edited by Birger Hjørland and Claudio Gnoli


IFLA Library Reference Model (LRM)
Harmonisation of the FRBR Family

by Maja Žumer

Table of contents:
1. Introduction
2. Principles of the consolidation process
3. User tasks
4. Entities
5. Attributes
6. Relationships
7. Aggregates
8. Conclusion

The IFLA Library Reference Model (IFLA LRM) consolidates the three models of the FRBR Family. First the major modelling and presentation issues are identified. The main part is the general description of IFLA LRM. Only the most important features are presented, with examples illustrating the modelling approaches.

1. Introduction

In 1998 the FRBR model (Functional Requirements for Bibliographic Records 1998) was developed under the auspices of the International Federation of Library Associations and Institutions (IFLA). The library domain finally developed its conceptual model of the bibliographic universe and thus the basis for the development of novel bibliographic information systems — such as library catalogues and bibliographies — which would better suit the changing environment. FRBR was a revolutionary step forward, influencing the foundations of cataloguing theory and practice. In the following years two complementary models were developed, focusing on authority data: Functional Requirements for Authority Data (Patton 2009), dealing with name authority and Functional Requirements for Subject Authority Data (Zeng, Žumer and Salaba 2011), focusing on the → subject relationship. The three models, usually referred to as the FRBR Family of models, were developed over a rather long period by different working groups. It is therefore not surprising that some different modelling decisions were made, resulting in some incompatible details. While relatively minor, these differences still present a major barrier for the development of implementations.

FRBR Review Group, the IFLA body responsible for the development and maintenance of the FRBR Family, started the consolidation process in 2011 and, in 2013, formally established the Consolidation Editorial Group (CEG). The task of CEG was to systematically and consistently combine the three models and thus create a unified model of the bibliographic universe.

In the beginning of 2016, CEG finished the first stable draft of the LRM model, which was issued for a two-month world-wide review, according to IFLA practice. All comments were collected and discussed and subsequently the CEG incorporated the revisions into the draft, which was then reviewed by the full FRBR Review Group at its annual meeting in August 2016. The Review Group made decisions on all outstanding issues, leading to a final draft accepted at the FRBR Review Group level by the end of 2016. In accordance with the IFLA standards process, this final draft was submitted for approval to the IFLA Committee on Standards in April 2017 and posted on the IFLA website. IFLA Professional Committee formally adopted IFLA LRM on August 18, 2017.

All figures in this article are from the IFLA LRM document.

[top of entry]

2. Principles of the consolidation process

The task of the CEG was to:

  • Prepare a high-level abstract model
  • Use the entity-relationship formalism
  • Develop a consistent model consolidating all three models of the FRBR Family
  • Consider implementation in the Semantic Web

The development of LRM was informed by user research (such as Pisanski and Žumer 2010a; 2010b), the work of FRBR/CRM Harmonisation Group and the resulting FRBRoo (Bekiari et al. 2015), experiences of FRBR/FRAD/FRSAD implementations and the Semantic Web and Linked Data environment.

The resulting model is described as:

The conceptual model as declared in IFLA LRM is a high-level conceptual model and as such is intended as a guide or basis on which to formulate cataloguing rules and implement bibliographic systems. Any practical application will need to determine an appropriate level of precision, requiring either expansion within the context of the model, or possibly some omissions. However, for an implementation to be viewed as a faithful implementation of the model, the basic structure of the entities and the relationships among them (including the cardinality constraints), and the attachment of those attributes implemented, needs to be respected. ((Riva, LeBoeuf, and Žumer 2017, 6)

[top of entry]

3. User tasks

The FRBR Family of models as well as LRM all start from the user tasks that need to be enabled and supported by bibliographic information systems. These user tasks provide the boundaries of the model and serve as the starting point for definitions of entities, attributes and relationships. Bibliographic and authority data are of interest to a broad user group — from library users (readers, researchers, students etc.) to librarians and other members of the information chain, such as publishers and booksellers. These user groups have different needs and different priorities. LRM follows FRBR in the choice of its primary user group: end-users and librarians acting on their behalf. Librarians creating and maintaining metadata may occasionally perform the same tasks as part of their work activities; they are included in this sense. On the other hand the model does not cover administrative data important for library operations, such as intellectual rights data, preservation data or acquisition processes. It also has to be mentioned that in this sense the scope of FRAD is different from the other two models, because it also models the cataloguing process, which is reflected in the tasks, particularly ‘justify’.

