A Hybrid Approach to Identify and Forecast Technological Opportunities based on Topic Modeling and Sentiment Analysis

This study proposes a hybrid approach to recognize both technological topics and application topics by mining intelligence separately from the different parts of Derwent patent document. Topic modeling method is introduced to recognize topics from patents while sentiment analysis combined with traditional bibliometric indicators are introduced to judge the value of topics from multi-aspect. The hybrid approach is demonstrated by a case study on dye sensitized solar cells. The main contributions of this study include three-fold. First, we explore both technical innovative opportunities and application opportunities by mining different parts of Derwent patent document. Second, we integrate sentiment analysis and bibliometric indicators to judge the value of topics from multi-aspect. Third, we propose a probability-based topic relation measurement method to identify the relationships of the applications with the core sub-technologies.