DIKW Hierarchy

A layered model that describes how raw symbols become data, then information, then knowledge, then wisdom --- each layer adding more context, meaning, and judgement.

---

What is it?

The DIKW hierarchy is a pyramid with four layers: Data, Information, Knowledge, and Wisdom. Each layer builds on the one below it by adding something the previous layer lacks --- context, structure, experience, or judgement. The model answers a deceptively simple question: what transforms raw symbols into something useful?¹

At the bottom sits data --- isolated symbols, numbers, or signals with no meaning attached. The number “38.5” is data. It could be a temperature, a stock price, or a shoe size. Without context, it is just a symbol. Moving up, information emerges when data is organised and contextualised. “The patient’s temperature is 38.5C” is information --- the same number now has structure and meaning.¹

Knowledge appears when information is integrated with experience and understanding. A doctor who sees “38.5C, post-surgical patient, day two” draws on years of training to recognise this as a potential sign of infection. Knowledge is information that has been internalised and connected to a web of other understanding. Finally, wisdom is knowledge applied with judgement --- knowing not just what could be done, but what should be done, weighing consequences, ethics, and context.²

The hierarchy has been influential in information science, knowledge management, and AI since Russell Ackoff formalised it in 1989, though its roots reach back further.¹ It remains a foundational mental model for understanding why data alone is never enough --- and why the hard work lies in the transformations between layers.

In plain terms

The DIKW hierarchy is like the journey from a pile of ingredients to a well-judged meal. Raw ingredients (data) mean nothing on their own. A recipe organises them (information). A skilled cook knows when to deviate from the recipe (knowledge). A wise host knows which dish suits which guest and occasion (wisdom).

---

At a glance

The four layers of DIKW (click to expand)

graph TD
    D[Data — Raw symbols without context] -->|Add context and structure| I[Information — Data with meaning]
    I -->|Integrate with experience| K[Knowledge — Information understood]
    K -->|Apply judgement and ethics| W[Wisdom — Knowledge used well]

Key: Each arrow represents a transformation. Data becomes information through structuring and contextualising. Information becomes knowledge through experience and integration. Knowledge becomes wisdom through reflective judgement. Each step requires something that cannot be automated trivially.

---

How does it work?

The hierarchy has four layers. Each layer transforms the one below it through a distinct process.

1. Data

Data consists of raw, unprocessed symbols --- numbers, characters, signals, measurements. On its own, data has no meaning. It is the “what” without the “so what.”¹

For example: 42, Paris, true, 2026-04-04. These are data points. They could mean anything or nothing without a frame of reference.

Think of it like...

Data is like individual Scrabble tiles scattered on a table. Each tile has a letter, but no tile is a word. The letters are real, but meaningless in isolation.

2. Information

Information is data that has been organised, structured, and given context. The transformation from data to information happens through processes like classifying, sorting, summarising, and relating data to other data.¹

“The temperature in Paris on 4 April 2026 was 42F” takes the isolated data points 42, Paris, and 2026-04-04 and assembles them into a statement with meaning. Information answers questions: who, what, where, when, how many.²

Think of it like...

Information is a sentence formed from Scrabble tiles. The tiles are now arranged in an order that communicates something. The same tiles in a different arrangement would communicate something different.

Example: data vs information (click to expand)

Data	Information
38.5	The patient’s temperature is 38.5C
200, OK	The server returned HTTP 200 OK
12, 47, 3	The team scored 12, conceded 47, and drew 3 matches

In each case, the raw number means nothing until it is placed in a structure with labels, units, and context.

3. Knowledge

Knowledge emerges when information is integrated with experience, understanding, and the ability to recognise patterns. Knowledge is not merely accumulated information --- it is information that has been internalised, connected to other knowledge, and made actionable.¹

A doctor who reads “38.5C, post-surgical, day two” does not just see a data point in context. They recognise a pattern: post-operative fever on day two could indicate atelectasis, wound infection, or a urinary tract infection depending on the procedure. This pattern recognition, drawn from years of study and clinical experience, is knowledge.²

Think of it like...

Knowledge is knowing which sentences to write --- and why. Not just forming grammatically correct sentences from Scrabble tiles, but choosing the right sentences because you understand their implications.

Key distinction

Information can be transferred directly (you can read a fact and immediately possess it). Knowledge cannot --- it requires integration with a person’s existing mental models and experience. This is why “reading the documentation” and “understanding the system” are very different things.

4. Wisdom

Wisdom is the reflective, evaluative application of knowledge. It involves judgement about what should be done, not just what can be done. Wisdom considers consequences, ethics, long-term effects, and context that knowledge alone does not capture.²

A knowledgeable doctor knows the antibiotic that would treat the infection. A wise doctor also considers the patient’s allergies, the risk of antibiotic resistance, the patient’s values and preferences, and whether observation might be more appropriate than immediate treatment.

Think of it like...

Wisdom is knowing when not to write a sentence, even though you can. It is the judgement layer --- understanding not just how to act, but whether you should.

