Patent trawler aims to predict next hot technologies

July 9, 2012 | Source: New Scientist
cluster_structure_patents_citations

Two-dimensional representation of patent similarity structure in National Bureau of Economic Research subcategory 11 (Agriculture, Food, Textiles). The colors encode the U.S. patent classes: red corresponds to class 8; green: 19; blue: 71; magenta: 127; yellow: 442; cyan: 504. (credit: Péter Érdi et al.)

A new tool automatically helps forecast emerging technologies, thanks to an innovative data-mining technique.

Developed by Péter Érdi at the Hungarian Academy of Sciences in Budapest and colleagues, it works by analyzing the frequency with which prior-art (previous patents) are cited by other patents.

Plotting how the frequency of these citations changes over time shows that patents can be grouped into related clusters. These clusters evolve, sometimes branching into new disciplines, sometimes merging with one another.

Érdi’s team has written software that charts this evolution and helps predict whether existing technological fields can combine or diverge to create new areas of innovation.