The schema matching process is a fundamental step in a schema integration system, and its quality impacts the overall performance of the system. Recently, a large number of schema matching approaches have been developed. Until today, the performance of schema matching is inherently uncertain and requires improvement. The most difficult task is inferring the realworld semantics of data from the information provided by schema labels in their representations. Usually, schemas with identical semantics are represented by different vocabularies and only their own designers can completely understand. A schema may contain synonyms and homonyms words. Therefore, it is necessary to understand how the schema elements are "presented"; it is often hard to get aware meaning associated with elements names, due to the semantic ambiguity of human language. Semantic ambiguity problem means the capability of being understood in two or more possible senses. Having more than one meaning for an individual schema element would cause confusion in interpretation of schema name. This may affect negatively on the matching result. Therefore, this paper aims to resolve this problem of semantic ambiguity and represent the intended meaning of the schema labels name, by introducing the CKBD (Context Knowledge-Based Disambiguation) approach. The CKBD is obtained by integrating two pieces of context knowledge: semantic domain and more frequency used into a disambiguation processor. Finally, the CKBD is implemented and is tested in a real dataset. The result is deeply grounded in the ability to detect schema name intended meaning.