A cognition graph approach for insights generation from event sequences

In recent years, cognition map techniques for human insights have already played a significant part in complex or ill-structured problem solving. There are increasing interests on computational methods rather than hand-drawing methods to build an cognition graph for insights generation. In this paper, a systematic approach called Temporal-IdeaGraph is proposed to build a directed cognition graph based on event sequences. Firstly, an algorithm of frequent sequence mining is employed to capture sequential patterns and a method is then designed to remove duplicate patterns. Secondly, relevant patterns are merged and visualized into a directed cognition graph. An algorithm is further proposed to identify bridge events and bridge patterns which would trigger human’s deeper insights for better decision making. Finally, two real case studies validate the effectiveness of proposed approach.