LRM defines five user tasks (Table 1) and lists the goals users want to reach when performing the tasks. The term resource is used in its broadest sense, meaning an instance of any entity defined in the model. The tasks are listed in an order that reflects typical user behaviour, which does not mean that all tasks need to be performed and that they cannot be repeated. Particularly identify and obtain often occur in parallel and in interaction.

Table 1: A summary of user tasks
Find To bring together information about one or more resources of interest by searching on any relevant criteria
Identify To clearly understand the nature of the resources found and to distinguish between similar resources
Select To determine the suitability of the resources found, and to be enabled to either accept or reject specific resources
Obtain To access the content of the resource
Explore To discover resources using the relationships between them and thus place the resources in a context

The first four tasks are the same as in FRBR, with slightly modified and broader definitions; explore, on the other hand, was first introduced by FRSAD. The need for ‘navigation’ was already mentioned in FRBR and in the following years researchers often emphasised that a modern bibliographic information system needs to support browsing and, consequently, serendipitous discovery of relevant resources.

[top of entry]

4. Entities

In an entity-relationship model, entities are defined as key objects of interest. They are abstract categories (also called classes) of conceptual objects, connected by relationships, and their characteristics are described by attributes.

The consolidation process included a detailed analysis of all entities defined by the FRBR Family, since they include virtually identical entities across models (Work, Expression, Manifestation, Item), similar entities (FRSAD Nomen and FRAD Name) or completely different ones (Person in FRBR and in FRAD). The decision was made to include only the entities with specific attributes and relationships. What is new in LRM is also the hierarchical structure of classes and subclasses, usually expressed with the isA relationship in formal modelling. This is a powerful mechanism enabling considerable simplification: attributes and relationships, declared on a higher level, are inherited by all subclasses and do not have to be repeated on lower levels.

The entities of the FRBR Group 1 remain the same; there are some minor differences in definitions and scope notes, though (Table 2).

Table 2: Work, Expression, Manifestation, Item
Work The intellectual or artistic content of a distinct creation
Expression A distinct combination of signs conveying intellectual or artistic content
Manifestation A set of all carriers that are assumed to share the same characteristics as to intellectual or artistic content and aspects of physical form. That set is defined by both the overall content and the production plan for its carrier or carriers.
Item An object or objects carrying signs intended to convey intellectual or artistic content

By declaring the superclass Agent, a hierarchical structure is introduced in the former FRBR Group 2 of entities. Since they are clearly subsumed in the entity Collective agent and have no specific attributes and relationships, Corporate body and Family are not included as entities in LRM (Table 3).

Table 3: Agents
Agent An entity capable of deliberate actions, of being granted rights, and of being held accountable for its actions
Person Individual human being
Collective agent A gathering or organization of persons bearing a particular name and capable of acting as a unit

The “agent” entities can best be presented showing the basic relationships (Fig. 1).

Figure 1: Relationships between Agent, Person and Collective agent

Agent therefore includes only entities Person and Collective agent and not any other named groups. LRM here follows FRBR in limiting Person to living persons and those who are assumed to have lived. Fictional, literary and legendary persons are therefore not included. They may be subjects of Works, but when they seem to appear as creators, it is in fact a real person or collective agent using this appellation in the context of the creation process. The name used does not change the nature of this person or collective agent.

The FRSAD model introduced two basic entities, Thema and Nomen to model the appellation relationship. LRM keeps both of them, with a change in label. Res is used instead of Thema to avoid the implicit restriction to the subject relationship.

Res is the superclass of all entities in the model and Nomen is the appellation used to refer to an instance of Res. Modelling Nomen as an entity enables us to assign specific attributes such as script, language or source vocabulary to appellations and establish relationships between them, such as the relationship between former and later name of a Person (Table 4).

