An exploration method for technology forecasting that combines link prediction with graph embedding: A case study on blockchain

【Author】 Wang, Liang; Li, Munan

【Source】TECHNOLOGICAL FORECASTING AND SOCIAL CHANGE

【影响因子】10.884

【Abstract】To keep pace with the latest technological changes and advancements, predicting future technological trends and topics has become a critical approach for high-tech companies and policy-making institutions. In this paper, we proposed an explorative method that integrates link prediction and Node2Vec graph embedding to predict future technology topics using co-occurrence data from patent keywords. Specifically, this method collects and preprocesses patent datasets, constructs network graphs that depict relationships among different technology topics, and builds a supervised link prediction model based on the time series of the graph to identify future technology graphs. Furthermore, node2vec graph embedding is conducted to obtain node vector representations, and then the clustering algorithms can be improved to identify the relevant topics, which could be interpreted as future technology. A case study on blockchain is conducted to validate the feasibility and practicality of the method to demonstrate the application of the method. Through the comparison of machine learning methods, we selected the Random Forest (RF) model, which presents the highest accuracy, for our experiments. The results show that the proposed method can be used to effectively visualize potential future topics related to a specific technology. Compared to traditional methods such as Latent Dirichlet Allocation (LDA), our method can identify more unique and differentiated technological topics, significantly reducing topic overlap. Additionally, the reported method can illustrate the internal relationships of topics through subgraphs, helping readers better understand the core concepts of each topic and vividly displaying the structure and composition of the topics. Furthermore, the proposed method can also depict potential relationships between different technology topics, which can facilitate the visualization of new directions of research and development.

【Keywords】Technology forecasting; Topic recognition; Link prediction; Graph representation learning; Emerging technology; Blockchain

【发表时间】2024 NOV

【收录时间】2024-09-23

【文献类型】案例研究

【主题类别】

区块链治理-市场治理-链上信息预测