Software to solve handwritten riddles

Researchers at the University at Buffalo, part of the State University of New York, have developed a computer program said to be 98 per cent percent effective in determining authorship of handwritten documents.

Funded by the US National Institute of Justice, it’s believed to be the first software program designed to develop computer-assisted handwriting analysis tools for forensic applications.

In criminal cases, human handwriting analysts solve the question of who penned a ransom note or forged a check. But because they are human, even the best graphologists cannot claim complete objectivity.

The UB software is the first that can identify who wrote a particular document based on purely scientific criteria.

‘A human expert may put in his or her own bias even unconsciously,’ said Sargur Srihari, Ph.D., principal investigator and SUNY Distinguished Professor in the Department of Computer Science and Engineering in the College of Arts and Sciences and School of Engineering and Applied Sciences at UB. ‘The idea that everyone’s handwriting is different is taken for granted. What we have done is to develop purely scientific criteria for that premise.’

It is the first time researchers have attempted to do that based on a large database of handwriting and by using a totally automated means of measuring specific features of human handwriting, said Srihari.

CEDAR is the world’s largest university-based research centre devoted to new technologies that can recognise and read handwriting.

Providing a scientific basis for establishing the individuality of handwriting has become essential for admitting handwriting evidence in U.S. courts due to a number of recent rulings concerning expert testimony said Srihari.

The UB researchers developed the software by first collecting a database of more than 1,000 samples of handwriting from a pool of individuals representing a cross section of the U.S. population.

Multiple samples of handwriting were taken from subjects, each of whom was asked to write the same series of documents in cursive.

Instead of analysing the documents visually, the way a human expert would, Srihari explained, the researchers deconstructed each sample, extracting features from the writing, such as measuring the shapes of individual characters, descenders, and the spaces between lines and words.

The researchers then ran the samples through their software program.

‘We tested the program by asking it to determine which of two authors wrote a particular sample, based on measurable features,’ said Srihari. ‘The program responded correctly 98 percent of the time.’

Srihari explained that human experts look for arcades and garlands, features that may distinguish one person’s penmanship from another’s.

The current software should be able to conduct that type of advanced analysis within the year, he concluded.