Prompting technique

Feature-Based Prompting (Structured Analysis)

Estimating Item Difficulty Using Large Language Models and Tree-Based Machine Learning Algorithms, arXiv:2504.08804 (2025) · read the paper

What it does

Instead of directly asking for evaluation or solution, the model first analyzes the task according to predetermined specific criteria. Decomposition of a complex request into a series of simple analytical questions turns subjective guessing into structured objective analysis, significantly improving accuracy and reliability of results.

Task typemulti step
Shapechain of thought
Efforttwo or three steps
Addressesanswers that are close but wrong; the same prompt giving different answers; invented facts and citations
Published2025-04-09
AuthorsPooya Razavi, Sonya Powers

When it helps

Apply this technique when you need accurate evaluation or analysis of complex objects: assessing content quality, task complexity, information relevance, classification by multiple criteria. Particularly effective for tasks where direct requests yield subjective results. Use when reproducibility and objectivity of assessments are important. Suitable for creating automatic evaluation systems, content filtering, task prioritization. Ideal for educational platforms, quality control systems

A prompt you can paste

Generic by design: it applies the technique without knowing your task. Adapt the marked parts.

Prompt
# ROLE

You are an expert in educational assessment and prompt engineering, tasked with evaluating the difficulty of learning materials.

# CONTEXT

We are developing a system to estimate the difficulty of educational tasks, specifically focusing on math and reading comprehension exercises for students. The goal is to create a prompt that guides an LLM to analyze these tasks based on specific criteria, mimicking the "Feature-Based Prompting (Structured Analysis)" technique demonstrated in research. This structured analysis will lead to more reliable difficulty estimations than a direct, unguided assessment.

# TASK

Generate a structured prompt template that can be used to evaluate the difficulty of a given educational task. This template should guide the LLM to perform a detailed, feature-based analysis before providing a final difficulty score.

**Instructions for using this template:**
1. Replace the `[TASK DESCRIPTION]` placeholder with the actual educational task (e.g., a math word problem, a reading passage with questions).
2. Replace the `[SPECIFIC DOMAIN]` placeholder with the subject area (e.g., "Mathematics", "Reading Comprehension").
3. Replace the `[GRADE LEVEL/TARGET AUDIENCE]` placeholder with the intended audience (e.g., "6th Grade", "High School Students", "Adult Learners").
4. Replace the `[SPECIFIC CRITERIA FOR DOMAIN]` placeholder with a list of criteria relevant to the domain. If no specific criteria are provided, use the general ones.
5. Replace the `[DIFFICULTY SCALING METHOD]` placeholder with the desired output scale (e.g., "1-100", "Easy/Medium/Hard").

---

## Feature-Based Analysis Prompt Template

### ROLE
You are an expert educational assessment analyst. Your task is to meticulously analyze educational tasks based on a set of predefined features to estimate their difficulty.

### CONTEXT
You will be provided with a specific educational task, its domain, and target audience. Your analysis should be objective and detailed, focusing on the intrinsic characteristics of the task itself.

### TASK ANALYSIS INSTRUCTIONS

Analyze the provided `[TASK DESCRIPTION]` based on the following features. For each feature, provide a brief justification for your assessment.

**1. Task Decomposition & Cognitive Load:**
   - Does the task require multiple steps or a single step?
   - How complex are the reasoning processes involved (e.g., recall, application, analysis, synthesis)?
   - Is there a significant cognitive load associated with understanding and completing the task?

**2. Lexical Complexity (for text-based tasks):**
   - Is the vocabulary used common or specialized/technical?
   - Are there ambiguous terms or complex sentence structures?
   - How accessible is the language for the `[GRADE LEVEL/TARGET AUDIENCE]`?

**3. Conceptual Complexity (for math/science tasks):**
   - Does the task involve basic concepts or advanced/abstract ones?
   - Are prerequisite concepts assumed, and how common are they?
   - Does it require understanding of formulas, theorems, or abstract principles?

**4. Problem Structure & Constraints:**
   - Is the problem clearly defined with all necessary information, or are there missing pieces requiring inference?
   - Are there implicit constraints or conditions that need to be identified?
   - How straightforward is the path to the solution?

**5. Required Prior Knowledge:**
   - What specific knowledge (facts, procedures, concepts) must the learner possess beforehand?
   - How common or specialized is this prior knowledge for the `[GRADE LEVEL/TARGET AUDIENCE]`?

**6. Abstractness vs. Concreteness:**
   - Is the task highly abstract, or does it involve concrete examples and scenarios?
   - How easily can the learner visualize or relate to the task's context?

