5.2 Locating resources by subject

Module 04 covered standards for metadata elements and their format. The focus now is on standardised vocabularies, particularly for describing the subject of information resources. These vocabularies are essential for effective subject access and are often more sophisticated and comprehensive, specifically designed to improve access to collections.

Information resources are often described using everyday ‘natural language’, such as abstracts or book titles. However, for greater consistency between the describer and searcher, controlled vocabularies are sometimes used instead. These standardised vocabularies, developed by librarians and other professionals, help ensure uniformity and precision, with some of the key examples being highlighted within Module 05.

Subject access

People seeking information often aim to find material on a specific subject. Librarians are accustomed to queries such as, ‘Do you have any books on Egyptology?’ or ‘I’m visiting China next year. What information do you have on China?’. Similarly, archivists may receive requests like, ‘I want to see everything you have on the ships that brought passengers to Sydney in 1843’. Research on library users shows that over half of online catalogue searches are for subjects. As libraries, archives, and other information centres are committed to providing relevant and timely information, it is essential that they develop effective tools for subject access.

Subject access tools can utilise different types of vocabularies (diagram below), ranging from uncontrolled to highly controlled. Tools that use these vocabularies, whether controlled or not, to help locate information resources are also referred to as indexing languages.

Nearly all of the words in the above diagram, apart from perhaps the last box, share a common feature: they involve words, i.e., text. Text-based information retrieval methods have been the most prevalent in both past and present practices.

Another helpful approach to understanding how we provide subject access to information resources is to consider the source of the words or text used in searches.

Derived indexing extracts words directly from the document itself, whereas assigned indexing uses words from an external source, usually a controlled vocabulary, to represent the document’s content.

It is important to remember that subject access is often provided using multiple methods. This is demonstrated in Figure 5.2, where a library catalogue record employs three distinct types of subject description, each utilising different vocabularies.

Figure 5.3 is a bibliographic record as it would appear in the SCIS catalogue.

SCIS catalogue records use several forms of subject description: natural language keywords found anywhere in the record, controlled vocabulary subject headings and thesaurus terms in the Subjects element, and controlled language classification numbers in the Call Numbers element. In school library catalogues, keyword searches may be limited to specific fields, such as title and subject, depending on the library management system used.

The controlled vocabularies used in the Subjects element include SCIS Subject Headings and Schools Online Thesaurus (ScOT) terms. SCIS Subject Headings are marked by the acronym ‘scisshl’, while Schools Online Thesaurus terms are indicated by ‘scot’, though these acronyms are not usually visible in school catalogues.

For classification, the Call Numbers element utilises the Abridged Dewey Decimal Classification, edition 15 (ADDC15), and the Dewey Decimal Classification, edition 23 (DDC23). The call numbers shown are based on SCIS standards for cataloguing and data entry. In the MARC view of a SCIS record, Figure 5.4, you might see ‘a15′ at the end of a Dewey classification number, indicating it is from ADDC15, and ’23’ indicating it is from DDC23.

Older SCIS records may include call numbers from previous editions of the DDC. In school library catalogues, usually only one call number is displayed, based on the chosen classification level, and the DDC edition indicators ‘a15′ or ’23’ are not typically visible.

Controlled vocabularies

Different people use different terms to describe the same things, including subjects, and would index and search using these varying terms unless we standardise (control) the indexing language. We have already discussed the benefits of controlling indexing terms. To illustrate, consider buying land to build a house. In New Zealand, this piece of land is referred to as a ‘section’, while in Australia it is called a ‘block’, ‘lot’, or ‘plot’. Similarly, a small holiday house in New Zealand is known as a ‘bach’ or ‘crib’, whereas in Australia it is called a ‘cottage’ or ‘shack’. If the house is elevated, the supporting structures are called ‘piles’ in New Zealand and ‘stumps’ in Australia.

As shown in Figure 5.1, controlled vocabularies are not the only way to provide subject access; other methods may be effective in different circumstances. One major drawback of controlled vocabularies is their cost—they can be expensive to implement and maintain. Since language evolves and new resources are continually produced, vocabularies need to be regularly updated, which can be costly. Despite this, the investment is often worthwhile, as it enhances access to valuable resources that might otherwise be overlooked.

