This article aims to explore the question: understanding, interpreting, translating, how to resolve ambiguity? Or: how does man-machine combination resolve ambiguity? In order to focus on the essence of the problem, the method is that: target analysis butterfly model and its use cases, macroscopic analysis ambiguity model and its use cases, microscopic analysis matrix model or search model within a series of bi-list and its use cases. The result is through the three examples, from manual translation to machine translation and translation memory on view, pointed out that the fundamental way to resolve ambiguity. Its significance is that the method can be advanced to the generalized translation and corresponding interpretation and final practical understanding, the specific performance is that through man-machine collaboration, and its verifiable results with this method, we can work to resolve various ambiguities better, to ensure accurate understand, prevent and eliminate all kinds of misunderstandings.