Researchers from the University of Granada have developed a new computer system that allows predicting what sales a book will have if it is published , very useful information for publishers, which could thus decide the print run of a work. The new tool, called PreTEL , is based on artificial neural networks .
To develop it, Pedro Ángel Castillo Valdivieso and his colleagues from the GeNeura research group previously trained the system considering data from thousands of books already published and thus obtaining a mathematical model for estimating data that is capable of interpreting the values of a new book. to estimate the optimal circulation and the sale.
Methods based on artificial intelligence – such as artificial neural networks, logistic regression models, decision trees, and ARIMA-type models – have the capacity to learn and adapt , "in such a way that if in the future they new data is available, they can be retrained to correct themselves and improve the prediction results ", points out Castillo Valdivieso. Thus, the larger the knowledge base in terms of the number of books contained, the better the model that makes the prediction is expected to be, and the higher the quality of the estimates.
