects, songs and folktales, as well as photographs, manuscripts and written histories.
Now, the challenge is to digitize everything for greater longevity. Digitization offers more than the safe storage of cultural content, however; it also offers wider availability. An ICT storage system needs to be more than a repository. It requires integrated search functions with well-organized metadata-tagging systems, interoperability with other IT systems to future-proof a collection, and easy access for archivists and curious laypeople alike.
Digitization is essential in modern management of heritage collections. The process can be as simple as photographing or scanning an extant document to be stored as a digital image, but it’s the IT architecture that’s the complicated part.
Additionally, meticulous planning is required for developing the metadata that will drive this architecture. If assets aren’t tagged appropriately, the smartest search function in the world will struggle to return them.
CHECKING FOR BIAS IN AI
Artificial intelligence can serve as a powerful tool in identifying and tagging digital-heritage assets, but it’s important to consider how bias in AI systems can affect the data collected. Careful cultivation of “cultural intelligence” is vital to developing an AI to avoid bias in identifying images of certain demographic characteristics and ensure appropriate and sensible tags and captions are generated.
Athol Yates, professor of humanities and social sciences at Khalifa University, uses examples from World War 2 archiving to explain:
“During WWII, there were U.S. aircraft and a facility located in Sharjah, UAE, but this isn’t a well-known fact. Confirmation bias was at play in the archive where I found a photo of an aircraft in Sharjah labeled as an aircraft in Egypt. Fortunately, this was found in the Australian War Memorial archive, which had a system to inform the archivist of the issue. There will be other examples of photos with an Arabic setting tagged as other areas of the Middle East, rather than Sharjah where it may actually have been.”
Not to mention, all of this tagging and archiving needs to be scalable. Digital library systems need to be able to expand to accommodate additional assets. Beyond the sheer volume of items, the system should also be able to handle different types of content, the number of users accessing the system at a given time, and the different ways the content could be accessed. The more complex they become, however, the more difficult to maintain, extend and reuse. The system needs to be flexible, not unwieldy.
When asked how difficult it is to build an IT system that stores, archives and makes easily accessible and searchable hundreds of thousands of photos, Dr. Ibrahim Elfadel answers with just three words: “Use the cloud.”
Now a professor of electrical engineering and computer science, Elfadel was a senior scientist at IBM before switching to academia. “The research on storing, archiving, indexing and searching digital media is quite old, but with the emergence of web services, early research work has become mainstream,” he says.
The research on storing, archiving, indexing and searching digital media is quite old, but with the emergence of web services, early research work has become mainstream.
– Ibrahim Elfadel
“For instance, when we place our smartphone photos on the cloud, they get stored, indexed and retrieved according to methods based on that early research. AI is now playing a larger role in retrieving digital media based on complex user queries.”
REGION’S LIBRARIES STEP UP
Libraries across the region are rising to the challenge of digitizing their history.
In 2021, Qatar Digital Library had a 2-million-page repository of culture, heritage and history. Each page highlights a historical report, letter, map, photograph or sound recording, with an explanation in both English and Arabic.
Art Jameel is a project preserving heritage and supporting creative enterprises rooted in local heritage in AlUla, Saudi Arabia. It trains photographers in advanced photogrammetry, a technique used to document objects in 3D, and is leading efforts to record and preserve the rich archaeological heritage of the region, especially the rock art.
Art Jameel has also operated abroad, using digital-documentation techniques to record evidence of Yazidi cultural heritage in Iraq. Working in the Sinjar region, the Art Jameel team used aerial photography and photogrammetry to construct 3D models of sites destroyed by conflict.
In Oman, an online training program seeks to turn young Omanis into keen digital-heritage curators, providing students and recent graduates from all disciplines with the digital skills and tools needed to conserve their history.
The National Library of Israel received a donation of more than 50 years of documentation of the Bedouin community living in the Sinai Peninsula. The donation is the work of Dr. Clinton Bailey, author and researcher of Middle Eastern and Islamic history, who spent his life recording and collecting materials from the last Bedouin generation to grow up in the pre-modern period to capture an orally transmitted ancient heritage. He conducted research throughout the deserts of Sinai and the Negev, speaking with the Bedouin people to collect their knowledge, memories and history.
COLLABORATION WITH UAE
The archivists involved in curating the Clinton Bailey Archive of Bedouin Culture are now working with archivists in the UAE to transcribe more than 350 hours of interviews in the various Arabic dialects spoken by the Bedouin people.
Dr. Samuel Thrope, curator of the Islam and Middle East collection at the National Library of Israel, explained that while the storage and cataloguing of these recordings were the technical challenges, they were simple compared with the careful transcription of the content: “We have all the recordings digitized and cataloged. They’re fully searchable with rich descriptions of the topics discussed and full names and tribal affiliations of the interviewees.
“The next stage is transcribing them. Robust cataloging — that’s not new. That wasn’t the challenge. The challenge was how to make the transcriptions available and searchable — and accurate. All these dialects, they’re difficult to transcribe and translate. Working with the professionals at the National Library and Archives in the UAE has been so helpful in solving this. We can learn from their experiences, from the work they’ve done in archiving rare and precious recordings of the Arabic language.”
As for the other side of archiving — the accessibility and democratization of knowledge, as Thrope puts it — the library is forward-thinking:
“All the vast collections we have are open and accessible to everyone everywhere around the world. Maybe we don’t know why people want to access the information and it’s OK that we don’t know. We’re collecting not just for the present, but for the future. We know we have to preserve, protect, record.”
We have all the recordings digitized and cataloged. They’re fully searchable with rich descriptions of the topics discussed and full names and tribal affiliations of the interviewees.
– Samuel Thrope
GIVING OLD PHOTOS NEW PUNCH
Artificial intelligence has revolutionized the way old images are not only tagged and archived but restored and cleaned. By using sophisticated algorithms, AI can analyze and improve the quality of old photographs, even those that have suffered from decades of wear and tear.
Machine learning techniques can be used by training models on large datasets of high-quality images to learn patterns and features that can be used to restore and enhance older images. By analyzing and identifying the differences between the old and the new, AI models can determine the best way to restore an old image.
Neural networks are one such technique. These networks are designed to mimic the structure and function of the human brain, allowing them to learn and adapt to different types of images. They can learn to identify patterns and features that are specific to high-quality images, such as sharpness and light balancing, and then apply these features to old and degraded images.
In this way, AI can enhance and clean the images by removing noise and other imperfections. Human photo editors can use these techniques and their own expertise to improve the overall quality and clarity of older images, bringing them into the future in ways that were previously impossible, preserving them for anyone interested in the past.