Although still few people, upon hearing the name archive, probably recall an image other than colorful files and large tapes, we all know that those things now sit somewhere in the long queue of our nostalgias, and the archive industry entered the digital world years ago (from the 1990s). Even so, selection, review and other archive processes are still carried out mainly by human labor, and this work is feasible for as long as the volume of archived data makes review by a human agent and appraiser possible. But the increase in data volume has changed the story, and the need for machine tools has emerged.
The history of using artificial intelligence in archives and records management goes back to the past decade, when it was tested in a limited way to assist part of the archive process. But recently, with the increase in research and investments in the field of artificial intelligence, we are witnessing a new trend in which artificial intelligence is used across the entire archive process. In light of this, the authors of this article, using a documentary method, have reviewed research conducted on the applications of artificial intelligence in archives and its future prospects.
The impact of artificial intelligence on the archive industry
With the integration of artificial intelligence into the archive industry, in the first instance its core concepts such as provenance, records preservation, and how records are arranged and organized are affected, because artificial intelligence provides new ways to extract and organize information. Also, given artificial intelligence's ability to analyze big data, the appraisal process and which records should be retained also change. Even so, debates arise about concepts such as authenticity, fixity, persistence, and uniqueness.
As a result of the presence of artificial intelligence, many routine and procedural tasks can be automated using artificial intelligence. At the same time, human skills such as storytelling, contextualization, and public engagement require skilled human labor more than ever. In these conditions, archivists will need new data and skills, and as a result of these skills a gap emerges between proficient and traditional people in this industry. Academically as well, the new field of "computational archival science" is emerging at the intersection of archival science and computer science.
Automating the archiving process
Artificial intelligence techniques such as machine learning can be applied to analysis and appraisal at large scales. For example, human appraisers typically determine which of recorded sounds (for example music, interviews, and so on) have archival value. But when we face big data, this labeling method is time-consuming, and natural language processing can help extract metadata and classify it. But despite the promising capabilities of artificial intelligence in automating archive processes, there are also limitations such as gaps between computer science and archival science, the need for very large training datasets, and cleaning of existing data, and this has made the development of artificial intelligence tools specific to archival work and interdisciplinary collaboration necessary. So although artificial intelligence can automate various stages of archive workflow and processing, it also faces limitations that must be addressed.
Organizing and accessing archives
Artificial intelligence provides new methods for organizing and accessing archives beyond traditional principles. Traditional access to archives is based on principles such as provenance and original order, which organize records by creator or original structure. In these conditions, artificial intelligence techniques have made it possible to analyze and extract meaningful information from archived content based on entities, topic modeling, word embeddings, visual analysis, and so on. Content-based indexing and organization have provided new ways to search and browse. For example, users can now search collections for a person or a topic. So, in other words, by going beyond the original classifications of archive creators, artificial intelligence has provided new ways to extract content and index it.
New forms of digital archives
Artificial intelligence plays a role in collecting and managing new types of digital archives beyond traditional institutional assets. Social media and web content generate a huge volume of digital documents of potential value for archives. Artificial intelligence techniques such as natural language processing and computer vision have made extensive collection and analysis of these emerging sources possible. For example, web and social media archives use artificial intelligence to take snapshots of digital content. So artificial intelligence has created new opportunities for the archive industry.
Even so, these new forms, which are mainly voluminous and unstructured data — compared with other documents — have created challenges such as ethical issues concerning privacy that require new management and solutions.
Discussion and conclusion
This article has comprehensively examined debates and applications at the intersection of artificial intelligence and archives and sets out the impacts, opportunities, and challenges of the presence of artificial intelligence in the archive industry.
The use of artificial intelligence in the archive industry has provided new methods for extracting and organizing information. Artificial intelligence has transformed the manner of appraisal given the possibility of using algorithms to analyze big data, and using it facilitates the capture of new digital records in places such as social media. At the same time, it also brings conceptual, practical, and ethical challenges.
Resolving the existing challenges and improving the current situation in the future requires interdisciplinary collaboration among archivists, humanists, and computer specialists.
Machine-translated from Persian by Namafar; the Persian original is authoritative. Read the original in Persian