The three main types of controlled vocabularies traditionally used in libraries and information centres are:

1. Subject headings lists
2. Subject thesauri
3. Subject classification schemes

Classification schemes are not typically used for indexing but for organising materials—such as arranging books on shelves—so that related resources are placed together. Instead of words, these schemes use notations to represent topics. For instance, in the Dewey Decimal Classification, ‘025’ represents ‘library operations’, so all books on this topic are grouped in that section.

In contrast, subject headings lists and thesauri are mainly used for indexing and rely on verbal language.

Read

‘Controlled subject vocabularies’ (pp. 175-178), in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

Controlled Subject Vocabularies Summary

Controlled subject vocabularies are used to standardise the terminology applied in describing the subject matter of information resources, which enhances the precision and consistency of information retrieval. Unlike uncontrolled natural language, where the same concept can be described in multiple ways, controlled vocabularies ensure that each concept is represented by one specific term, and each term refers to only one concept. This standardisation is beneficial for both indexers and users, as it allows for consistent indexing and more effective searches. However, the application of controlled vocabularies can be challenging due to their limitations and the subjective nature of identifying a resource’s subject matter.

One of the primary issues with controlled vocabularies is determining the appropriate level of coverage and specificity for a resource. For example, a book might be broadly described as being about ‘dogs’, but it could also include specific chapters on subtopics such as ‘dog breeds’, ‘dog training’, or ‘canine health’. Indexers must decide whether to describe the resource as a whole or to include these subtopics, a practice referred to as the level of ‘exhaustivity’. The question of how much detail to include in the subject description is crucial, as it affects how well the resource can be retrieved by users. For instance, if a resource includes substantial content on both volcanoes and earthquakes, both topics should be indexed to reflect the main themes adequately.

Specificity is another significant challenge. The principle is that the index term should match the level of detail of the topic. For instance, a book about ‘dogs’ should be indexed as ‘dogs’ rather than the broader category ‘pets’ or the overly specific ‘golden retriever’. The risk of using overly broad or overly narrow terms is that it can either generalise too much or provide an interpretation that is too specific for the intended use. For example, a picture of a ‘golden retriever’ should be indexed as ‘dogs’ if it is primarily used to represent dogs in general rather than the specific breed. In such cases, using both broader and narrower terms may be necessary to cover the resource comprehensively.

Controlled vocabularies can be applied using different approaches, such as pre-coordination or post-coordination. In pre-coordination, terms are combined into specific strings before indexing and searching. For example, the term ‘Birds – Conservation – Australia’ pre-coordinates three concepts into one structured term. This approach provides greater precision but is less flexible. In contrast, post-coordination allows terms to be combined during the search process, such as using ‘Birds’ and ‘Australia’ separately and then linking them in the search query. This method is more flexible and often easier to use, especially with modern search technologies.

The application of controlled vocabularies also involves translating the analysis of a resource’s subject into controlled terms, which can be influenced by how the subject itself is defined. For example, a resource might address multiple related topics, but not all of them may be represented in the controlled vocabulary. If a book covers both ‘volcanoes’ and ‘earthquakes’ extensively, indexers might decide to represent both subjects. However, they must also consider whether sub-topics, such as ‘lava flows’ or ‘quake-proof buildings’, should be included, depending on how much these aspects are covered.

In practice, controlled vocabularies can significantly improve access to resources, but they require careful application and regular updates to stay relevant as language evolves and new subjects emerge. The use of controlled vocabularies, while sometimes costly and complex to maintain, is often justified by the improved retrieval accuracy and the ability to make valuable resources more discoverable and accessible to users.

Read

‘Subject headings’ (pp. 178-184), in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

Subject Headings

The section on “Subject Headings” explains the development and use of controlled subject vocabularies in libraries. These vocabularies originated in the 19th century alongside descriptive cataloguing codes, which were discussed in an earlier chapter. The aim was to help patrons who sought information on specific topics but didn’t know the exact title of a book. Melvil Dewey and Charles Ammi Cutter were instrumental in establishing controlled subject headings to organise library collections, starting with Cutter’s Rules for a Dictionary Catalog in 1876.