Table 4: Res and Nomen
Res Any entity in the universe of discourse
Nomen An association between an entity and a designation that refers to it

In order to model more precisely the temporal and spatial aspects, LRM introduces two additional entities, Place and Time-span (Table 5).

Table 5: Place and time-span
Place A given extent of space
Time-span A temporal extent having a beginning, an end and a duration

[top of entry]

5. Attributes

Attributes enable assigning values to characteristics of entities. The FRBR Family of models treated attributes with various levels of detail. Since LRM has to cover all types of library materials, the decision was made to include only the most frequent and general attributes; the list is therefore not exhaustive and none of the attributes are mandatory. In an implementation attributes may be added to record additional relevant characteristics or add more detail to the existing attributes. Cataloguing rules determine how the attributes values are determined and their values: from a controlled vocabulary, as free text in a particular language and script, as a numeric value. Multiple values of attributes are possible.

Category and Note are the two attributes declared for Res and are therefore inherited by all entities of the model.

As an illustration, Tables 6 and 7 show attributes of Expression and Manifestation respectively.

Table 6: Attributes of Expression
Extent A quantification of the extent of the expression
Intended audience A class of users for which the expression is intended
Use rights A class of use restrictions to which the expression is submitted
Language A language used in the expression
Key A pitch structure (musical scale, ecclesiastic mode, raga, maqam, etc.), that characterizes the expression
Medium of performance A combination of performing tools (voices, instruments, ensembles, etc.) stated, intended, or actually used in the expression
Cartographic scale A ratio of distances in a cartographic expression to the actual distances they represent

Table 7: Attributes of Manifestation
Category of carrier A type of material to which all physical carriers of the manifestation are assumed to belong
Extent A quantification of the extent observed on a physical carrier of the manifestation and assumed to be observable on all other physical carriers of the manifestation as well
Intended audience A class of users for which the physical carriers of the manifestation are intended
Manifestation statement A statement appearing in exemplars of the manifestation and deemed to be significant for users to understand how the resource represents itself
Access Information as to how any of the carriers of the manifestation are likely to be obtained
Use rights A class of use and/or access restrictions to which all carriers of the manifestation are assumed to be submitted

Two attributes need to be mentioned in particular: Representative expression attribute (Work attribute) and Manifestation statement (Manifestation attribute).

Representative expression attributes are the attributes that are deemed essential in characterizing the work and whose values are taken from a representative or canonical expression of the work. According to user studies the key characteristics of a work are associated with the original or canonical expression, which is considered to be the best representation of the work (Pisanski and Žumer 2010a; 2010b). The values of these attributes are inferred either from a particular expression considered to best represent the work, or from characteristics abstracted from set of similar expressions. There is no requirement to precisely identify an expression or expressions which serves as source, nor does that expression need to be recorded in the system.

Manifestation statement provides a mechanism to record information found on a manifestation, which is important to understand how the resource represents itself. Typical examples are responsibility statements; not always complete, sometimes even fictitious, but nonetheless important to identify a manifestation. LRM thus enables both the transcription from the manifestation itself and relationships to creators (e.g. publishers), places (of publication) and time-spans (dates of publication).

[top of entry]

6. Relationships

Relationships are an essential component of the model, connecting entities and placing them in context. Some relationships (such as between Works, Expressions, Manifestations and Items) are virtually identical in all three models, others differ particularly in the level of detail. In line with other modelling decisions, relationships in LRM are modelled on a general and abstract level. Again, for an implementation, additional relationships may be added within the general framework when needed. Since relationships support exploration, they are essential in every implementation and should be recorded as much as possible.

Relationships between Works, Expressions, Manifestations and Items remain the core of the model and are in essence required. It is also important to mention, though, that while relationships are declared between entity types, they really occur between instances.

Formally a relationship is declared between its domain and range and always in both directions. If the domain and range are the same, the relationship is recursive. When a relationship is the same in both directions, it is called symmetrical.

Cardinality is another term that needs to be explained: it specifies the number of instances of the domain and range for each relationship. The cardinality 1 to M (meaning ‘many’) means that one instance of domain is related to many instances of range and, consequently for the reverse relationship, many instances are related to one and only one instance of the range.

