Ancient excrement is an important source of information for archaeological research. It can provide information about people’s identity, diet, health status and so on. However, dog excrement similar in size and shape to human excrement is also common in many archaeological sites, which often puzzles researchers. “It’s hard to distinguish them,” said Maxim Boli of the Max Planck Institute of human history and Science in Germany
As we all know, dogs may eat human and other animal feces, so human DNA information will appear in dog feces; In ancient times, many places had the habit of eating dog meat, so human feces often contained dog DNA information.
In order to solve this problem, using the existing data of human and canine microorganisms, the researchers trained a machine learning system called coproid (ancient fecal identification). The coproid system can enhance its recognition ability by “feeding” a large number of human and dog DNA, as well as the data of modern human feces and dog feces such as microorganisms and plants in the intestine.
After machine learning, the coproid system successfully identified 7 non fecal control samples, accurately determined 5 human fecal samples, and even identified a dog feces that archaeologists originally thought was human feces. Lisa Marie Shillito of the University of Newcastle in the UK said: “this is a great study, but there are some clear problems to be solved before the method is widely used.”