We introduce graph Transformer into text representation learning and apply it to the ABSA task. To tightly combine the 2 text learning processes in graph view and sequence view, we propose a novel ...