Who is Olivier Danvy? He is the most thanked person in computer science, according to an automated analysis of almost half-a-million acknowledgements in science papers.
The technique, which relies on text-mining software, offers a new way to investigate the influence of individuals, research agencies and companies within different areas of science.
Until now, research on acknowledgements has been hampered by the lack of a data repository to study. In contrast, citation statistics (which track which papers are referenced by others) have been manually compiled over the past four decades by the company Thomson ISI in Philadelphia.
Lee Giles, a computer scientist at Pennsylvania State University in University Park, Philadelphia, and his colleagues have now developed software that automatically extracts information on who is thanking who from science papers. They used it to analyse 335,000 papers within the lab's CiteSeer archive of computer-science papers to produce rankings of the most thanked funding agencies and individuals in computer science.
Something that once seemed like it would require an enormous amount of manual labour suddenly seems feasible.
Jon Kleinberg Cornell University, Ithaca
The work opens up a largely untapped mine of information, says Jon Kleinberg, a computer scientist at Cornell University, Ithaca. "Something that once seemed like it would require an enormous amount of manual labour suddenly seems feasible," he says.
Due thanks
Danvy, a French researcher who works on programming languages at the University of Aarhus, Denmark, says that at first he was "stunned" to find his name at the top of the list.
But on reflection, he puts it down to "a series of coincidences". He is multidisciplinary, well-travelled, is involved with an international PhD programme, and belongs to a university department that encourages international visitors.
"It's a snowball effect," says Danvy, who admits to being a helpful sort of fellow. "I encourage people a lot, and advise many students on their papers."
Acknowledgement analysis could allow this type of inspirational or motivational person to be more routinely appreciated, says Rémi Barré, former head of the French science-assessment agency, Observatoire des Sciences et des Techniques. "This could bring a welcome recognition of the role," he says.
Acknowledged problems
Experts are divided over the significance of the new approach. Acknowledgements are more slippery than citations - there are many reasons to acknowledge somebody, different countries and fields have different traditions of who they acknowledge, and there is no requirement to mention everyone who deserves recognition.
"Peer review requires you to incorporate the latest and relevant citations. However no one is checking whether all contributors are referenced in the acknowledgements," points out Erik van Mulligen, chief technology officer of Collexis, a Dutch text-mining firm.
I was stunned to find my name at the top of the list.
Oliver Danvy University of Aarhus, Denmark
Social scientists have already tackled some of these problems. They have divided acknowledgements into six main types, including support from funders, and the, perhaps more tantalizing, Danvy-type 'conceptual' support of individuals.
This type of categorization is essential in any analysis of acknowledgements, insists Eugene Garfield, the founder of ISI. "Otherwise you'll get a mishmash," he says.
Much obliged
ADVERTISMENT
Bill Valdez, director of the Office of Planning & Analysis at the US Department of Energy's Office of Science, says that when it comes to looking at the influence of different funding agencies in various fields, Giles's work is an "important breakthrough".
Meanwhile Chris Rosin, chief executive of Parity Computing in San Diego, sees the technique as a powerful way to trace social networks among scientific collaborators, and to identify clusters of related research, something that researchers already try to do by looking at the groups of authors on different papers.
"Accurate acknowledgement data for individual scientists would add a useful layer of informal relationships to this social network," Rosin says. "This would identify connections between groups of collaborators who might otherwise appear isolated."
Something that once seemed like it would require an enormous amount of manual labour suddenly seems feasible. 
