Affiliation:
1. Jiangsu University of Science and Technology
Abstract
Abstract
This paper amis to provide support for the prediction and recommendation of technological innovation cooperation relations. A patent-based offshore engineering equipment technological innovation cooperation network is built at first. The trend of network predictability and its upper limit are explored through the normalized shortest compression length of the network structure, and the actual prediction accuracy of classical link prediction methods is compared with the upper limit. Results show that the change of normalized shortest compression length of the network structure is fully consistent with the development trend of the link predictability of offshore engineering equipment technological innovation cooperation network, and the normalized shortest compression length can also be used as the basis for quantitative characterization of the upper limit of network link predictability. Futhermore, the prediction accuracy of the classical link prediction methods is about 30% lower than the upper limit calculated according to the normalized shortest compression length.
Publisher
Research Square Platform LLC