Using Tasks

Back to Main Help

Click Pipelines > Tasks for the Tasks Page. You can view/edit Task properties and create new Tasks from this page.

The Template column shows the associated Template containing this Task. The Type column shows the Task Type.

n

Task Types

There are 7 Task Types, each of which takes one column and generates one or more Pipeline Columns. Click on a Task Type below for more information.

      Count Task

Counts the matches of a single Pattern or a Term List of Patterns

o   Input: A Text Column

o   Task Parameters: a single Count Pattern or a Term List

o   Other Required Objects: a Term List of patterns (if selected)

o   Result: one Numeric Column

      Extract Task

Extracts matches of either a single Pattern or a Term List of Patterns

o   Input: A Text Column

o   Task Parameters: a single Extract Pattern or a Term List

o   Other Required Objects: a Term List of patterns (if selected)

o   Result: one Text Column

      Edit Task

Performs all substitutions in a Substitution List. Each Substitution has a Pattern and Replacement

o   Input: A Text Column

o   Task Parameters: a single Pattern/Replacement or a Substitution List

o   Other Required Objects: a Substitution List (if selected)

o   Result: one Text Column

      Term Vector Task

Creates a Term Vector column from a Text column, using NLP parameters, such as Drop/Go List, Stemming, Bi-grams.

o   Input: A Text Column

o   Task Parameters: Stem Mode, Bi-Grams, Drop/Go List, Negation Clause Elimination, Tokenizer Regex, Max Count

o   Result: one Term Vector Column

      Lookup Task

Finds each Lookup Term in the Lookup Column of a Lookup Dataset and returns the Target Column value from the same row.

o   Input: A Text Column of Lookup Terms

o   Task Parameters: Lookup Column, Target Column, Lookup Dataset

o   Other Required Objects: a Lookup Dataset

o   Result: one Text Column of Target values

      Rules Task

For each input Text, this task selects one or more Tags, based on highest Tag Scores calculated by applying a Ruleset containing Rules of the form (Pattern, Tag, Score).

o   Input: A Text Column

o   Task Parameters: Ruleset name, Tags-Display parameters (Score Threshold, Single/Multi-Tag, Null Tags)

o   Other Required Objects: a Tagset, a Ruleset

o   Results: Up to 4 columns:

  Tag: Text

  Count: Numeric (integer)

  Matches: Text

  Score Total: Numeric (double).

      ML Task

For each input Term Vector, chooses one or more Tags, based on highest Tag Confidence values calculated as a similarity measure between the Term Vector word counts and a Trained Word Cloud for each Tag stored in a previously trained ML Model.

o   Input: A Term Vector Column

o   Task Parameters: ML Model name, Tags-Display parameters (Confidence Threshold, Single/Multi-Tag, Null Tags)

o   Other Required Objects: a Tagset, a trained ML Model

o   Results: Up to 2 columns:

  Tag: Text

  Confidence: Numeric (double).

Creating a Task

Click Create Task button

      Enter a Name and Description

      Choose a Template to which this task will be added

      Choose an Input Column for this task

      Select Task-specific parameters (see individual task help pages)

      Check Change Output Columns? to change the output Pipeline column names and their display status.

      Check Update Immediately? to run an Update Job to generate new column values after creating this task.

Next: Updating a Dataset