RDF [[RDF-CONCEPTS]] describes a graph-based data model for making claims about the world and provides the foundation for reasoning upon that graph of information. At times, it becomes necessary to compare the differences between sets of graphs, digitally sign them, or generate short identifiers for graphs via hashing algorithms. This document outlines an algorithm for normalizing RDF datasets such that these operations can be performed.

This document is a work in progress.


When data scientists discuss canonicalization, they do so in the context of achieving a particular set of goals. Since the same information may sometimes be expressed in a variety of different ways, it often becomes necessary to be able to transform each of these different ways into a single, standard format. With a standard format, the differences between two different sets of data can be easily determined, a cryptographically-strong hash identifier can be generated for a particular set of data, and a particular set of data may be digitally-signed for later verification.

In particular, this specification is about normalizing RDF datasets, which are collections of graphs. Since a directed graph can express the same information in more than one way, it requires canonicalization to achieve the aforementioned goals and any others that may arise via serendipity.

Most RDF datasets can be normalized fairly quickly, in terms of algorithmic time complexity. However, those that contain nodes that do not have globally unique identifiers pose a greater challenge. Normalizing these datasets presents the graph isomorphism problem, a problem that is believed to be difficult to solve quickly. More formally, it is believed to be an NP-Intermediate problem, that is, neither known to be solvable in polynomial time nor NP-complete. Fortunately, existing real world data is rarely modeled in a way that manifests this problem and new data can be modeled to avoid it. In fact, software systems can detect a problematic dataset and may choose to assume it's an attempted denial of service attack, rather than a real input, and abort.

This document outlines an algorithm for generating a normalized RDF dataset given an RDF dataset as input. The algorithm is called the Universal RDF Dataset Canonicalization Algorithm 2015 or URDNA2015.

Uses of Dataset Canonicalization

There are different use cases where graph or dataset canonicalization are important:

A canonicalization algorithm is necessary, but not necessarily sufficient, to handle many of these use cases. The use of blank nodes in RDF graphs and datasets has a long history and creates inevitable complexities. Blank nodes are used for different purposes:

Further more, RDF semantics dictate that deserializing the same RDF document results in the creation of unique blank nodes, unless it can be determined that on each occasion, the blank node identifiers the same resource; this is due to the fact that blank node identifiers are an aspect of a concrete RDF syntax and are not intended to be persistent or portable. Within the abstract RDF model, blank nodes do not have identifiers (although some RDF store implementations may use stable identifiers and choose to make them portable). See Blank Nodes in [[!RDF-CONCEPTS]] for more information.

RDF does have a provision for allowing blank nodes to be published in an externally identifiable way through the use of Skolem IRIs, which allows a given RDF store to replace the use of blank nodes in a concrete syntax with IRIs, which serve to repeatably identify that blank node within that particular RDF store, however, this is not generally useful for talking about the same graph in different RDF stores, or other concrete representations. In any case, a stable blank node identifier defined for one RDF store, serialization is arbitrary, and typically not relatable to the context within which it is used.

This specification defines an algorithm for creating stable blank node identifiers repeatably for different serializations possibly using individualized blank node identifiers of the same RDF graph (dataset) by grounding each blank node through the nodes to which it is connected, essentially creating Skolem blank node identifiers. As a result, a graph signature can be obtained by hashing a canonical serialization of the resulting normalized dataset, allowing for the isomorphism and digital signing use cases. As blank node identifiers can be stable even with other changes to a graph (dataset), in some cases it is possible to compute the difference between two graphs (datasets), for example if changes are made only to ground triples, or if new blank nodes are introduced which do not create an automorphic confusion with other existing blank nodes. If any information which would change the generated blank node identifier, a resulting diff might indicate a greater set of changes than actually exists.

TimBL has a design note on problems with Diff which should be referenced.

Jerremy Carroll has a paper on signing RDF graphs.