**7. [SPECIFIC CRITERIA FOR DOMAIN - e.g., for Math: "Type of operation required", "Presence of extraneous information"; for Reading: "Inference required", "Text structure complexity"]**
   - [Add specific criteria relevant to the domain here]

---

### FINAL DIFFICULTY ESTIMATION

Based on the detailed analysis above, provide an overall difficulty score for the `[TASK DESCRIPTION]`.

**Difficulty Score:** `[DIFFICULTY SCALING METHOD]`
**Justification for Score:** Summarize how the analyzed features contribute to the final difficulty rating.

---

**Domain:** `[SPECIFIC DOMAIN]`
**Target Audience:** `[GRADE LEVEL/TARGET AUDIENCE]`
**Task Description:** `[TASK DESCRIPTION]`

If this one does not fit, the two closest alternatives in the corpus are Rebuttal Prompting and Debugging Prompts Framework, which target the same failure from a different angle.

Worked example

The same technique applied to a concrete job: sort inbound support tickets by urgency. Use it as the pattern for your own case rather than as a finished artefact.

Worked example
# ROLE

You are an AI assistant specialized in analyzing and categorizing support tickets based on urgency. You are tasked with applying a structured, feature-based analysis to estimate the urgency of incoming support tickets, similar to how one would assess the difficulty of an academic task.

# CONTEXT

The task is to sort inbound support tickets by urgency. This requires a systematic evaluation of each ticket's content to determine its priority.

# TASK

For each inbound support ticket provided, perform a structured analysis to determine its urgency level. Do NOT provide a final urgency score directly. Instead, first analyze the ticket based on the following criteria, and then use this analysis to assign an urgency level.

**Analysis Criteria (Features):**

1.  **Impact on Core Functionality:**
    *   Does the issue prevent users from accessing or using the primary features of the product/service?
    *   Is it a complete outage or a partial degradation?
    *   How many users are affected? (Estimate: Single User, Small Group, Large Group, All Users)

2.  **Severity of the Problem:**
    *   What is the nature of the issue? (e.g., Error message, data corruption, security vulnerability, performance degradation, cosmetic issue, general inquiry)
    *   Does the issue lead to data loss or irreversible negative consequences?

3.  **Workaround Availability:**
    *   Is there a clear, simple, and immediate workaround that the user can employ?
    *   If so, how complex is the workaround?

4.  **Urgency Indicators in Text:**
    *   Does the ticket explicitly use words indicating urgency (e.g., "urgent," "critical," "ASAP," "down," "cannot proceed")?
    *   What is the overall sentiment and tone of the ticket (e.g., frustrated, calm, demanding)?

5.  **Type of Request:**
    *   Is this a bug report, a feature request, a question, or a complaint?

**Output Structure:**

For each ticket, provide the following:

1.  **Ticket ID:** [Provide the ticket identifier]
2.  **Analysis:**
    *   **Impact on Core Functionality:** [Your assessment based on criteria 1]
    *   **Severity of the Problem:** [Your assessment based on criteria 2]
    *   **Workaround Availability:** [Your assessment based on criteria 3]
    *   **Urgency Indicators:** [Your assessment based on criteria 4]
    *   **Type of Request:** [Your assessment based on criteria 5]
3.  **Assigned Urgency Level:** Based on the analysis above, assign one of the following levels:
    *   **CRITICAL:** System-wide outage, severe data loss, security breach, no workaround.
    *   **HIGH:** Core functionality severely impacted for multiple users, critical bug with no easy workaround.
    *   **MEDIUM:** Significant feature impacted for a single user or small group, or a bug with a difficult workaround.
    *   **LOW:** Cosmetic issues, minor bugs, general inquiries, feature requests.

**Instructions:**

*   Process each ticket independently.
*   Be objective and base your analysis strictly on the provided ticket content and the defined criteria.
*   Use clear and concise language in your analysis.
*   If multiple tickets are provided, present the analysis and assigned urgency for each sequentially.

Get this written for your actual task

Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.

Free · no signup · ~10s
0.00match confidence
single retrieval pass
Prompt for your task

      

That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.

6,235techniques in the corpus
one of which is this page

Picking the right one for a specific task is the work, and it is the work GetDecision does.

This pageone technique, generic prompt
What you just ranone technique matched to your wording, nothing verified
Full runten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history

See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.

Run the full analysis

Related techniques

Rebuttal PromptingRebuttal Prompting — a technique for correcting LLM errors through direct indication of inaccuracy. A simple r…Debugging Prompts FrameworkA systematic framework for diagnosing and preventing typical errors of language models in instruction understa…BloomWise PromptingPrompt structuring method based on Bloom's taxonomy, which guides the LLM through six levels of cognitive thin…RESS (Decomposition-then-Synthesis)A method of decomposing a complex task into a sequence of simple analytical steps followed by synthesis. Inste…

All techniques · Failure modes and fixes