Overview of predefined tasks
overview.RmdBefore creating your own tasks, you might want to explore the
predefined tasks available in the quallmer package. These
tasks are designed to streamline common annotation needs and can be
easily integrated into your workflow.
Predefined tasks
The quallmer package currently includes the following
predefined tasks:
| Task | Function | Description | Output |
|---|---|---|---|
| Sentiment analysis | task_sentiment() |
Rates overall sentiment of text. | Sentiment score (-1 to 1) and
explanation |
| Stance detection | task_stance() |
Identifies stance/position taking toward a topic. | Stance label (pro/neutral/contra) and explanation |
| Ideological scaling | task_ideology() |
Places text on an ideological scale (0–10) for a specified dimension. | Ideological score (0 to 10) and
explanation |
| Topic salience | task_salience() |
Ranks salience of specified topics in text. | Ranked list of topics and explanation |
| Fact-checking | task_fact_check() |
Verifies factual claims in text. | Truthfulness score (0-10, with 10 signalling highest confidence in truthfulness), misleading topics (if any) and explanation |
If you wish to create custom tasks tailored to your specific research
questions and data types, you can use the task() function.
This function allows you to specify the system prompt and output
structure, enabling you to customize the annotation process according to
your needs. For a tutorial on how to create custom tasks, please refer
to our custom
tasks tutorial.