The objective of this paper is to propose a cooperative query answering approach that relaxes query conditions and provides approximate answers by utilizing semantic relationships between data values. The proposed fuzzy abstraction hierarchy (FAH) adopts the integrated notion of data abstraction and fuzzy relation to represent similarity relationship. Also, to be tolerant of incorrect input for queries, FAH refers to a thesaurus for synonym information about database token. For query processing based on FAH, query relaxation operators such as query generalization, approximation, and specialization are developed. Compared with existing approaches, FAH supports more effective information retrieval by processing various kinds of cooperative queries through elaborate relaxation control, and by providing ranked query results according to fitness scores.