By analyzing language and syntax used online, the new technology can reveal if an adult is masquerading as a child as part of the victim "grooming" process.
The scheme, known as Project Isis, will also be able to keep track of secret code words used by pedophiles as file names for child pornography.
"The main technique is something called authorship attribution. Research has been done which indicates that there are differences in how people of a particular age group write," the Telegraph quoted Professor Awais Rashid, of Lancaster University, as saying.
"You can distinguish when someone is 25 when they are claiming to be 14, for example.
"The project uses a lot of artificial intelligence and a lot of algorithms. We are using language analysis tools to identify someone who is masquerading as a child, and therefore identify people who may be a risk to children.
"We are looking at being able to monitor traffic in file-sharing networks to try to identify core distributors, who are of interest to law enforcement agencies because they have access to children and are preparing photographs of these children in abusive situations.
"Pedophiles use very specific ways of marking these files and searching for them.
"To the untrained eye they can look like innocent searches but with our analysis you can isolate them and study how they change," Rashid added.
A first prototype system has already being devised and is being tested on non-sensitive data.
The next step of the three-year project will be to use the software on real-life pedophile material provided by police and other agencies.
If it proves effective, Project Isis could raise Internet privacy concerns. For it to work in practice, it would require a wide range of Internet sites such as chat rooms and peer-to-peer networks to be monitored for tell-tale pedophile language.
"We are setting up a stakeholder ethics group of internet service providers, users and other groups who may have concerns about the ethics of monitoring," Rashid said.