New Delhi: The digital archive consists of an incredible amount of archaeological reports. However, a person who wants to search for information in these archives has to do it manually, which is a tedious activity.
Archaeologist Alex Brandsen, postdoctoral researcher in digital archeology at the University of Leiden, the Netherlands, used deep learning, a form of artificial intelligence, to develop a search engine capable of very precisely searching for archaeological information. The current search engine, called AGNES, can only search general keywords and titles of PDF files, such as “Middle Ages” and “pottery”, according to a statement released by Leiden University.
“If you are looking for axes in the Middle Ages, you should now download everything related to the Middle Ages and carry out an annual search,” said the Dutch researcher quoted in the press release.
The smart search engine is a kind of “Google for archaeologists”, the statement said.
Since the Treaty of Valletta (1992), archaeologists have produced an enormous amount of reports. The Treaty of Valletta, formally the European Convention for the Protection of the Archaeological Heritage (revised), is a multilateral treaty of the Council of Europe which aims to protect Europe’s archaeological heritage as a source of European collective memory and as as an instrument of historical memory and scientific study. The 1992 treaty regulates the management of Europe’s archaeological heritage.
If one intends to start construction work anywhere, he or she should first check if there is any archaeological heritage in the ground.
How Brandsen Developed AGNES, a Google for Archaeologists
In order to develop AGNES, Brandsen used deep learning and trained a language model to recognize words in archaeological reports, the statement said. An important factor taken into consideration was that the model should be able to recognize synonyms and distinguish between different meanings of a word.
Brandsen said the word bijl can refer to an artifact you can cut things with, but it can also be a last name. He explained that if one searches for the bijl artifact, that is all the person will find and more Mr. Bijls.
The smart search engine can also be used to geographically search archaeological data, which will retrieve information about a user-specified area, according to the release.
How accurate is AGNES?
Brandsen tested the search engine with a colleague and said the latter had been given a database of cremations in the early Middle Ages in the Netherlands by an expert on the period. Brandsen added that the professor spent his whole life collecting this data, but with the search engine, archaeologists found 30% more early medieval cremations. “So you see even an expert doesn’t know everything because there’s so much data,” Brandsen said.
Now a rough version of AGNES is available online, the statement said. AGNES can perform searches with approximately 80% accuracy.
Brandsen will make the search engine more accurate and expand it by also allowing searches in other languages, Leiden University said on its website.