Optimizing Collaboration Decisions in Technological Innovation Through Machine Learning: Identify Trend and Partners in Collaboration-Knowledge Interdependent Networks

Jianyu Zhao,Zhenjie Dong,Xinlin Yao,Xi

DOI: https://doi.org/10.1007/s10479-024-05867-z

IF: 4.82

2024-01-01

Annals of Operations Research

Abstract:Technological innovation, which is exemplified by the convergence of diverse knowledge domains and collaborative efforts across industries, has emerged as the primary catalyst for organizations seeking to enhance their business operations and attain competitive advantages. Consequently, optimizing decision-making processes is imperative for organizations to adeptly forecast technological convergence trends and strategically identify suitable collaboration partners. Since a firm is embedded in both a collaboration network and a knowledge network, it is necessary to consider a dual network structure and the interdependence of the two subnetworks in terms of technological convergence. Regrettably, limited attention has been given to this aspect and the evolving nature of knowledge networks in the existing studies. To address these issues, this study proposes an innovative research framework for technological convergence, combining a dynamic link prediction model using a transformer with the theory of interdependent networks, which enhances predictive algorithms regarding the convergence of new technologies and simultaneously identifies potential collaborative partners. Patent data from the medical device field are collected to conduct experiments. The results indicate that the proposed approach yields a 15% improvement in the area under the curve ( AUC ) metric over that of the recurrent neural network (RNN)-based dynamic model and a gain of more than 25% compared to the machine learning-based static model. Our study has important theoretical and practical implications for improving business operations and innovation strategies.