How to Read this Document

This document is a detailed specification for an RDF dataset canonicalization algorithm. The document is primarily intended for the following audiences:

To understand the basics in this specification you must be familiar with basic RDF concepts [[!RDF-CONCEPTS]]. A working knowledge of graph theory and graph isomorphism is also recommended.


There are a number of ways that one may participate in the development of this specification:


General Terminology

A string is a sequence of zero or more Unicode characters.
true and false
Values that are used to express one of two possible boolean states.
An IRI (Internationalized Resource Identifier) is a string that conforms to the syntax defined in [[RFC3987]].
A subject as specified by [[!RDF11-CONCEPTS]].
A predicate as specified by [[!RDF11-CONCEPTS]].
An object as specified by [[!RDF11-CONCEPTS]].
RDF triple
A triple as specified by [[!RDF11-CONCEPTS]].
RDF graph
An RDF graph as specified by [[!RDF11-CONCEPTS]].
graph name
A graph name as specified by [[!RDF11-CONCEPTS]].
A tuple composed of subject, predicate, object, and graph name. This is a generalization of an RDF triple along with a graph name.
RDF dataset
A dataset as specified by [[!RDF11-CONCEPTS]]. For the purposes of this specification, an RDF dataset is considered to be a set of quads
blank node
A blank node as specified by [[!RDF11-CONCEPTS]]. In short, it is a node in a graph that is neither an IRI, nor a literal.
blank node identifier
A blank node identifier as specified by [[!RDF11-CONCEPTS]]. In short, it is a string that begins with _: that is used as an identifier for an blank node. Blank node identifiers are typically implementation-specific local identifiers; this document specifies an algorithm for deterministically specifying them.


Canonicalization is the process of transforming an input dataset to a normalized dataset. That is, any two input datasets that contain the same information, regardless of their arrangement, will be transformed into identical normalized dataset. The problem requires directed graphs to be deterministically ordered into sets of nodes and edges. This is easy to do when all of the nodes have globally-unique identifiers, but can be difficult to do when some of the nodes do not. Any nodes without globally-unique identifiers must be issued deterministic identifiers.

Strictly speaking, the normalized dataset must be serialized to be stable, as within a dataset, blank node identifiers have no meaning. This specification defines a normalized dataset to include stable identifiers for blank nodes, but practical uses of this will always generate a canonical serialization of such a dataset.

In time, there may be more than one canonicalization algorithm and, therefore, for identification purposes, this algorithm is named the "Universal RDF Dataset Canonicalization Algorithm 2015" (URDNA2015).

This statement is overly prescriptive and does not include normative language. This spec should describe the theoretical basis for graph canonicalization and describe behavior using normative statements. The explicit algorithms should follow as an informative appendix.

Canonicalization Algorithm Terms

input dataset
The abstract RDF dataset that is provided as input to the algorithm.
normalized dataset
The immutable, abstract RDF dataset and set of normalized blank node identifiers that are produced as output by the algorithm. A normalized dataset is a restriction on an RDF dataset where all nodes are labeled, and blank nodes are labeled with blank node identifiers consistent with running this algorithm on a base RDF dataset. A concrete serialization of an normalized dataset MUST label all blank nodes using these stable blank node identifiers.
identifier issuer
An identifier issuer is used to issue new blank node identifier. It maintains a blank node identifier issuer state.
The lowercase, hexadecimal representation of a message digest.
hash algorithm
The hash algorithm used by URDNA2015, namely, SHA-256.

Canonicalization State

When performing the steps required by the canonicalization algorithm, it is helpful to track state in a data structure called the canonicalization state. The information contained in the canonicalization state is described below.

blank node to quads map
A data structure that maps a blank node identifier to the quads in which they appear in the input dataset.
hash to blank nodes map
A data structure that maps a hash to a list of blank node identifiers.
canonical issuer
An identifier issuer, initialized with the prefix _:c14n, for issuing canonical blank node identifiers.
Mapping all blank nodes to use this identifier spec means that an RDF dataset composed of two different RDF graphs will use different identifiers then that for the graphs taken independently. This may happen anyway, due to automorphisms, or overlapping statements, but an identifier based on the resulting hash along with an issue sequence number specific to that hash would stand a better chance of surviving such minor changes, and allow the resulting information to be useful for RDF Diff.

