The ACORN provides an agent-based architecture for information retrieval and provision across networks. The main objective of this paper is to present the design and implementation of information sharing in a community of mobile agents in ACORN, based on keyphrase sharing and comparison among agents. In keyphrase-based information sharing, keyphrases and their weights are used to represent user interests and document contents. Agents use these keyphrase-weight pairs to find other relevant agents. Two similarity measures, the substring indexing method and the Cosine measure method, are compared. The ACORN with Cosine measure is found to be more efficient in terms of meaningful information exchange and execution time.