Disruptive Technology Identification Based on News Influence and Enhanced Graph Attention Network Clustering Method
[Purpose/Significance]Many existing studies rely on patent data to identify disruptive technolo-gies,but these methods still have potential for further optimization in terms of topic clustering analysis of patent texts.[Method/Process]By constructing a graph attention network based on the enhancement of news influence and adaptive allocation of attention weights,the node relationship of the co-occurrence network of technical subject words was effectively captured and fully utilized.After generating representative node vectors,it conducted the-matic clustering analysis of patent text,which could further assist in identifying potential disruptive technologies.[Result/Conclusion]In order to further verify the effectiveness of the method,it selects two emerging technology fields of smart city and industrial Internet for empirical testing.The theoretical and empirical analysis shows that this clustering method of graph attention network,which integrates the influence of news,can further enrich the current methodological system on the identification of disruptive technologies.
万方数据
维普