Blank Node Identifier Issuer State

During the canonicalization algorithm, it is sometimes necessary to issue new identifiers to blank nodes. The Issue Identifier algorithm uses an identifier issuer to accomplish this task. The information an identifier issuer needs to keep track of is described below.

identifier prefix
The identifier prefix is a string that is used at the beginning of an blank node identifier. It should be initialized to a string that is specified by the canonicalization algorithm. When generating a new blank node identifier, the prefix is concatenated with a identifier counter. For example, _:c14n is a proper initial value for the identifier prefix that would produce blank node identifiers like _:c14n1.
identifier counter
A counter that is appended to the identifier prefix to create an blank node identifier. It is initialized to 0.
issued identifiers list
A list that tracks previously issued identifiers in the order in which they were issued. It also tracks the existing identifier that triggered the issuance, to prevent issuing more than one new identifier per existing identifier, and to allow blank nodes to be reassigned identifiers some time after issuance.

Canonicalization Algorithm

The canonicalization algorithm converts an input dataset into a normalized dataset. This algorithm will assign deterministic identifiers to any blank nodes in the input dataset.

Documenting the algorithm is a WIP, various steps will become more detailed in time.

See the note for the hash first degree quads algorithm. We should either remove the loop based on simple here but indicate that the original design of the algorithm was to have such a loop, or leave it but inform implementers that it is safe to break after one iteration of the loop (again, indicating why). A future version of this algorithm should make the loop effectual.



  1. Create the canonicalization state.
  2. For every quad in input dataset:
    1. For each blank node that occurs in the quad, add a reference to the quad using the blank node identifier in the blank node to quads map, creating a new entry if necessary.
      It seems that quads must be normalized, so that literals with different syntactic representations but the same semantic representations are merged, and that two graphs differing in the syntactic representation of a literal will produce the same set of blank node identifiers.
  3. Create a list of non-normalized blank node identifiers non-normalized identifiers and populate it using the keys from the blank node to quads map.
  4. Initialize simple, a boolean flag, to true.
  5. While simple is true, issue canonical identifiers for blank nodes:
    1. Set simple to false.
    2. Clear hash to blank nodes map.
    3. For each blank node identifier identifier in non-normalized identifiers:
      1. Create a hash, hash, according to the Hash First Degree Quads algorithm.
      2. Add hash and identifier to hash to blank nodes map, creating a new entry if necessary.
    4. For each hash to identifier list mapping in hash to blank nodes map, lexicographically-sorted by hash:
      1. If the length of identifier list is greater than 1, continue to the next mapping.
      2. Use the Issue Identifier algorithm, passing canonical issuer and the single blank node identifier in identifier list, identifier, to issue a canonical replacement identifier for identifier.
      3. Remove identifier from non-normalized identifiers.
      4. Remove hash from the hash to blank nodes map.
      5. Set simple to true.
  6. For each hash to identifier list mapping in hash to blank nodes map, lexicographically-sorted by hash:
    1. Create hash path list where each item will be a result of running the Hash N-Degree Quads algorithm.
    2. For each blank node identifier identifier in identifier list:
      1. If a canonical identifier has already been issued for identifier, continue to the next identifier.
      2. Create temporary issuer, an identifier issuer initialized with the prefix _:b.
      3. Use the Issue Identifier algorithm, passing temporary issuer and identifier, to issue a new temporary blank node identifier for identifier.
      4. Run the Hash N-Degree Quads algorithm, passing temporary issuer, and append the result to the hash path list.
    3. For each result in the hash path list, lexicographically-sorted by the hash in result:
      1. For each blank node identifier, existing identifier, that was issued a temporary identifier by identifier issuer in result, issue a canonical identifier, in the same order, using the Issue Identifier algorithm, passing canonical issuer and existing identifier.
  7. For each quad, quad, in input dataset:
    1. Create a copy, quad copy, of quad and replace any existing blank node identifiers using the canonical identifiers previously issued by canonical issuer.
    2. Add quad copy to the normalized dataset.
  8. Return the normalized dataset.

