A collective textbook-guide by Colaboratoria
Prompting
- Olga AkimovaAuthor
In academic research, using prompts correctly makes it possible to automate data analysis, conduct comparative studies, and much more. In this chapter, we will look at the key principles of prompting adapted to digital diplomacy tasks and provide examples of effective queries.
/01
Core principles of prompting
Prompting is the process of formulating queries to large language models (LLMs) in order to obtain the desired result.
Role
Defining the model’s role is the first step toward an effective prompt. In academic research, it is important to specify that the model acts as an expert in a particular field. For example, if you are analyzing international treaties, the model should be an "expert in international law and digital diplomacy."
Defining the role helps an LLM focus only on the data that, during training, was associated with the domain you specify.
Bad: "Tell me about international treaties."
Good: "You are an expert in international law and digital diplomacy. Analyze the text of an international treaty and highlight the key provisions related to digital security."
Good: "You are an expert in international law and digital diplomacy. Analyze the text of an international treaty and highlight the key provisions related to digital security."
Format
Specifying the output format helps the model structure its response according to your requirements. In academic research, this may be a report, an analytical memo, a comparative table, or even a presentation script.
Bad: "Write about digital diplomacy."
Good: "Write a 500-word plan for a future conference paper on the impact of digital diplomacy on international relations, using data from recent research in this field."
Good: "Write a 500-word plan for a future conference paper on the impact of digital diplomacy on international relations, using data from recent research in this field."
Context
Context includes keywords, the target audience, conditions, and the expected length/volume of the text. In digital diplomacy, it is important to consider the specifics of international relations, political aspects, and cultural features.
Bad: "How do international organizations work?"
Good: "Analyze the uploaded article and explain how international organizations use digital technologies to improve interaction between countries, taking into account cultural and political specifics."
Good: "Analyze the uploaded article and explain how international organizations use digital technologies to improve interaction between countries, taking into account cultural and political specifics."
/02
Prompting formulas
There are different formulas for effective prompts. In this section, we will look at the most common ones.
Role – Task – Format
Task – Actions – Goal
Role – Information – Steps – Expected Result
Role – Task – Format / R – T – F
Define the role, set the task, and specify the response format.
- /promptRole: You are an expert in digital diplomacy.
Task: Analyze the text of an international treaty and highlight the key provisions related to digital security.
Format: Present the result as a table with sections and subheadings.
Task – Actions – Goal / T – A – G
Describe the task, the actions required, and the final goal.
- /promptTask: Compare two international treaties on digital security.
Actions: Identify key differences in approaches to data protection and cybersecurity.
Goal: The goal is to understand which treaty is more effective under modern conditions.
Role – Information – Steps – Expected Result / R – I – S – E
Define the role, provide the data, list steps, and describe the expected output.
- /promptRole: You are an expert in digital diplomacy.
Information: You have the texts of two international treaties on digital security.
Steps: Compare the approaches to data protection and cybersecurity in these treaties.
Expected Result: Present the result as a comparative table with the key differences.
Effective prompts used in academic research can be quite long.
- /promptRole:
You are an expert in digital diplomacy and user research. Your task is to create realistic user personas that represent types of people from the target audience whom I would most likely encounter if I were conducting user interviews on the topic of digital diplomacy. These personas will help better understand the needs, goals, and problems of potential users, which will make it possible to improve the development and implementation of digital diplomacy tools and solutions.
Information:
Here is a description of my product/offer:
[a general ethical code for writing posts in the media]
And here is the target audience for this product:
[all individuals holding official positions in countries whose media posts can influence relations between countries]
Steps:
Create 5 different user personas based on the provided information about the product and the target audience. For each persona, include the following information:
Name:
Core demographic information: age, gender, occupation, location, level of education, professional experience in diplomacy or international relations.
Primary goals and motivations: what drives this person in the context of digital diplomacy? What tasks are they trying to solve?
Main challenges and pain points: what difficulties does this person face in their work or when using digital tools in diplomacy?
Daily activities: describe what a typical day looks like for this person, including their professional duties and use of digital technologies.
Persona quotes: provide several quotes that reflect this person’s thoughts, opinions, or statements about digital diplomacy or the use of technology in their work.
Make sure each persona is realistic and represents a specific segment of the target audience. Include relevant details from the product and audience descriptions I provided.
The personas should be different enough to cover a wide range of potential users I may encounter in the field of digital diplomacy. Focus on creating useful and realistic personas that will help in conducting user research and making decisions to improve the product.
Do not add any fictional details beyond what is necessary to create recognizable user profiles.
