What are the methods for the generation of concept hierarchies for nominal data?

Data MiningDatabaseData Structure

There are various methods for the generation of concept hierarchies for nominal data as follows −

Specification of a partial ordering of attributes explicitly at the schema level by users or professionals − Concept hierarchies for nominal attributes or dimensions generally contains a set of attributes. A user or professionals can simply represent a concept hierarchy by defining a partial or total governing of the attributes at the schema level.

For instance, suppose that a relational database includes the following set of attributes such as street, city, province or state, and country. A data warehouse location dimension can include the same attributes. A hierarchy can be represented by describing the total ordering between these attributes at the schema level including street < city < province or state < country.

Specification of a portion of a hierarchy by explicit data grouping − This is basically the manual description of a portion of a concept hierarchy. In a huge database, it is unrealistic to describe a whole concept hierarchy by explicit value enumeration.

Specification of a set of attributes, but not of their partial ordering − A user can define a set of attributes forming a concept hierarchy, but exclude to explicitly state their partial ordering. The system can attempt to automatically create the attribute ordering so as to make a significant concept hierarchy.

Consider the observation that because higher-level concepts usually cover multiple subordinate lower-level concepts, an attribute describing a high concept level (e.g., country) will generally include a smaller number of distinct values than an attribute describing a lower concept level (e.g., street).

It depends on this observation, a concept hierarchy can be automatically created based on the multiple distinct values per attribute in the given attribute set. The attribute with the most distinct values is located at the lowest hierarchy level.

The lower the multiple distinct values an attribute has, the larger it is in the generated concept hierarchy. This heuristic rule operates well in several cases. Some local-level swapping or adaptations can be used by users or experts, when essential, after the analysis of the generated hierarchy.

Specification of only a partial set of attributes − Sometimes a user can be inaccurate when describing a hierarchy, or have only a vague concept about what must be contained in a hierarchy. Consequently, the user can included only a small subset of the relevant attributes in the hierarchy description.

It can manage such partially specified hierarchies, it is essential to embed information semantics in the database design so that attributes with fast semantic links can be pinned together. In this method, the representation of one attribute can trigger an entire group of semantically tightly connected attributes to be “dragged in” to design a complete hierarchy. Users should have the option to reverse this feature, as essential.

Updated on 18-Feb-2022 07:11:31