Issue Identifier Algorithm



This algorithm issues a new blank node identifier for a given existing blank node identifier. It also updates state information that tracks the order in which new blank node identifiers were issued.

This algorithm takes an identifier issuer and an existing identifier as inputs. The output is a new issued identifier. The steps of the algorithm are:

  1. If there is already an issued identifier for existing identifier in issued identifiers list, return it.
  2. Generate issued identifier by concatenating identifier prefix with the string value of identifier counter.
  3. Append an item to issued identifiers list that maps existing identifier to issued identifier.
  4. Increment identifier counter.
  5. Return issued identifier.

Hash First Degree Quads

Add note that the result of this algorithm for a particular blank node will always be the same. This is only true because there was a typo in the spec that has now been implemented by many implementations. The design of the algorithm was to use the assigned canonical blank node identifier, if available, instead of _:a or _:z, similar to how it is used in the related hash algorithm, but this text never made it into the spec before implementations moved forward. Therefore, the hashes here never change, making the loop based on the simple flag that calls this algorithm unnecessary; it needs to only run once. A future version of this algorithm should correct this mistake.



This algorithm takes the canonicalization state and a reference blank node identifier as inputs.

  1. Initialize nquads to an empty list. It will be used to store quads in N-Quads format.
  2. Get the list of quads quads associated with the reference blank node identifier in the blank node to quads map.
  3. For each quad quad in quads:
    1. Serialize the quad in N-Quads format with the following special rule:
      1. If any component in quad is an blank node, then serialize it using a special identifier as follows:
        1. If the blank node's existing blank node identifier matches the reference blank node identifier then use the blank node identifier _:a, otherwise, use the blank node identifier _:z.

      Note potential need to normalize literals to their canonical representation here as well, if not done on the original input dataset.

  4. Sort nquads in lexicographical order.
  5. Return the hash that results from passing the sorted, joined nquads through the hash algorithm.

Hash Related Blank Node



This algorithm creates a hash to identify how one blank node is related to another. It takes the canonicalization state, a related blank node identifier, a quad, an identifier issuer, issuer, and a string position as inputs.

  1. Set the identifier to use for related, preferring first the canonical identifier for related if issued, second the identifier issued by issuer if issued, and last, if necessary, the result of the Hash First Degree Quads algorithm, passing related.
  2. Initialize a string input to the value of position.
  3. If position is not g, append <, the value of the predicate in quad, and > to input.
  4. Append identifier to input.
  5. Return the hash that results from passing input through the hash algorithm.

Hash N-Degree Quads

The relationship and difference between this algorithm and the hash first degree quads algorithm should be better explained. There may also be better names for the two algorithms.

The 'path' terminology could also be changed to better indicate what a path is (a particular deterministic serialization for a subgraph/subdataset of nodes without globally-unique identifiers).


Usually, when trying to determine if two nodes in a graph are equivalent, you simply compare their identifiers. However, what if the nodes don't have identifiers? Then you must determine if the two nodes have equivalent connections to equivalent nodes all throughout the whole graph. This is called the graph isomorphism problem. This algorithm approaches this problem by considering how one might draw a graph on paper. You can test to see if two nodes are equivalent by drawing the graph twice. The first time you draw the graph the first node is drawn in the center of the page. If you can draw the graph a second time such that it looks just like the first, except the second node is in the center of the page, then the nodes are equivalent. This algorithm essentially defines a deterministic way to draw a graph where, if you begin with a particular node, the graph will always be drawn the same way. If two graphs are drawn the same way with two different nodes, then the nodes are equivalent. A hash is used to indicate a particular way that the graph has been drawn and can be used to compare nodes.

