26 October 2019 | New Scientist | 15Machine learning Archaeology
Gege Li Paul Cooper
ARTIFICIAL intelligence is learning
to decipher damaged ancient Greek
engravings. The AI is better than
humans at filling in missing words,
but may be most useful as a
collaborative tool for narrowing
down the options.
Many ancient inscriptions have
become eroded or damaged over
the centuries, resulting in segments
of text being lost. Figuring out what
was once in the gaps is a tricky task
that involves looking at the rest of
the inscription and similar texts.
Yannis Assael at DeepMind
and his colleagues trained a neural
network, a type of AI algorithm, to
guess missing words or characters
from Greek inscriptions, on surfaces
including stone, ceramic and metal,
that were 1500 to 2600 years old.
The AI, called Pythia, learned to
recognise patterns in 35,000 relics,
containing more than 3 million
words. The patterns it picks up
on include the context in which
different words appear, the
grammar and also the shape
and layout of the inscriptions.
Given an inscription with
missing information, Pythia
provides 20 different suggestions
that could plug the gap, with the
idea that someone could then
select the best one using their own
judgement and subject knowledge.
“It’s all about how we can help the
experts,” says Assael.
The team tested the system in
a head-to-head against humans.
When filling in text for almost
3000 damaged inscriptions, human
experts made 30 per cent more
mistakes than the AI (arxiv.org/
abs/1910.06262).
This shows the potential of
AI-assisted restoration, says Thea
Sommerschield at the University of
Oxford, who was part of the team.
“The reward is huge because it tells
us about almost every aspect of the
religious, social and economic life
of the ancient world,” she says. ❚
AI beats humans at
filling in damaged
ancient inscriptions
ARCHAEOLOGISTS are
endeavouring to find a lost
Sumerian city that was once
home to a mysterious ancient
cult before it is emptied of its
artefacts. In recent years,
thousands of clay tablets have
been rescued from smugglers,
suggesting that looters must
have discovered the site
somewhere in Iraq.
Irisaĝrig reached its zenith
about 4000 years ago. “It was
the capital of a major province
of the Neo-Sumerian state,” says
Manuel Molina at the Spanish
National Research Council in
Madrid, who has spent much
of his life searching for it.
The city was frequently
visited by Sumerian kings.
One set of tablets recovered
from smugglers tells the story
of a woman from Irisaĝrig called
Ninsaga, who managed her own
large estate. This changed
historians’ views of the role
of Sumerian women. The city
is also thought to have been
the centre of a cult that
worshipped the Sumerian
mother goddess Ninhursag.
If the city were discovered, it
could reveal much more about
how these ancient people lived,and the beliefs they held about
their mother goddess, says
Eckart Frahm at Yale University.
“Perhaps even some new
mythological texts related
to her would be discovered.”
Irisaĝrig was probably
abandoned during a period
of social collapse along with
many other cities. As a result,
southern Iraq is dotted with
ruin-mounds, known as tells,
where towns once stood.
Irisaĝrig could be any of them.Over the past few years,
many more tablets that seem
to have come from Irisaĝrig
have turned up. In August,
the British Museum returned
to Iraq a set of 156 cuneiform
tablets that had been
confiscated from smugglers at
London’s Heathrow airport.
Molina first tried to figure
out the location of the city in
the late 2000s, based on a text
discovered through excavations
in the ruins of the city of Umma.
The text records a journey toIrisaĝrig from Umma taken
four millennia ago by a group
of workers. They went on a
round trip of 23 days, towing
a barge upstream, loading it
with barley and floating it back.
Molina retraced this journey
using satellite images and
landed at a location known as
Site 1056, a heavily looted tell
73 kilometres north of Umma
on the banks of the river Tigris.
Now Maurizio Viano at the
University of Turin in Italy
has taken this strategy further.
“I was not convinced by
Molina’s reconstruction of
the direction of the current
along the canals,” he says.
In 2014, an expedition
discovered inscriptions that
allowed Viano to pinpoint the
location of another ancient lost
city called Keš, which is known
to have been close to Irisaĝrig.
Using this information, he tried
to plot the route of the boat trip.
He concluded that the lost
city wasn’t on the Tigris, but
connected to it by a now-
dried-up 30-kilometre canal
known as the Mama-šarrat.
When Viano followed this
vanished watercourse, it led
to an enormous ruin-mound
named Tell al-Wilaya (Journal
of Cuneiform Studies, doi.org/
dcvs).
“There is sufficient reason
to suggest the identification
of Irisaĝrig with the site of
Tell al-Wilaya,” he says.
No excavations are planned
at the moment due to the
unstable situation in Iraq.
But Frahm remains hopeful.
Maybe Irisaĝrig has now been
found, he says. “Maybe it really
is Tell al-Wilaya. It is exciting.” ❚The race to find the
lost city of Irisaĝrig
TRUSTEES OF THE BRITISH MUSEUMCuneiform tablets
taken from smugglers
at Heathrow“ A text records a boat
journey to Irisaĝrig
taken by workmen
four millennia ago”