Expected Result:
Present 5 user personas, each reflecting a unique profile within the target audience in the field of digital diplomacy. Each persona should be described in the following format:
Persona 1:
Name: [Persona name]
Core demographic information:
Age: [age]
Gender: [gender]
Occupation: [occupation, e.g., diplomat, international relations analyst, digital security expert, etc.]
Location: [city, country]
Level of education: [MA in International Relations, PhD in Political Science, etc.]
Professional experience: [e.g., 10 years at the Ministry of Foreign Affairs, specialization in digital diplomacy, etc.]
Primary goals and motivations:
[Describe what drives this person, e.g., a desire to improve international cooperation through digital technologies, a desire to increase the efficiency of diplomatic processes, etc.]
Main challenges and pain points:
[Describe difficulties this person faces, e.g., lack of digital tools for big data analysis, challenges in coordinating international projects, etc.]
Daily activities:
[Describe a typical day for this person, e.g., participating in online conferences, analyzing international relations data, developing digital diplomacy strategies, etc.]
Persona quotes:
[Provide several quotes, e.g., "Digital diplomacy is the future of international relations, but we lack tools to work effectively with big data."]
Persona 2:
[Repeat the same format for the second persona]
Persona 3:
[Repeat the same format for the third persona]
Persona 4:
[Repeat the same format for the fourth persona]
Persona 5:
[Repeat the same format for the fifth persona]
Make sure each persona is unique and reflects different aspects of the target audience in the field of digital diplomacy. These personas should help in understanding user needs and making decisions to improve the product.
- /answer by DeepSeekPersona 1:
1. Name: Anna Ivanova
2. Core demographic information:
Age: 35
Gender: Female
Occupation: Diplomat, digital diplomacy specialist
Location: Moscow, Russia
Level of education: MA in International Relations
Professional experience: 8 years at the Ministry of Foreign Affairs, specialization in digital diplomacy and public diplomacy
3. Primary goals and motivations:- Aspiration to improve the country’s image through digital platforms
- Desire to increase the effectiveness of communication with foreign partners via social media
- Lack of tools to analyze the effectiveness of digital campaigns
- Difficulties in coordinating public messages across different government agencies
- Participation in online conferences on digital diplomacy
- Analysis of international audience reactions to social media posts
- Development of strategies to improve engagement with foreign media
- "Digital diplomacy is not just a trend — it is a necessity in today's world."
- "We need tools that help measure the impact of our publications on international audiences."
Persona 2:
1. Name: John Smith
2. Core demographic information:
Age: 42
Gender: Male
Occupation: Ambassador to the UN
Location: New York, USA
Level of education: PhD in Political Science
Professional experience: 15 years in diplomacy, including leadership positions in international organizations
3. Primary goals and motivations:- Strengthening international cooperation through digital platforms
- Desire to use technology to enhance transparency in diplomatic processes
- Lack of unified standards for digital diplomacy across countries
- Risks related to cybersecurity and leakage of confidential information
- Participation in international summits and negotiations
- Working with analytical reports on the impact of digital technologies on international relations
- Coordinating public messaging for the United Nations
- "Digital diplomacy must be secure and transparent."
- "We must use technology to make diplomacy more accessible to everyone."
Persona 3:
1. Name: Sofia Müller
2. Core demographic information:
Age: 33
Gender: Female
Occupation: Digital security expert
Location: Berlin, Germany
Level of education: MA in Cybersecurity
Professional experience: 7 years in cybersecurity, specializing in data protection within diplomatic institutions
3. Primary goals and motivations:- Ensuring the security of digital communications in diplomacy
- Desire to develop standards for protecting confidential information
- Insufficient resources for implementing advanced cybersecurity technologies
- Difficulties in coordination between different agencies
- Analyzing cybersecurity threats to diplomatic institutions
- Developing data protection recommendations
- Participating in international cybersecurity conferences
- "Security of digital communications is the foundation of modern diplomacy."
- "We must protect our data to maintain the trust of international partners."
Persona 4:
1. Name: Ahmed Al-Farisi
2. Core demographic information:
Age: 50
Gender: Male
Occupation: Minister of Foreign Affairs
Location: Riyadh, Saudi Arabia
Level of education: MA in International Relations
Professional experience: 20 years in diplomacy, including leadership positions at the Ministry of Foreign Affairs
3. Primary goals and motivations:- Strengthening the country’s image on the international stage through digital platforms
- Desire to use technology to improve interaction with foreign partners
- Shortage of qualified personnel to work with digital tools
- Risks related to cybersecurity and leakage of confidential information
- Participation in international negotiations and summits
- Working with analytical reports on the impact of digital technologies on international relations
- Coordinating public messaging for the Ministry of Foreign Affairs
- "Digital diplomacy is the key to success in the modern world."
- "We must invest in technology to remain competitive on the international stage."
