We were discussing two new operators:
one for the use of additional columns to generate groups that facilitate hierarchical view of entities and another for additional rows to describe new pseudo-entities transformed from exiting entities that is helpful to view entities in the form of layers.
The ability to view the entities as a hierarchical organization let us expand our logic to perform tree-based searches and to assign different roles to entities so that the data may have augmented information to improve the queries. Let us take an example. If we had an employee table that listed all the employees of an organization then additional columns such as manager_level_001_id and manager_level_002_id and so on could essentially provide information that does not repeatedly need to be calculated in subsequent queries. More importantly the addition of columns for groups may be a temporary table so that we don't necessarily write on the original data. Besides if storage is unlimited, it helps to memoize results of processing that is going to be repeated. This now favors a non-recursive tree algorithm to list the managers all the way to the root for any employee as well as the reuse of the logic to determine the next level manager as opposed to recursive queries across levels. This form of data extract, transform and load is popular in data reporting stacks that are used to publish charts and graphs from data.
The ability to view the entities as not just those existing but composites that behave the same as existing entities but also in simpler abstraction of entities purposes different from those involving original entities. We could take another example here. Let us say there was a table of email addresses. We could list different email addresses for the same person and also list different distribution lists that are made up of one or more email addresses. We could also dynamically combine individual and distribution lists into new temporary entries. Combinations may be multi-leveled. Since there is a separation of concern involving the original and the composites from the query, this is considered layering. Since the selection of original entities determine the composites for use in the higher layer, this comes as a different technique from grouping and hierarchies.
one for the use of additional columns to generate groups that facilitate hierarchical view of entities and another for additional rows to describe new pseudo-entities transformed from exiting entities that is helpful to view entities in the form of layers.
The ability to view the entities as a hierarchical organization let us expand our logic to perform tree-based searches and to assign different roles to entities so that the data may have augmented information to improve the queries. Let us take an example. If we had an employee table that listed all the employees of an organization then additional columns such as manager_level_001_id and manager_level_002_id and so on could essentially provide information that does not repeatedly need to be calculated in subsequent queries. More importantly the addition of columns for groups may be a temporary table so that we don't necessarily write on the original data. Besides if storage is unlimited, it helps to memoize results of processing that is going to be repeated. This now favors a non-recursive tree algorithm to list the managers all the way to the root for any employee as well as the reuse of the logic to determine the next level manager as opposed to recursive queries across levels. This form of data extract, transform and load is popular in data reporting stacks that are used to publish charts and graphs from data.
The ability to view the entities as not just those existing but composites that behave the same as existing entities but also in simpler abstraction of entities purposes different from those involving original entities. We could take another example here. Let us say there was a table of email addresses. We could list different email addresses for the same person and also list different distribution lists that are made up of one or more email addresses. We could also dynamically combine individual and distribution lists into new temporary entries. Combinations may be multi-leveled. Since there is a separation of concern involving the original and the composites from the query, this is considered layering. Since the selection of original entities determine the composites for use in the higher layer, this comes as a different technique from grouping and hierarchies.
No comments:
Post a Comment