Controlled vocabularies, such as the Library of Congress Subject Headings (LCSH), were created to ensure consistency in the way subjects were described. These controlled terms became essential as libraries began distributing record cards that standardised how topics were indexed. The LCSH, for example, continues to be widely used today, containing hundreds of thousands of subject headings and cross-references.

LCSH and other controlled vocabularies allow resources to be grouped systematically under specific subject headings, regardless of how the title might describe them. While this offers benefits such as consistency and easier searching, it also has limitations. Some critics argue that LCSH is biased towards American culture and language, and that it doesn’t always reflect the diverse range of topics covered in modern collections. Nevertheless, LCSH remains one of the most comprehensive and widely used systems for organising library resources.

There are other subject heading lists like the Sears List of Subject Headings for smaller libraries, and specialised lists like Medical Subject Headings (MeSH) for health-related materials. Some national libraries have also developed their own controlled vocabularies, such as RAMEAU in France and Répertoire de vedettes-matières in Canada.

Subject headings can often include subdivisions to provide more specific information about a topic, such as geographic areas or time periods. The section highlights the importance of controlled vocabularies in ensuring that libraries can manage large collections effectively, allowing users to find resources on a particular subject through systematic and structured terms.

Thesauri

While subject thesauri, like subject heading lists, are essentially sets of terms used to describe subjects, these sets are systematically structured in a way that previous lists were not. The American National Information Standards Organization, NISO, in its Guidelines for the Construction, Format, and Management of Monolingual Controlled Vocabularies (ANZI/NISO Z39.19-2005 (R2010) defines a thesaurus as:

It goes on to say:

“the use of hierarchical relationships is the primary feature that distinguishes a thesaurus from other, simpler forms of controlled vocabularies.”

Thesauri have the same aims in whatever setting they’re used: to provide standardised terms and to structure these terms so that they can be readily navigated by both indexer and searcher. Most thesauri are used to cover particular subject areas, though some cover very broad subject areas.

Read

‘Subject thesauri’ (pp. 185-188), in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

Subject Thesauri

The “Subject Thesauri” section explains a form of subject vocabulary often used by library cataloguers and database indexers, which was developed in the 1960s with automated retrieval in mind, unlike the older card index system. Instead of using traditional headings, thesauri employ “descriptors” and feature structures based on systematic cross-referencing. This makes them distinct from subject headings lists like the Library of Congress Subject Headings (LCSH), which primarily compile terms without the deep interrelationships that thesauri emphasise.

A subject thesaurus is similar to a standard language thesaurus, such as Roget’s, but more selective, focusing on specific fields and the terms most likely to describe particular information resources. The inclusion of a term in a thesaurus can be based on two main types of “warrant”: literary warrant, which examines the actual content of the resources (the literature), and user warrant, which considers the terms and concepts people are searching for. Subject thesauri often balance these two aspects but may give more weight to user warrant than older systems like LCSH.

Thesauri are typically constructed by analysing the structure of knowledge in a particular field, identifying the various facets or conceptual components of the subject area. For example, in cooking, the facets might include method, ingredients, and cuisine style. The terms are then linked to one another through hierarchical (taxonomic) or associative relationships, and the concepts are given preferred terms or descriptors. Cross-references and scope notes are also commonly included to guide users and clarify meanings.

Subject thesauri are particularly valuable for indexing periodical articles, technical papers, and similar types of documents, which is why they are widely used in database indexing, though less so in standard library cataloguing. Examples of thesauri in use today include the ERIC Thesaurus for education and the STW Thesaurus for economics. They allow for precise searching and systematic access to specialised collections of information.

Some thesauri are developed with a specific focus, such as the Getty Art and Architecture Thesaurus (AAT) for visual arts, while others may be designed to support multilingual access, like the AGROVOC Thesaurus for agriculture. Despite their advantages, thesauri require significant intellectual effort to create and maintain, which is why they are generally only developed when no existing tool suffices.

Thesauri are increasingly available online and integrated into database systems to improve accessibility for both indexers and end-users, helping to maintain consistency in how resources are categorised and searched.

Classification schemes

The DDC is the most recognised and widely used library classification system worldwide. It is implemented in over 135 countries and translated into more than 30 languages, including Arabic, Chinese, French, Greek, Hebrew, Italian, Persian, Russian, Spanish, and Turkish. In the United States, 95% of public and school libraries, 25% of college and university libraries, and 20% of special libraries use the DDC system.