---

Why do we use it?

Key reasons

1. Diagnosing where systems and processes fail. Most failures in information systems happen at the transitions: data that never gets contextualised into information, information that never gets integrated into knowledge. The DIKW model helps locate the breakdown.¹

2. Designing better knowledge systems. When building AI, search, or knowledge management tools, the hierarchy clarifies what each component must do. A database stores data. A search engine surfaces information. A knowledge graph captures knowledge. Each serves a different layer.³

3. Understanding the limits of automation. The lower layers (data, information) are relatively easy to automate. The upper layers (knowledge, wisdom) resist automation because they require experience, judgement, and context that are hard to encode. The model makes these limits visible.²

---

When do we use it?

• When designing a data pipeline and needing to clarify what each stage should produce
• When evaluating whether an AI system is truly “intelligent” or merely processing data
• When building a knowledge management system and deciding what to capture and how
• When explaining to stakeholders why more data does not automatically mean better decisions
• When assessing what a large language model can and cannot do at each layer

Rule of thumb

If someone claims their system produces “knowledge” or “wisdom,” ask what transformations it performs. If it only organises and retrieves data, it is an information system, not a knowledge system.

---

How can I think about it?

The library analogy

A library contains millions of data points (individual words on pages). The books organise those words into information (structured arguments, narratives, facts in context). A well-read scholar who has spent years studying those books has knowledge --- they can synthesise across sources, recognise contradictions, and draw novel conclusions. A wise mentor who has lived through the situations the books describe can advise on what to do with that knowledge, considering human consequences that no book fully captures.

• Data = individual words
• Information = the books themselves
• Knowledge = the scholar’s integrated understanding across books
• Wisdom = the mentor’s judgement about when and how to apply that understanding

The GPS analogy

A GPS sensor generates data: raw latitude and longitude coordinates streamed every second. Mapping software transforms this into information: “You are on Rue de Bourg, Lausanne, heading north.” A knowledgeable driver integrates this with experience: “This road gets congested at 5pm, I should take the side street.” A wise driver also considers: “But my passenger gets carsick on winding roads --- the main road is worth the delay.”

• Data = raw coordinates
• Information = your position on a labelled map
• Knowledge = knowing which route is faster based on experience
• Wisdom = choosing the route that balances speed, comfort, and context

---

Concepts to explore next

Concept	What it covers	Status
knowledge-graphs	Graph structures that represent knowledge as connected nodes and edges	complete
structured-data-vs-prose	The difference between machine-readable structure and human-readable text	complete
embeddings	How machines represent meaning as vectors in high-dimensional space	complete

Some cards don't exist yet

A broken link is a placeholder for future learning, not an error.

---

Check your understanding

Test yourself (click to expand)

1. Explain the difference between data and information using an example not found on this card.
2. Distinguish between knowledge and wisdom. Why is the boundary between them the hardest to define?
3. Name the four layers of the DIKW hierarchy and describe the transformation that connects each pair.
4. Interpret this scenario: a company collects millions of customer feedback records but cannot improve its product. At which DIKW layer is the process likely failing, and why?
5. Connect the DIKW hierarchy to knowledge-graphs: at which layer of the hierarchy does a knowledge graph operate, and what transformations does it encode?

---

Where this concept fits

Position in the knowledge graph

graph TD
    KE[Knowledge Engineering] --> DIKW[DIKW Hierarchy]
    KE --> KG[Knowledge Graphs]
    KE --> MRF[Machine-Readable Formats]
    style DIKW fill:#4a9ede,color:#fff

Related concepts:

• knowledge-graphs --- knowledge graphs operate at the knowledge layer of the hierarchy, encoding not just data but relationships and meaning
• structured-data-vs-prose --- the transformation from data to information often involves moving from unstructured prose to structured formats
• embeddings --- embeddings represent meaning as vectors, capturing some aspects of the information-to-knowledge transition

---

Sources

---

Further reading

Resources

• DIKW: A Semiotic Theoretical and Empirical Exploration (ResearchGate) --- Academic paper tracing the hierarchy’s theoretical roots through semiotics and information science
• The Theoretical Foundations of Metadata in Knowledge Management (Fehlau) --- Connects the DIKW model to metadata theory and knowledge management practice
• DIKW Pyramid in Modern Data Management (Preprints.org) --- Contemporary analysis of the hierarchy’s relevance to data engineering and AI systems
• The DIKW Model (Wikipedia) --- Concise overview with historical context and criticisms of the model

Footnotes

1. Liew, A. (2013). DIKIW: Data, Information, Knowledge, Intelligence, Wisdom and their Interrelationships. Business Management Dynamics, 2(10), 49-62. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

2. Fehlau, M. (2025). The Theoretical Foundations of Metadata in Knowledge Management. Fehlau.de. ↩ ↩² ↩³ ↩⁴ ↩⁵

3. Ochab, M. et al. (2025). DIKW Pyramid in the Context of Modern Data Management. Preprints.org. ↩