Fusion-knowledge metric based on ontology is presented to control the scale of new knowledge causing by knowledge fusion. The compact degree between knowledge units is analyzed by using tacit relationship strength. Semantic relevancy is formulated based on the construction rule of lexical chains, and calculated with the semantic distance between ontology concepts. The semantic entropy is analyzed by using the maximum entropy model. The utility weight is studied by
analyzing the effect of attribute value on fusion-knowledge. On the basis of the above analysis, the fusion-knowledge metric is formulated to guide the design of the knowledge fusion algorithm, and some properties of the fusion-knowledge metric, such as symmetry, determinacy, non-negativity and expansibility, are studied. Finally, the effectiveness of fusion-knowledge metric is demonstrated by an illustrative example, and the important effect of the fusion-knowledge metric on the knowledge evaluation mechanism is discussed.