Prompting technique

RICE Framework (Role, Instruction, Context, Examples with Rationale)

Classifier or Prompt: A Case Study on Legal Requirements Traceability, arXiv:2502.04916 (2025) · read the paper

What it does

A framework for structuring prompts from four components: Role, Instruction, Context, and Examples with Rationale. Key feature — adding explanation of logic to examples, which allows the model to understand not only what to do, but why, significantly improving accuracy of complex tasks.

It turned out that a carefully structured prompt, built according to the RICE framework, showed incomparably better results, increasing the completeness of connection detection from 15% to 84%.checked against the paper
Task typeclassification
Shapesingle shot
Efforttwo or three steps
Addressesanswers that are close but wrong; the same prompt giving different answers; the model losing what was agreed earlier
Published2025-02-07
AuthorsRomina Etezadi, Sallam Abualhaija, Chetan Arora et al.

When it helps

Use the RICE framework when you need to solve a complex analytical task of classification or matching, especially in specialized domains (jurisprudence, medicine, finance). Ideal for cases where training a specialized model is too costly or there is insufficient data. Apply when both accuracy and explainability of solutions are important. Particularly effective for requirement linking tasks, compliance analysis, document classification, and extraction of structured information.

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 AI assistant specializing in legal compliance and software development. Your task is to meticulously trace legal requirements against software features.

**[Context]**
You are analyzing the relationship between a set of legal regulations and the features of a software product. The goal is to ensure full compliance by identifying which features address which legal requirements. Accuracy and completeness are paramount.

**[Examples with Rationale]**

*   **Example 1: Legal Requirement vs. Feature**
    *   **Legal Requirement:** "Article 15 of GDPR (General Data Protection Regulation) states that data subjects have the right to access their personal data."
    *   **Software Feature:** "User Profile Page: Allows users to view and download all personal data stored by the application."
    *   **Traceability Analysis:**
        *   **Link:** Direct Match
        *   **Rationale:** The "User Profile Page" directly implements the right of access by providing a mechanism for users to view and download their personal data, fulfilling Article 15 of GDPR.

*   **Example 2: Legal Requirement vs. Feature (Partial/Indirect)**
    *   **Legal Requirement:** "The Children's Online Privacy Protection Act (COPPA) requires parental consent for collecting personal information from children under 13."
    *   **Software Feature:** "User Registration Form: Includes fields for date of birth and parent's email address."
    *   **Traceability Analysis:**
        *   **Link:** Partial/Indirect Support
        *   **Rationale:** While the feature collects necessary information (date of birth, parent's email), it doesn't explicitly detail the *process* of obtaining parental consent itself. It supports the requirement but doesn't fully implement the consent mechanism. Further features or workflows are needed.

*   **Example 3: Legal Requirement NOT Addressed by Feature**
    *   **Legal Requirement:** "The Payment Card Industry Data Security Standard (PCI DSS) mandates regular vulnerability scanning."
    *   **Software Feature:** "Secure Payment Gateway Integration: Encrypts all credit card transactions."
    *   **Traceability Analysis:**
        *   **Link:** No Match
        *   **Rationale:** The "Secure Payment Gateway Integration" addresses data encryption, which is a part of PCI DSS, but it does not cover the requirement for regular vulnerability scanning. This legal requirement remains unaddressed by the described feature.

**[Instruction]**
Analyze the following Legal Requirement and Software Feature. Determine the type of link between them (Direct Match, Partial/Indirect Support, No Match). Provide a clear rationale explaining your assessment, focusing on how the feature addresses or fails to address the specific aspects of the legal requirement. Prioritize identifying *all* potential links and supporting evidence, even if they seem minor, to maximize completeness.

**[Constraints]**
*   Focus solely on the provided Legal Requirement and Software Feature.
*   The rationale should be concise but comprehensive.
*   If a feature only partially supports a requirement, clearly state what is missing.

**[Output Indicator]**
Present your analysis in the following structured format:

*   **Legal Requirement:** [The provided legal text]
*   **Software Feature:** [The provided software feature description]
*   **Traceability Analysis:**
    *   **Link:** [Direct Match / Partial/Indirect Support / No Match]
    *   **Rationale:** [Your explanation]

If this one does not fit, the two closest alternatives in the corpus are Goal-Reversed 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
**[Контекст и Роль]**
Ты — опытный специалист по поддержке клиентов и классификации запросов. Твоя задача — анализировать входящие тикеты поддержки и сортировать их по степени срочности, чтобы обеспечить своевременное решение наиболее критичных проблем.

**[Примеры с Обоснованием]**
Вот несколько примеров, как нужно выполнять сортировку тикетов.

*   **Пример 1:**
    *   **Тикет:** "Здравствуйте, я не могу войти в свой аккаунт уже второй час. Все мои рабочие данные там!"
    *   **Анализ:**
        *   **Срочность:** Высокая
        *   **Категория:** Доступ к аккаунту (блокировка/потеря доступа)
        *   **Суть проблемы:** Пользователь полностью лишен доступа к критически важным рабочим данным.
    *   **Обоснование:** Потеря доступа к рабочему аккаунту напрямую влияет на возможность выполнения работы и доступ к данным, что требует немедленного вмешательства.

*   **Пример 2:**
    *   **Тикет:** "Подскажите, пожалуйста, где в настройках найти опцию экспорта данных? Не могу найти."
    *   **Анализ:**
        *   **Срочность:** Низкая
        *   **Категория:** Вопрос по функционалу (инструкция/подсказка)
        *   **Суть проблемы:** Пользователю нужна информация о расположении функции.
    *   **Обоснование:** Запрос является информационным и не препятствует основной работе пользователя или доступности сервиса.

*   **Пример 3:**
    *   **Тикет:** "Мое приложение начало работать очень медленно после последнего обновления. Иногда зависает на несколько минут."
    *   **Анализ:**
        *   **Срочность:** Средняя
        *   **Категория:** Производительность (замедление/зависания)
        *   **Суть проблемы:** Снижение производительности и зависания влияют на удобство использования, но не блокируют полностью.
    *   **Обоснование:** Проблемы с производительностью ухудшают пользовательский опыт и могут привести к потере данных или времени, но не являются полной блокировкой.

**[Инструкция]**
Теперь проанализируй следующий тикет поддержки по тому же принципу. Определи его срочность (Низкая, Средняя, Высокая), выдели основную категорию проблемы и кратко опиши суть проблемы. Отдавай приоритет выявлению проблем, которые полностью блокируют работу пользователя или доступ к критически важным данным.

**[Индикатор Вывода]**
Предоставь ответ в следующем формате:
- **Срочность:** <Низкая/Средняя/Высокая>
- **Категория:** <название категории>
- **Суть проблемы:** <описание проблемы>

---
**Тикет для анализа:**
"Я пытаюсь оплатить подписку, но платежная система выдает ошибку 'Недостаточно средств', хотя на карте достаточно денег. Не могу продолжить пользоваться сервисом без активной подписки."

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

Goal-Reversed PromptingA method that consists of inverting the query goal: instead of asking the LLM to find the best option, it is a…Debugging Prompts FrameworkA systematic framework for diagnosing and preventing typical errors of language models in instruction understa…Modular Definition PromptingA method of explicitly defining complex concepts in a prompt through decomposition into modular conceptual ele…Multi-Aspect Cueing (MAC)The Multi-Aspect Cueing (MAC) method involves explicitly listing key aspects in the prompt as a checklist. Ins…

All techniques · Failure modes and fixes