This algorithm works in concert with the main canonicalization algorithm to produce a unique, deterministic identifier for a particular blank node. This hash incorporates all of the information that is connected to the blank node as well as how it is connected. It does this by creating deterministic paths that emanate out from the blank node through any other adjacent blank nodes.


An additional input to this algorithm should be added that allows it to be optionally skipped and throw an error if any equivalent related hashes were produced that must be permuted during step 5.4.4. For practical uses of the algorithm, this step should never be encountered and could be turned off, disabling canonizing datasets that include a need to run it as a security measure.

The inputs to this algorithm are the canonicalization state, the identifier for the blank node to recursively hash quads for, and path identifier issuer which is an identifier issuer that issues temporary blank node identifiers. The output from this algorithm will be a hash and the identifier issuer used to help generate it.

  1. Create a hash to related blank nodes map for storing hashes that identify related blank nodes.
  2. Get a reference, quads, to the list of quads in the blank node to quads map for the key identifier.
  3. For each quad in quads:
    1. For each component in quad, where component is the subject, object, or graph name, and it is a blank node that is not identified by identifier:
      1. Set hash to the result of the Hash Related Blank Node algorithm, passing the blank node identifier for component as related, quad, path identifier issuer as issuer, and position as either s, o, or g based on whether component is a subject, object, graph name, respectively.
      2. Add a mapping of hash to the blank node identifier for component to hash to related blank nodes map, adding an entry as necessary.
  4. Create an empty string, data to hash.
  5. For each related hash to blank node list mapping in hash to related blank nodes map, sorted lexicographically by related hash:
    1. Append the related hash to the data to hash.
    2. Create a string chosen path.
    3. Create an unset chosen issuer variable.
    4. For each permutation of blank node list:
      1. Create a copy of issuer, issuer copy.
      2. Create a string path.
      3. Create a recursion list, to store blank node identifiers that must be recursively processed by this algorithm.
      4. For each related in permutation:
        1. If a canonical identifier has been issued for related, append it to path.
        2. Otherwise:
          1. If issuer copy has not issued an identifier for related, append related to recursion list.
          2. Use the Issue Identifier algorithm, passing issuer copy and related and append the result to path.
        3. If chosen path is not empty and the length of path is greater than or equal to the length of chosen path and path is lexicographically greater than chosen path, then skip to the next permutation.
      5. For each related in recursion list:
        1. Set result to the result of recursively executing the Hash N-Degree Quads algorithm, passing related for identifier and issuer copy for path identifier issuer.
        2. Use the Issue Identifier algorithm, passing issuer copy and related and append the result to path.
        3. Append <, the hash in result, and > to path.
        4. Set issuer copy to the identifier issuer in result.
        5. If chosen path is not empty and the length of path is greater than or equal to the length of chosen path and path is lexicographically greater than chosen path, then skip to the next permutation.
      6. If chosen path is empty or path is lexicographically less than chosen path, set chosen path to path and chosen issuer to issuer copy.
    5. Append chosen path to data to hash.
    6. Replace issuer, by reference, with chosen issuer.
  6. Return issuer and the hash that results from passing data to hash through the hash algorithm.

Use Cases





A previous version of this algorithm has light deployment. For purposes of identification, the algorithm is called the "Universal RDF Graph Canonicalization Algorithm 2012" (URGNA2012), and differs from the stated algorithm in the following ways:


The editors would like to thank Jeremy Carroll for his work on the graph canonicalization problem, Gavin Carothers for providing valuable feedback and testing input for the algorithm defined in this specification, Sir Tim Berners Lee for his thoughts on graph canonicalization over the years, Jesús Arias Fisteus for his work on a similar algorithm.