Computer systems becoming better at cataloging science
Computers are quickly catching up to, and perhaps even surpassing, humans in the field of scientific data extraction and assessment, according to a Monday university news release.
By searching for key words and sifting through the myriad of articles on a given topic, advanced computer systems are now able to draw connections between terms and catalog them in databases.
PaleoDeepDive, a machine reading system project led by current UW-Madison computer science professor Miron Livny and former UW-Madison computer science professor Christopher Ré, imitates human cataloging techniques to enter data into the Paleobiology Database.
The approach PaleoDeepDive takes to catalog the data is “to look at the entire problem of extraction as a probabilistic problem,” Ré said in the release. Therefore, though it cannot read context like humans can, it can connect different terms based on the likelihood they are related.
The ultimate hope is that one day, a machine that can quickly read and synthesize related facts will be created to tackle big scientific questions.Subscribe to The Daily Cardinal Newsletter