Research on the identification of generic technology of eco-friendly materials based on text mining.

The overconsumption of natural resources and the production of waste are causing severe degradation of our environment. Generic technologies (GTs) of eco-friendly materials could alleviate environmental pollution, solve resource and environment-related conflicts, and promote society's sustainable development. The identification of GTs is the first step towards GT innovation and establishing a supply of eco-friendly materials; thus, how to accurately identify GTs is an important challenge for governments and enterprises. In this paper, a new method that integrates latent Dirichlet allocation (LDA), term frequency-inverse document frequency (TF-IDF), social network analysis (SNA), and a hidden Markov model (HMM) is proposed for GT identification. The LDA model was employed to extract hidden information of eco-friendly materials. In addition, scientific linkage, betweenness centrality, technology co-occurrence rate, the number of patents, and the number of valid countries or territories designated by the patent were selected to analyse the technology topic characteristics. Then, the fundamentality, pervasiveness, and benefit characteristics of GTs were combined to identify GTs of eco-friendly materials. On this basis, HMM was employed to predict the evolution trend of GTs. The results show that sixteen technologies are GTs of eco-friendly materials. Furthermore, degradable composite materials and cellulose extraction methods will become the focus of research in the future. These studies can provide a new method for the identification of GTs of eco-friendly materials and help nations or enterprises to make effective decisions to develop GTs, minimizing the burden on the environment.