Persona 5:
1. Name: Maria Gonzalez
2. Core demographic information:
Age: 29
Gender: Female
Occupation: International relations analyst
Location: Madrid, Spain
Level of education: MA in International Relations
Professional experience: 5 years at a think tank, specializing in digital diplomacy and media analytics
3. Primary goals and motivations:- Analyzing the impact of digital technologies on international relations
- Desire to develop recommendations to improve the country’s digital strategy
- Lack of data to assess the effectiveness of digital campaigns
- Difficulties in interpreting international audience reactions to publications
- Analyzing data from social networks and media platforms
- Preparing reports for the government on digital diplomacy
- Participating in research on the impact of digital technologies on international relations
- "Digital diplomacy is not only about publishing content, but also about analyzing its impact."
- "We need more precise tools to measure the effectiveness of our strategies."
It is important to understand that creating an effective prompt that you can reuse in your work can be no less labor-intensive than collecting materials for the study.
/03
Limitations of LLMs in academic research
Despite the capabilities of large language models, using them in academic research comes with a number of limitations. These limitations must be considered to avoid errors and incorrect conclusions.
Let’s look at the main ones:
Large language models have a limited context window—i.e., the amount of information the model can process and "remember" at once. This means that if the prompt context is too large, the model may "forget" part of the information, resulting in incomplete or inaccurate answers.
For example, when analyzing lengthy international treaties or large academic articles, the model may miss important details if the text exceeds the allowable context length.
For example, when analyzing lengthy international treaties or large academic articles, the model may miss important details if the text exceeds the allowable context length.
Using LLMs in academic research requires compliance with ethical and legal standards, especially regarding data confidentiality and copyright. Models trained on large datasets may inadvertently reproduce confidential information or violate copyright if the source data included such materials. In digital diplomacy, where work often involves classified or sensitive documents, this can lead to serious consequences. Therefore, it is important to carefully check data sources and comply with legal requirements.
One of the most serious problems with LLMs is their tendency to "hallucinate"—to generate information that looks plausible but is not true. Models can invent facts, quotations, or even entire studies, which is especially dangerous in academic work.
For example, when analyzing international treaties or political documents, a model may add non-existent clauses or distort the meaning of the text. Therefore, it is always essential to verify and re-check information produced by an LLM.
For example, when analyzing international treaties or political documents, a model may add non-existent clauses or distort the meaning of the text. Therefore, it is always essential to verify and re-check information produced by an LLM.
Large language models are trained on huge datasets that may contain biases and stereotypes. These biases can be inherited by the model and appear in its answers. For example, when analyzing political documents or international relations, the model may inadvertently reinforce existing stereotypes or biased viewpoints. This is especially important in digital diplomacy, where objectivity and neutrality are crucial.
You can learn more about bias in large language models and its impact on data analysis results in Bender et al., 2021.
You can learn more about bias in large language models and its impact on data analysis results in Bender et al., 2021.
LLM developers often impose restrictions on topics the model can discuss to avoid generating harmful or inappropriate content. However, this can cause the model to avoid important but controversial topics, limiting its usefulness in academic research.
For example, when analyzing conflict situations in international relations, the model may omit key aspects due to internal filters.
For example, when analyzing conflict situations in international relations, the model may omit key aspects due to internal filters.
Using LLMs in academic research, including digital diplomacy, requires a conscious approach and awareness of these limitations. Limited context, ethical and legal aspects, hallucinations, bias, and censorship can all affect the quality and reliability of results. Therefore, it is important not only to write prompts properly, but also to critically evaluate model outputs, verify information, and consider context.
Practicum
Using the R-I-S-E formula, write a query to analyze a public speech by a politician (for example, a foreign minister’s speech at the UN General Assembly).
Include:
Include:
- Role (e.g., "an expert in international relations and discourse analysis").
- Information (you can use a real speech from open sources).
- Steps (e.g., "identify key narratives, rhetorical devices, and implicit political messages").
- Expected result (e.g., "a table with themes, tone, and potential impact on an international audience").
Run the prompt in ChatGPT / DeepSeek / Claude and analyze the response.
Identify possible:
Identify possible:
- Hallucinations (are there factual errors?)
- Omissions (does the model ignore important aspects?)
- Bias (are there signs of stereotypes in the interpretation?)
Write a short conclusion (200−300 words) about how suitable an LLM is for such tasks and what could improve the reliability of the analysis.
- Prompting guide by Yandex Education: https://education.yandex.ru/handbook/prompting
- Online course by SberUniversity "Working with LLM GigaChat": https://courses.sberuniversity.ru/llm-gigachat/
- Prompt Engineering Guide от DAIR.AI: https://www.promptingguide.ai/ru
