MIT researchers use artificial intelligence to make it easier for scientists to keep track of their work
MIT researchers have used artificial intelligence techniques to make the process of tracking research more efficient and less error-prone.
The researchers say their new system, called a probabilistic analytics system, is more accurate and more efficient than current methods, including those used by academic journals.
The technology uses statistical algorithms to track the performance of a small set of researchers on their own and those of their collaborators and peers.
It can be used to help identify errors, identify problems with research and recommend action.
It has the potential to make research more transparent, easier to manage and more cost-effective.
“We are using machine learning to automate the task of tracking what people do,” said Professor James Pomerantz, a professor of computer science at MIT.
“The system learns what a researcher is doing and how they are doing it and uses that information to give a more complete picture of what they are looking at.”
For instance, a machine learning algorithm might look at a paper that is published by a research group and suggest a different approach.
If that is done by a collaborator, the system can then make a recommendation to the journal, which will publish that paper in the journal.
The system can also be used by an academic or a professional who wants to keep a track of how their research is being done.
This could be a colleague, a student or even a researcher themselves.
“It’s very difficult to track down what a research team is doing, especially as they have different teams,” said co-author Dr. Daniel Bielen, a postdoctoral fellow at MIT and an assistant professor in MIT’s Department of Computer Science and Artificial Intelligence.
“We have to know where their work is going and what the challenges are, so that we can be able to give advice and make decisions to the team.”
To be sure, the probabilistics system doesn’t provide all the answers.
For instance, it doesn’t predict whether an author is likely to publish their results in a prestigious journal.
But the researchers say it is an effective way to track research progress and to ensure that researchers are doing their work correctly.
“This system can be applied to anything,” said Bielens.
“You can use it to check a report, or if a researcher gets sick and needs to take a leave, you can use the system to check whether the person who’s sick is actually in the lab or in a clinical setting.
It’s a very flexible way to use machine learning.”
The MIT researchers said they expect the system will become even more accurate as it matures and they add more features to it.