Types of Indexing Languages: Classification, Subject Headings, and Thesauri

Walk into any library and you will notice something quietly remarkable: thousands of books, each on a different topic, are arranged in a way that lets you find exactly what you need. Behind this order lies a powerful concept called an indexing language. An indexing language is an artificial language designed specifically to represent the subject content of documents so they can be stored and retrieved efficiently. Without it, the vast collections in libraries, research databases, and digital archives would be little more than unsearchable heaps of information. In this post, we will explore the major types of indexing languages, how each one works, and why some clever combinations of these systems were created to overcome their individual limitations.

What is an indexing language? 🔗

An indexing language is a controlled set of terms or codes used to describe what a document is about. The key word here is controlled. Instead of letting indexers use whatever words come to mind, an indexing language fixes the vocabulary, decides which terms are preferred, and shows how different terms relate to one another. This solves a basic problem: the same idea can be expressed in many ways. A document about “cars” might also be described using “automobiles” or “motor vehicles.” If indexers used these terms randomly, a searcher might miss half the relevant material.

Indexing languages broadly fall into two families based on the nature of their vocabulary. The vocabulary of an indexing language is either verbal or coded. Subject heading lists and thesauri use verbal vocabulary, meaning the terms are ordinary words arranged alphabetically. Classification schemes use coded vocabulary, where each subject is represented by a notation or symbol. For example, the subject “Indian History” is rendered as the notation V44 in Colon Classification, but appears as the verbal heading “India – History” in a subject heading list. Both point to the same subject; they simply use different languages to express it.

This distinction between coded and verbal vocabulary gives us a natural way to understand the main types of indexing languages, along with a fascinating set of hybrid systems that blend both approaches.

Classification schemes: organising subjects through codes 🔗

At the foundation of subject organisation sits the classification scheme, which arranges knowledge into a systematic order using notation. In a classification scheme, each subject is allocated a piece of notation, and that notation is used to represent the subject and to fix its place in a logical sequence. Related subjects are grouped together, and broad topics are progressively broken into narrower ones. This is what allows physical books on the same subject to sit side by side on a shelf.

The major classification schemes 🔗

Three classification systems dominate library practice, and understanding their differences reveals how the very idea of classification evolved.

The Dewey Decimal Classification (DDC), first published by Melvil Dewey in 1876, is among the most widely used schemes in the world. It divides all knowledge into ten main classes, each represented by numbers and further subdivided using decimals. For instance, the 500s cover natural sciences and mathematics, which then break down into physics, chemistry, astronomy, and so on. DDC is largely an enumerative scheme, meaning it tries to list ready-made numbers for subjects in advance, which makes it simple to apply.

The Universal Decimal Classification (UDC) was developed by Paul Otlet and Henri La Fontaine, beginning in 1895, using DDC as its base. UDC combines both enumerative and faceted characteristics, which is why it is often called an “almost-faceted” scheme. It uses punctuation marks such as the colon, plus sign, and parentheses to link concepts relationally, allowing indexers to build highly specific numbers that connect a subject with its form, place, or time. This expressiveness makes UDC popular in special and scientific libraries.

The Colon Classification (CC), devised by Dr. S. R. Ranganathan in 1933, took a radically different path. Ranganathan argued that rigid hierarchical schemes were too limiting because many documents deal with more than one subject. His solution was facet analysis: instead of listing every possible subject, the classifier identifies the fundamental aspects of a topic and combines them. CC is built on five fundamental categories, often remembered as PMEST – Personality, Matter, Energy, Space, and Time. These facets are joined using connecting symbols, with the colon giving the scheme its name. Colon Classification is widely regarded as the first comprehensive analytico-synthetic scheme of classification.

Enumerative versus faceted thinking 🔗

The contrast between DDC and CC illustrates a deep divide in classification theory. Enumerative schemes attempt to provide a complete, pre-built list of subject numbers. They are easy to use but struggle with new or complex topics that were never anticipated. Faceted schemes instead provide building blocks that can be combined freely, making them far more hospitable to interdisciplinary and emerging subjects. Modern systems like DDC, the Library of Congress Classification, and UDC all use facets at various points within their schedules, showing that the two approaches are not entirely separate but exist on a spectrum.

Subject heading lists: organising topics through words 🔗

While classification schemes use codes, subject heading lists use words. A subject heading list is a controlled, standardised set of verbal terms used to describe what documents are about. These headings are arranged alphabetically, and there are strict rules for how new terms are added and how they link to existing ones. This consistency ensures that materials on the same topic are indexed under the same heading, so a searcher reliably finds everything relevant in one place.

Subject heading lists power the alphabetical subject approach in catalogues. When you search a library catalogue by subject and type “Physics,” you are using a subject heading. The system has decided that “Physics” is the preferred term and will guide you there even if you started with a related or alternative word.

Well-known subject heading systems 🔗

Several subject heading lists have become standards in the field. The Library of Congress Subject Headings (LCSH) is one of the most comprehensive and is used in libraries worldwide. Sears List of Subject Headings is a simpler list designed for small and medium-sized libraries, including school libraries. For medicine, the Medical Subject Headings (MeSH), created by the United States National Library of Medicine, organises biomedical literature with great precision.

A defining feature of these lists is the use of cross-references. They guide a searcher from a term that is not used to one that is, and they connect broader, narrower, and related headings. This network of references is what turns a flat alphabetical list into a navigable structure.

Thesauri: defining relationships between terms 🔗

A thesaurus in information science is not the synonym dictionary you might use while writing an essay. It is a controlled vocabulary in which each term represents a single concept, and the relationships between terms are explicitly mapped out. A thesaurus adds a layer of meaning, or semantics, to indexing by spelling out exactly how concepts connect.