The relationships between Works, Expressions, Manifestations and Item are shown in Fig. 2.

Figure 2: Relationships between Works, Expressions, Manifestations and Items

As an example, all work-to-work relationships are shown in Table 8.

Table 8: Work-to-work relationships
Domain Relationship name Inverse name Range Cardinality
Work has part is part of Work M to M
Work precedes succeeds Work M to M
Work accompanies / complements is accompanied / complemented by Work M to M
Work is inspiration for is inspired by Work M to M
Work is a transformation of was transformed into Work M to 1

Each instance of an entity in the model has one or more appellations, the relationship between entities Res and Nomen has to be established (Table 9).

Table 9: Appellation relationship
Domain Relationship name Inverse name Range Cardinality
Res has appellation is appellation of Nomen 1 to M

Agent as a new entity enabled a considerably simplified model of responsibility relationships (Table 10).

Table 10: Responsibility relationships
Domain Relationship name Inverse name Cardinality
Work was created by created Agent M to M
Expression was created by created Agent M to M
Manifestation was created by created Agent M to M
Manifestation was manufactured by manufactured Agent M to M
Manifestation is distributed by distributes Agent M to M
Item is owned by owns Agent M to M
Item was modified by modified Agent M to M

Modelling of the subject relationship, introduced by FRSAD, remains essentially the same (Fig. 3).

Figure 3: Subject relationship

Since Place and Time-span were introduced as entities and any entity in the model may have a temporal and/or spatial component, two general relationships are necessary (Table 11).

Table 11: Temporal and spatial relationships
Domain Relationship name Inverse name Range Cardinality
Res has association with is associated with Place M to M
Res has association with is associated with Time-span M to M

An overview of all LRM relationships is shown in Figure 4. The isA relationships between all entities and the entity Res is not shown. For the sake of simplicity, relationships are shown in one direction only.

Figure 4: Overview of LRM relationships. The implicit isA relationship of all entities to Res is not shown. For simplicity, the relationships are shown in one direction only and cardinality is not specified

[top of entry]

7. Aggregates

Aggregates are manifestations, embodying more than one distinct expression. The FRBR model does not provide clear guidance on how to model such publications, which in general are quite common. According to the Aggregates report (Working Group on Aggregates 2011), three types of aggregates exist:

  • Aggregate collections of expressions
    Aggregate collections are manifestations embodying distinct expressions of the same type or genre. Typical examples include collected and selected works and anthologies, but also journals (aggregates of articles), monographs with independently created chapters, compilations of music recordings on a CD, several essays published in a book etc.
  • Aggregates resulting from augmentation
    This type of aggregates occur when additional content is added to a distinct expression, such as illustrations, forewords, notes, commentaries
  • Aggregates of parallel expressions
    Aggregates of parallel expressions occur, when two or more expressions of the same work are embodied in a manifestation. Typical examples are multilingual manuals and official publications, an original published with a translation, but also multilingual websites.

Cardinality of the ‘Expression is embodied in Manifestation’ relationship, which is many-to-many, explicitly indicates that a manifestation may include several expressions. Research (O’Neill, Žumer and Mixter 2015) shows that aggregates are, even as identified from existing bibliographic records, very common. Considering that current cataloguing practice does not particularly encourage systematical recording of components of collections and augmentations such as illustrations or forewords, the actual occurrence of aggregates is likely much higher.

Modelling of aggregates as manifestations embodying several distinct expressions is straightforward, since works and their respective expressions are completely independent of the manifestations they are published in. An expression does not change when embodied on its own or with other expressions in a manifestation.

When modelling aggregates, we need to consider not only the aggregated expressions, but also the intellectual contribution of selecting and arranging the expressions. This intellectual contribution should be considered a work on its own right and it is called ‘aggregating work’ in the model. It may be rather insignificant in the case of combining two recordings on a CD, but it may also be essential for a publication like an anthology. It has to be emphasized that the aggregating work does not include the works being aggregated. When an aggregating work is not significant enough, it is typically not recorded. The same is true for minor augmentations such as a brief foreword. But if such a foreword is republished as an independent essay, it should also be recorded in the aggregate.