Read

‘Subject classification schemes’, pp. 189-201, in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

Subject Classification Schemes

The section on “Subject Classification Schemes” discusses the organisation of resources using artificial notations rather than natural language terms. Unlike subject headings or thesauri, classification schemes arrange groups of resources into a logical order, which helps users gain a broad overview of a collection’s subject coverage. By using notations (such as numbers or codes), similar topics can be placed near each other, aiding effective browsing.

While classification schemes can be created in various ways, they are time-consuming to develop and maintain. For this reason, libraries often opt for established schemes rather than developing new ones locally. Established schemes also provide the advantage of allowing for standardised notations that can be used in records sourced from external databases.

Subject classification schemes, like other vocabularies, are often based on principles such as literary and user warrant. In earlier times, some schemes followed an objectivist approach, which treated knowledge as static and waiting to be discovered. However, this view shifted in the 20th century to a more subjectivist perspective, recognising that knowledge depends on cultural and social perspectives, which evolve over time.

These classification schemes, similar to thesauri, are used to collocate resources by grouping them according to specific descriptors or notations. In addition to organising resources, classification schemes provide linear arrangements of subject groups, which facilitates browsing and the discovery of related resources in a library or digital collection.

In summary, subject classification schemes help arrange resources systematically by using notations rather than words, with well-established schemes like Dewey Decimal Classification (DDC) and Library of Congress Classification (LCC) serving as the most commonly used frameworks in libraries.

Taxonomies and ontologies

Read

‘Taxonomies’ and ‘Ontologies’ (pp. 201-205), in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

Taxonomies

In the context of information organisation, taxonomies are used to categorise online resources, such as websites and intranet content, into structured classes (or taxa). Unlike classification schemes for physical resources, taxonomies for online materials do not require notations to maintain order. This flexibility allows resources to be linked in multiple places within a website’s structure, enabling easier navigation and polyhierarchical systems, where a single resource may be accessible through various paths. However, overly complex webs of links can cause confusion for users.

Taxonomies are often developed based on the knowledge domain of a site and aim to provide an intuitive structure for users to navigate. Information architects play a key role in determining the best ways to arrange content, often relying on user warrant—how users think about and search for information. Websites or directories may use subject, organisational, or functional taxonomies depending on the content. Some large-scale taxonomies, like the Audit Commission Taxonomy, function as external standards for organising extensive collections.

Ontologies

Ontologies represent a more advanced form of knowledge organisation compared to taxonomies, characterised by greater complexity and precision. While taxonomies primarily show hierarchical relationships (e.g., “x is a type of y”), ontologies include formal associations and can describe intricate relationships between concepts. Ontologies are often employed in fields like computer science and artificial intelligence to model domains of knowledge with a structure that mimics human thought.

Ontologies go beyond simple hierarchical relationships by specifying associative relationships and applying inference rules. For example, an ontology might express that “x produces y” or “x consumes y.” These formalised relationships enable systems to make logical inferences, which enhances information retrieval. However, constructing and maintaining ontologies is more demanding than simpler vocabularies like taxonomies or thesauri.

Topic maps are a sophisticated form of ontology, providing detailed mappings of relationships between concepts. They are advantageous for information retrieval because of their precision and flexibility, and are particularly valuable for evolving and expanding knowledge domains. Despite their complexity, ontologies and topic maps play a crucial role in supporting more accurate and detailed information organisation, especially in digital environments.

Natural Language Approaches

This section discusses the other end of the subject access spectrum, focusing on natural language searching, where terms from the documents themselves are used rather than controlled vocabularies or classification schemes. Two factors make natural language approaches appealing:

1. Economic factors – Indexing is quicker and cheaper because it doesn’t involve translating terms into a controlled vocabulary.
2. Language factors – Natural language might better reflect the terms searchers naturally use.

However, relying solely on natural language has its downsides, particularly terminological scatter, where similar terms (synonyms or quasi-synonyms) are spread across multiple access points. Controlled vocabularies help consolidate similar concepts under one access point, ensuring that users don’t need to search for various synonyms. This control is similar to how authority control ensures users don’t have to search for all variants of a personal name when looking for a particular author.

