0 ×

Modelflow Configuration

KNeval version 0.1.0 by University of Konstanz

Configuration of a model workflow. This node allows one to specify certain criteria that determine how other workflows will be evaluated against the one that contains this node. In order to evaluate workflows, set up a model workflow containing this node, configure the node accordingly, and run "Evaluate..." in the context menu of the KNIME Explorer on multiple selected workflows (including the model workflow). Alternatively, the evaluation on exported workflows (zip-files) can be performed by using the "Evaluation"-node.

Node configuration: basically, certain nodes in the model workflow can be marked as "required", meaning that they are expected to be present in the workflows that are to be evaluated. Furthermore, the node settings as well as the node ports of "required" nodes can further be modeled (see below).

Options

Configuration

Nodes
The list of all nodes that are part of this workflow. Here it can be specified whether a node is regarded as "required" and a certain amount of credits can be set. The credits help to determine a final score as an evaluation result (e.g. a number of achieved credits out of a maximum number of credits). Only the credits of nodes that are required and present in the workflow to be evaluated are summed up.

Nodes that are marked as "required" can further be modeled (e.g. their settings and ports), see "Model Node Settings" and "Model Node Ports".
Additional Options
Node Insertion Penalty : The specified value will subtracted from the overall score for each additionally inserted node (i.e. for all nodes that are present in the test-workflow but not in this model-workflow).
Model Node Settings
Allows one to model the node settings of a given node. The model settings can either be newly created ("New Entry") or added from the list of all available node settings for this particular node ("Add"). Only the settings that are selected for evaluation (upper table) are used. The others are ignored.
Settings Key : The key of the settings that will be evaluated. If left empty, it will be ignored for matching.
Parent Keys : The parent keys of nested settings ("/key1/key2"). If specified, the parent keys have to match as well. If left empty, the parent keys will be ignored entirely.
Type : The type of the settings (e.g. xstring).
Specification : Specifies how the settings value of a workflow to be evaluated has to match the given settings. It can either be
  • a crisp value that is compared directly
  • a regular expression (in case of a xstring-type), see https://docs.oracle.com/javase/7/docs/api/java/util/regex/Pattern.html , or
  • a list of numbers and intervals (in case of type xdouble, xfloat, xint, xshort, or xbyte). The numbers are separated by comma, the intervals are specified by '[' and ']'. Example: "1,3,4,[5,10]" - the values 1, 3, 4 and 5 to 10 are allowed.
Model Node Ports
Allows one to specify whether and how node ports are to be matched during the evaluation with the according ports in the model workflows. So far only data tables are available for configuration. Every option can also be provided with a number of credits that determines to what extend it will contribute to the final evaluation score.
NOTE : The configuration of model ports is only possible if the port specification is available. That is, some nodes might have to be executed before their model ports can be configured.
For every in- and outport table the following options are available:
  • Ignore : the respective table is ignored during the evaluation
  • Match Entire Table: The tables of the model- and test workflow at the respective port are only considered as match if the two tables match exactly (i.e. every single data cell)
  • Match Properties: Column count and/or row count of both tables are compared. If equal, it is considered as a match.
  • Match Table Columns: A more fine-grained modeling. For individual table columns it can be specified whether and how they should match. If "Match Name" is checked, the column name of the respective column (of the model and test table) have to match. If "Match Type" is checked, the data cell type (e.g. String) and the column name have to match. Additionally, the actual content of a individual column can be used for matching (e.g. by exactly comparing the values of the model- and test column). Please refer to the documentation of the "Column Content Match"-configuration (can be opened by clicking into the "Match Content"-column for the respective table column).

Credits and Hints

Summary Tree
A tree that displays the summary of all set configurations for the nodes that are marked as 'required'. Additionally the credits (in case of a match) and a hint (in case of a mismatch) can be adjusted and set.

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.