With the rapid development of garments recommendation systems, more and more garment knowledge base have been widely developed. The research in this paper aims to build a garment knowledge base in order to help general consumers to identify the most relevant products satisfying their specific requirements. We design four experiments for building this knowledge base by 8 pairs of normalized sensory evaluation criteria for describing both consumers’ expectations and product profile. The theory of fuzzy composition technology is applied for setting up garment knowledge base which can be used for consumer-oriented intelligent garment recommendation system. Compared with the other knowledge base, this knowledge base is more robust and more interpretable owing to its capacity of handling vague, imprecise, uncertain, or ambiguous.