The model of aggregates is shown in Fig. 5.

Figure 5: Aggregates

[top of entry]

8. Conclusion

This is a short presentation of the IFLA LRM model. It does not include all the necessary details and is not complete in any way. To understand the model in full, the readers should consult the full model description as published on the IFLA website (IFLA LRM 2017).

IFLA LRM presents an important step forward; we now finally have a complete model of the bibliographic universe, which can and should serve as the foundation for the development of cataloguing rules and bibliographic formats. The next steps include the declaration of namespaces, which will enable Semantic Web compliant implementations and mapping to existing namespaces. Important future tasks include extensions for specific material types, different target audiences and other circumstances important for the design of bibliographic information systems.

With IFLA LRM we finally have a modern model, compatible with the Semantic Web. Only with an immediate development of new library catalogs we can hope to exploit fully the wealth of library data and stop the trend of decreasing use (or even avoidance) of current catalogues.

[top of entry]


Bekiari, C. et al. (Eds.). 2015. Definition of FRBRoo : a conceptual model for bibliographic information in object-oriented formalism / International Working Group on FRBR and CIDOC CRM Harmonisation. Version 2.4. November 2015. http://www.ifla.org/files/assets/cataloguing/FRBRoo/frbroo_v_2.4.pdf

Final Report of the Working Group on Aggregates. 2011. https://www.ifla.org/files/assets/cataloguing/frbrrg/AggregatesFinalReport.pdf

Functional requirements for bibliographic records: final report. 1998. München: K.G. Saur. (UBCIM publications; new series, vol. 19). As amended and corrected through February 2009. http://www.ifla.org/files/assets/cataloguing/frbr/frbr_2008.pdf

O'Neill, Ed T., Maja Žumer, and Jeff Mixter. 2015. “FRBR aggregates: their types and frequency in library collections”. Library Resources & Technical Services, 59(3), 120-129.

Patton, Glenn E. (Ed.). 2009. Functional requirements for authority data: a conceptual model. München: K.G. Saur. (IFLA series on bibliographic control ; vol. 34). As amended and corrected through July 2013. http://www.ifla.org/files/assets/cataloguing/frad/frad_2013.pdf

Pisanski, Jan and Maja Žumer. 2010a. “Mental models of the bibliographic universe: part 1: mental models of descriptions”. Journal of Documentation, 66(5), 643–667. 

Pisanski, Jan and Maja Žumer. 2010b. “Mental models of the bibliographic universe: part 2: comparison task and conclusions”. Journal of Documentation, 66(5), 668–680.

Riva, Pat, Patrick LeBoeuf, and Maja Žumer. 2017. IFLA Library Reference Model. https://www.ifla.org/files/assets/cataloguing/frbr-lrm/ifla-lrm-august-2017.pdf

Zeng, Marcia Lei, Maja Žumer, and Athena Salaba (Eds.). 2011. Functional requirements for subject authority data (FRSAD) : a conceptual model. München: De Gruyter Saur. (IFLA series on bibliographic control ; vol. 43). http://www.ifla.org/files/assets/classification-and-indexing/functional-requirements-for-subject-authority-data/frsad-final-report.pdf. Errata for section 5.4.2, October 2011, http://www.ifla.org/files/assets/cataloguing/frsad/FRSADerrata2011.pdf

[top of entry]


Visited Hit Counter by Digits times since 2019-11-14 (2 years after first publication).

Version 1.1 (= version 1.0 with uniformed reference format); published 2017-11-07, last edited 2019-11-14
Article category: KOS general issues

This article (version 1.1) is also published in Knowledge Organization. How to cite it:
Žumer, Maja. 2018. "IFLA Library Reference Model (IFLA LRM): Harmonisation of the FRBR Family". Knowledge Organization 45, no. 4: 310-8. Also available in Hjørland, Birger and Gnoli, Claudio eds., ISKO Encyclopedia of Knowledge Organization, http://http://www.isko.org/cyclo/lrm

©2017 ISKO. All rights reserved.