There is sometimes confusion between two natural language approaches: natural language indexing and keyword searching. Keyword searching enables users to search databases using natural language terms, but it doesn’t mean the database uses natural language indexing. An item might be unindexed, relying entirely on the search and retrieval capabilities of the system, similar to how word-processed documents on a computer are searched. Keyword searching is common in most online databases and library catalogues (OPACs), and while convenient and powerful, it has its limitations.

Natural language indexing, on the other hand, involves selecting specific terms to represent the key concepts of an item, whether done by a human or a machine. These terms are still selected to reflect the topics within the resource.

Read

‘Natural language approaches’, pp. 154-163, in Hider, P., & Harvey, R. (2008). Organising knowledge in a global society: Principles and practice in libraries and information centres. Centre for Information Studies, Charles Sturt University. This book is available as an ebook through the CSU library.

Natural Language Approaches

The “Natural Language Approaches” section highlights how natural language searching differs from using controlled vocabularies. In this method, the vocabulary, or terms used for searching, are derived directly from the documents themselves. This allows access to the author’s original terminology, which is often more familiar to users and more current than terms in controlled vocabularies. Natural language searching complements, rather than replaces, controlled vocabulary searches.

Several mechanisms support natural language searching, including:

1. Keyword searching: This involves searching terms found in document titles, subtitles, and other metadata fields.
2. Records enhancement: Terms are also sourced from catalogue records, such as abstracts and content lists.
3. Automatic indexing: This uses either specified parts of the document, such as titles, or the entire document to extract searchable terms.

Keyword searching has become standard in many library information retrieval systems. These systems typically search through fields like title, author, and subject, but improvements in searching facilities, such as allowing searches across more fields or enabling full-text searches, are ongoing. Enhancing records by adding more searchable terms, like those from content pages or abstracts, improves subject access and retrieval success.

Despite its advantages, natural language searching can lead to challenges such as terminological scatter, where different synonyms or related terms might spread out results, making searching less precise. While controlled vocabularies help reduce this issue by standardising terms, natural language processing tools are still useful for capturing a wider range of search terms that users might naturally employ.

Natural language approaches are further supported by Boolean operations, truncation options, automatic synonym switching, and spell-checking, all of which help to refine and improve search precision.

Social Tagging and Folksonomies

Social tagging, or folksonomies, involves indexing resources using terms suggested by authors, contributors, or users, rather than relying solely on a controlled vocabulary. This allows for multiple people to tag a resource, potentially making it easier for searchers to find what they’re looking for. While this approach can be more cost-effective and reflective of the language searchers use, it also has limitations. Social tagging can lead to inaccurate or irrelevant tags, the use of synonyms or homonyms, and inconsistent vocabulary. While it may increase recall, it can reduce the precision of search results. Websites like Flickr and YouTube use this method, allowing users to assign searchable keywords. However, the vocabulary of the tagging community may not always align with that of the searchers, which can limit the effectiveness of the system.

Watch

Teresa Pelkie defines folksonomies and tagging and explains how they can be used to find and add meaning to web content.

Folksonomies and Tagging User: n/a – Added: 26/02/09 YouTube URL: http://www.youtube.com/watch?v=e8zajIMPVQE

Understanding Folksonomies in Web 2.0

Summary

In this introduction to folksonomies, Teresa Pelon discusses the concept, its significance, and applications in the digital age. The term “folksonomy,” combining “folk” and “taxonomy,” was coined by Thomas Vanderwal around 2004. A folksonomy refers to the collaborative method of creating tags to annotate and organise content. Unlike a taxonomy, which is a structured classification system governed by rules and authorities, a folksonomy allows users to freely choose and create keywords or tags without restrictions.

Folksonomies are essential for navigating the vast amount of information available on the internet. Traditional methods of information retrieval have limitations, particularly as content now exists in various media formats. Folksonomies enhance the search process by providing meaning or metadata to content, facilitating communication about it, and aiding in the retrieval of information. Users can tag content with labels that help define and locate it later, making the tags publicly accessible and open for collaboration.

