-
Notifications
You must be signed in to change notification settings - Fork 63
Result Post Processing
The result post processing is based on the work of a master project.
They analyzed and interpreted the result of data profiling algorithms.
Therefore, they add some visualizations and statistics to Metanome.
Everything located in de.metanome.backend.result_postprocessing is based on the code from the master project.
After an algorithm execution is done, the results are written to disk.
To show them in the frontend they have to be read again.
This is the first step, which is done in the ResultPostProcessor.
Afterwards, the results may be further analyzed and ranked.
A ResultAnalyzer is responsible for analyzing the results.
In a first step the ResultAnalyzer converts the results into a new result type based on the original one.
The new result type basically holds the old result type and adds some additional statistics to it.
The conversion of the result is done in the ResultAnalyzer.
If the user decides to further analyze the result a ResultRanker is created and the additional statistics are calculated.
Those statistics are then stored in the new result.
By default, Metanome is not using the extended result, because its calculation can take a while.
Having the result, a ResultStore is created.
The ResultStore is responsible for keeping the result of an execution in memory.
Having the result in the ResultStore, the frontend can request it via an api call.