Researchers from Bar-Ilan University have built a computer program that makes it possible to identify the gender of an unknown writer. The success of the identification - five out of six times * The research will be published in the coming days in a journal published by the University of Oxford
Avi Blizovsky
Direct link to this page: https://www.hayadan.org.il/aibarilan.html
Researchers from Bar-Ilan University succeeded in building a computer program that can identify the gender of the author of a particular article or book.
Prof. Moshe Kopel from the Computer Science Department at Bar-Ilan University and Prof. Shlomo Argman (now at the Illinois Institute of Technology-Chicago), together with the linguist Prof. Jonathan Payne (from the English Department at Bar-Ilan University) and the student Anat Shimoni, report that the use Savvy with linguistic signals, it allows to identify the gender of the writer of an article in English, with a success rate of five out of six attempts.
The new discoveries will be published this week in the journal Literary and Linguistic Computing, published by Oxford University Press.
Various techniques from an area of computer science known as "machine learning" are used to program a computer to distinguish between male and female writing samples.
The computer is programmed to learn by itself how to differentiate between these spellings based on statistical determinations it finds in the examples. After the computer learns from these examples, it applies the things to other documents it has not seen before.
The books of writers such as Deborah Tannen and John Gray popularized the idea that women and men have different speaking styles. The new discoveries show that gender differences even extend to writing styles in books and articles.
This is surprising because unlike participating in a conversation, in writing books and articles there is no direct social interaction.
Continue on p. 2…
The research at Bar-Ilan was done on about 600 books and articles taken from the British National Corpus - a large collection of written texts collected for the purpose of linguistic research. The researchers found that the same differences between writers of both sexes existed across the wide range of subjects examined, including art, politics, science, biographies, and more.
In an article to be published this summer in the prestigious journal, Text, the authors show that in writing fiction and non-fiction, women use in their writing words that mark the relationship between the writer of the book and the reader more than men, while men tend to use words that describe and quantify things more than women. The reasons for these differences are still unclear.
One potential use of these findings is in the field of forensics.
Donald Foster of Vassar College in the USA, an expert in identifying anonymous authors, identified the author Joe Klein as the author of the book - Primary Colors, by comparing the style of this book with known writings of the same author.
However, it should be remembered that usually, law enforcement researchers, who are trying to identify an anonymous writer of some document, do not have a selection of written examples like the poster had.
In these cases, the researchers will be satisfied with assembling a profile of the writer according to his gender, age, linguistic, educational background, etc.
The team of researchers from Bar-Ilan says that the automatic techniques that led to the solution of the problem of identifying the gender of the anonymous writer, could also be used to identify other demographic characteristics.
Knowledgeable in robotics and artificial intelligence
https://www.hayadan.org.il/BuildaGate4/general2/data_card.php?Cat=~~~500870851~~~207&SiteName=hayadan
