This paper proposes new approach based on the breakdown of metadata repository into taxonomy based metadata classifiers to classify the information. Due to the raising quality issues, it results in avoiding metadata from being processed correctly. The inconsistent metadata makes it difficult to locate relevant information. In the multitier architecture of data warehousing, there is a need to break metadata repository to handle the information. From warehouse, information like data names and definitions of that warehouse is marked by metadata. The reason for construction of metadata is also discussed. Information like data warehouse structure, operational metadata, algorithms, mapping, system performance related data and business metadata are contained by the repository. This storage of information and management should be persistent. This approach will split the heavily populated data warehouse into data marts to control and manage data in immensity which results in controlling of time consuming and slow working. New method is introduced here based on dividing the metadata repository into data marts. This paper is discussed as follows. First part is the introduction of the metadata and taxonomies. In the second part, need of breaking metadata repository into data marts is discussed. Statistical framework from metadata repository's point of view is delineated in third part of this paper. How data warehouse is managed and its components are conversed in next section. Methodology is conferred as steps in implementing a data mart. After the related work, conclusion and future directions are given at the last part of the paper.

Data warehouse Data marts Metadata repository Taxonomy based data marts Taxonomy based metadata classifier