Pelon highlights several platforms that utilise folksonomies, such as Flickr, which is recognised as one of the first to implement this system. Other notable examples include Delicious, YouTube, LibraryThing, and Amazon. As the course progresses, students will explore how these applications leverage folksonomies for better content categorisation and discovery.

The video also explains the concept of tags in greater detail. A tag is defined as a label that associates meaning with various types of content, including photos, videos, books, and web pages. These user-defined tags are informal and publicly visible. For instance, an animated photo of a dog could be tagged with phrases like “white dog,” “bouncing ball,” or “dog playing ball.” Such tags help convey meaning and facilitate the search for related information.

Additionally, Pelon introduces the idea of a tag cloud, a visualisation of popular or frequently used tags. This tool showcases the most significant tags associated with a particular topic, allowing users to quickly identify key areas of interest. For example, in a blog about Web 2.0, the tag cloud could highlight “Web 2.0” as the most prominent tag, with others varying in size based on usage frequency.

The session concludes with a promise to explore the practical applications of folksonomies, starting with Flickr, to illustrate how tags can be effectively used in real-world scenarios. This foundational understanding of folksonomies sets the stage for exploring their implications in the broader context of digital information sharing and organisation.

Read

‘End-users’ (pp. 85-87), in Hider, P. (2018). Information resource description: Creating and managing metadata (2nd ed.). Facet Publishing.

End-users

Historically, end-users seldom contributed metadata to information systems, with their input generally limited to reviews of books, films, and music in print media. However, the digital age has made it easier for end-users to directly add metadata in the form of comments, reviews, and ratings, as seen on platforms like Amazon and YouTube. Sites such as Flickr also allow users to add their own tags to resources, which is particularly valuable for audiovisual content where textual descriptions may be lacking.

In Web 2.0, communities of end-users have created collections of links through social bookmarking platforms like Diigo and Pinboard. In some cases, users can catalogue their personal book collections on platforms like LibraryThing and Goodreads, using their own tags alongside standard cataloguing data. This form of indexing, called “social tagging,” has revolutionised the organisation of information by allowing users to tag resources, creating a more democratic—though potentially anarchic—system of indexing.

Despite the potential benefits of social tagging, including scalability and the low cost of implementation, it has its drawbacks. User-generated tags may lack precision, and participation rates for tagging in systems like library catalogues have been low. Issues also arise from the uncontrolled nature of tags, with inconsistent vocabulary and varying levels of effort from users. While social tagging may increase recall, it can reduce precision by generating a large number of less relevant search results.

Nonetheless, user tagging can complement professional indexing, offering alternative ways to access information resources. It may also help improve controlled vocabularies and allow for more diverse perspectives in the way resources are categorised. With support tools like tag suggestion and disambiguation, social tagging can be made more effective, combining the strengths of both professional and user-driven approaches.

5.3 Subject access in school libraries

References

When searching by subject in school library catalogues in Australia or New Zealand, you are likely using SCIS Subject Headings, a controlled language system. These searches should also provide related references to help locate the correct subject headings. The subject headings and related terms are managed in the library’s subject authority file, which ensures more effective subject access through a controlled vocabulary.

Unfortunately, some school library catalogues only offer subject headings taken directly from catalogue records, without the additional related terms found in subject authority records, which would enhance resource discovery. Related terms must be specifically added to the subject authority file for them to be searchable online.

While learning about assigning and creating subject headings, it’s important to recognise that subject authority control, not just subject heading assignment, is crucial for effective searches. SCIS references, such as ‘Use’, ‘Broader Terms’, ‘Narrower Terms’, and ‘Related Terms’, help guide users from non-allowed terms to preferred subject headings, improving the search experience. Without these references, finding the correct terms and resources would be much harder.

References in library catalogues are essential to guide users to the correct subject headings, yet they are often underappreciated. Some catalogues contain few or no references, which undermines the effort of assigning subject headings, reduces the catalogue’s usefulness, and frustrates users, making them less likely to use it.

For instance, if users search for the subject ‘Magnets’ but the catalogue only uses ‘Magnetism’, without references directing them to the correct term, they may fail to find relevant resources. However, with proper references, such as:

• Magnets
  Use: Magnetism

or

• Physics
  Narrower Term: Magnetism

users are more likely to successfully locate the resources they need.