The relationships within a thesaurus typically fall into three categories. Equivalence relationships handle synonyms, telling the indexer which of several terms is preferred. Hierarchical relationships show broader terms (BT) and narrower terms (NT), placing a concept within a chain of generality. Associative relationships link related terms (RT) that are conceptually connected but not hierarchically. These relationships help overcome the problem of synonymy, where multiple words describe the same concept, and improve both the recall and precision of searches.

Well-known thesauri include the Art and Architecture Thesaurus and the ERIC Thesaurus for education. Thesauri are especially valuable in post-coordinate indexing systems and digital databases, where searchers combine terms at the time of searching rather than relying on a pre-built order.

Thesaurofacet: marrying classification and thesaurus 🔗

For a long time, classification schemes and thesauri were treated as separate tools. Classification gave a logical, hierarchical structure; thesauri gave rich alphabetical access and relationship mapping. The Thesaurofacet was a pioneering attempt to combine the best of both in a single tool.

Developed by Jean Aitchison and colleagues for the English Electric Company in 1969, the Thesaurofacet covers engineering and related scientific, technical, and management subjects. It is commonly described as the first faceted thesaurus. Its novel feature is the integration of faceted classification schedules with a thesaurus. Each term appears both in the classified schedules and in the alphabetical thesaurus. In the schedules, the term sits in its most appropriate facet and hierarchy. The thesaurus then supplements this by indicating alternative hierarchies and other relationships that cut across the classified arrangement, while also controlling synonyms and word forms.

In effect, the thesaurus also serves as the alphabetical index to the class numbers. This dual structure makes the Thesaurofacet genuinely multipurpose: it can be used for shelf arrangement, classified card catalogues, coordinate indexing, and computerised retrieval alike.

Classaurus: a unique Indian contribution 🔗

Around the same time that the Thesaurofacet emerged in Britain, a parallel innovation was taking shape in India. The Classaurus is a vocabulary control tool developed by G. Bhattacharyya at the Documentation Research and Training Centre (DRTC) in Bangalore. It was designed as the vocabulary control device for POPSI – Postulate-based Permuted Subject Indexing – a pre-coordinate indexing system also developed by Bhattacharyya. The term “classaurus,” denoting an indexing tool that combines classification and thesaurus, was introduced by Bhattacharyya in 1982.

How the Classaurus works 🔗

The Classaurus is best described as a category-based, or faceted, systematic scheme of hierarchical classification operating in the verbal plane, while incorporating the essential features of a conventional information retrieval thesaurus. In other words, it organises terms hierarchically the way a classification scheme would, but the terms themselves are words rather than codes, and it controls synonyms and quasi-synonyms the way a thesaurus does.

Structurally, the Classaurus rests on the postulates of Ranganathan’s general theory of classification. It has separate hierarchical schedules for the elementary categories used in POPSI – Discipline, Entity, Property, and Action – along with their species, parts, and special modifiers, as well as common modifiers such as form, time, place, and environment. Each term in these schedules is enriched with its synonyms, quasi-synonyms, antonyms, and a scope note where required.

The systematic part is supplemented by an alphabetical index that gives the address of each term in the schedules. This two-part design – a systematic hierarchical portion and an alphabetical index – is a feature it shares in spirit with the Thesaurofacet.

What makes the Classaurus distinctive 🔗

One interesting design choice sets the Classaurus apart from a conventional thesaurus: related terms (RTs) are deliberately not shown in it. Bhattacharyya argued that relationships between terms should be dictated by the actual thought-content of each document being indexed, not pre-decided by the designer of the tool. Any term might relate to any other term depending on the document at hand, so fixing these relationships in advance would be artificial.

Because of its category-based structure, the arrangement of terms within each array does not require any complex principle-based ordering. Even a simple alphabetical arrangement within an array serves the purpose, which makes the Classaurus particularly suitable for computerisation. Computerised versions of POPSI and the Classaurus have indeed been developed, with one notable later version known as the Deep Structure Indexing System (DSIS).

How these tools fit together 🔗

It helps to see these indexing languages not as rivals but as answers to different needs. Classification schemes give a coded, browsable order ideal for arranging collections. Subject heading lists offer a familiar word-based approach for catalogue searching. Thesauri provide precise relationship mapping for database retrieval. And hybrid tools like the Thesaurofacet and Classaurus recognise that the boundary between classification and verbal vocabulary is artificial, blending coded structure with verbal flexibility to get the strengths of both.

The fact that two such hybrids emerged independently – one in Britain, one in India – within roughly a decade of each other tells us something important. As information grew more complex and computers entered the picture, indexers everywhere reached the same conclusion: the richest retrieval comes from tools that combine the logical depth of classification with the descriptive clarity of words.

What do you think? If you were designing an indexing system for a brand-new digital archive today, would you lean toward the structured logic of a classification scheme or the flexible relationships of a thesaurus – and why might a hybrid like the Classaurus be the wiser choice? And do you think modern search algorithms have made traditional indexing languages obsolete, or do they still quietly depend on these decades-old principles?

References

1. https://en.wikipedia.org/wiki/Colon_Classification
2. https://en.wikipedia.org/wiki/Dewey_Decimal_Classification
3. https://www.lisedunetwork.com/universal-decimal-classification-udc/
4. https://en.wikipedia.org/wiki/Library_of_Congress_Subject_Headings
5. https://en.wikipedia.org/wiki/Sears_Subject_Headings
6. https://en.wikipedia.org/wiki/Medical_Subject_Headings
7. https://en.wikipedia.org/wiki/Art_and_Architecture_Thesaurus
8. https://www.britannica.com/topic/Thesaurofacet
9. https://www.sciencedirect.com/science/article/pii/0306457385901244