POPSI: A Faceted Approach to Subject Indexing

Subject indexing sounds simple until you try to do it well. How do you take a document about “the chemical treatment of tuberculosis of the lungs” and turn it into index entries that a reader can find from any angle they choose? For decades, librarians relied on Chain Indexing, but it had a stubborn flaw: it leaned entirely on a classification scheme and suffered from the problem of disappearing links in the chain. POPSI was designed to fix exactly that. It is a homegrown indexing system that broke free from classification numbers while still drawing on the deep logic of Indian classification theory.

What POPSI actually is 🔗

POPSI stands for Postulate-Based Permuted Subject Indexing. It is a pre-coordinate indexing system, which means the indexer combines the component terms of a subject in a fixed, meaningful order before the entries are filed, rather than leaving that combining to the searcher. The system grew out of work at the Documentation Research and Training Centre (DRTC) in Bangalore, where information scientists set out to overcome the weaknesses of Chain Indexing.

The two big problems with Chain Indexing were its dependence on a scheme of classification and the phenomenon of the “disappearing chain,” where certain links could not generate usable index entries. POPSI addressed both, minimising the disappearing-chain problem and providing a systematic structure for generating subject headings that does not need to be tagged to any particular classification scheme.

Concept and development 🔗

POPSI was developed by Ganesh Bhattacharyya at the DRTC, with its formal introduction traced to a 1969 DRTC seminar and continued refinement through the following decade. Bhattacharyya built directly on S. R. Ranganathan’s General Theory of Library Classification, particularly the postulational and facet-analysis approach that underpins Colon Classification.

The key word here is “postulate.” In this context, a postulate is a basic assumption or guiding principle that tells the indexer how subjects are structured and how their parts should be arranged. Bhattacharyya set out the fundamentals of POPSI on the basis of an experiment, formulating a set of general postulates about the elementary and syntactic structures of a compound subject. Where Ranganathan’s Colon Classification produced a class number, POPSI uses the same underlying logic to produce a verbal subject heading instead. That single shift, from notation to natural-language terms, is what makes POPSI so flexible.

The deep structure of subject indexing language 🔗

POPSI is grounded in what Bhattacharyya called the General Theory of Subject Indexing Languages (SIL). To understand POPSI, you first have to understand what a “subject indexing language” is and why it has a deep structure.

A subject indexing language is a controlled, rule-governed system for naming and arranging subjects so that documents on the same topic are brought together and can be retrieved consistently. Bhattacharyya argued that beneath the surface words of any subject statement lies a logical deep structure: a predictable arrangement of conceptual building blocks and the relationships between them. POPSI was derived through a logical interpretation of this deep structure of the subject indexing language, which is what gives the system its theoretical strength.

This deep structure has three layers worth knowing. The semantic structure deals with the meaning of terms and the vocabulary control needed to handle synonyms. The elementary structure identifies the basic conceptual categories that any subject can be broken into. The syntactic structure governs the order in which those categories are arranged to form a meaningful subject proposition. POPSI’s procedure is, in effect, a practical method for surfacing this hidden structure and writing it down as index entries.

The elementary categories 🔗

At the heart of the elementary structure sit the Elementary Categories (ECs), the fundamental kinds of idea that combine to form any subject statement. Bhattacharyya postulated four core categories:

Discipline (D): a conventional field of study or any aggregate of such fields, for example Physics, Medicine, or Library Science. Entity (E): a manifestation with a perceptual correlate or a purely conceptual existence, distinct from the properties and actions associated with it, for example Lungs, Plant, or Energy. Action (A): a manifestation denoting the concept of “doing” or a process, for example Treatment, Migration, or Education. Property (P): a manifestation denoting an attribute, whether qualitative or quantitative, for example Power, Capacity, or Efficiency.

Alongside these categories, POPSI uses modifiers to qualify a term without changing its essential character, and it prescribes apparatus words such as prepositions, conjunctions, and participles where they are needed to keep a heading readable. These categories and modifiers are the raw material from which every POPSI subject heading is assembled.

How POPSI works: the indexing mechanism 🔗

The real character of POPSI shows in its working procedure. Index entries are generated through a fixed series of steps, so two trained indexers working on the same document should arrive at very similar results. The standard steps are content analysis, formalisation, modulation, standardisation, preparation of the entry for organising classification, a decision about terms of approach, preparation of entries for associative classification, and finally alphabetisation.

Step one: content analysis 🔗

Content analysis involves identifying the different component ideas in the document and assigning each to its elementary category and any modifiers. Take the classic teaching example, “Chemical treatment of tuberculosis of lungs.” Analysis sorts the ideas as Discipline = Medicine, Entity = Lungs, Property = Tuberculosis, and Action = Chemical treatment.

Step two: formalisation 🔗

Formalisation arranges these components into a formal sequence according to the rules of syntax, with each term tagged by its status. The basic chain becomes: Medicine (D), Lungs (E), Tuberculosis (P of E), Chemical treatment (A on P). Notice how the syntax records not just the terms but their role and relationship to one another, such as a property “of” an entity, or an action “on” a property.

Step three: modulation 🔗

Modulation enriches each component by interpolating and extrapolating its successive superordinates, the broader terms in its hierarchy, so the context is clear. The chain expands to something like: Medicine (D), Man. Respiratory System. Lungs (E), Disease. Tuberculosis (P of E), Chemical treatment (A on P). This is where POPSI captures the hierarchical context that Chain Indexing often lost.

Step four: standardisation 🔗

Standardisation handles semantics. It decides the standard term for any manifestation that has synonyms and prepares the basis for cross-reference entries, drawing on a vocabulary-control device called the Classaurus. In the example, “Chemical treatment” is standardised to “Chemotherapy,” with the synonym recorded so a searcher using either word still finds the document.

Role operators and semantic relationships 🔗

POPSI uses a set of role operators and punctuation marks to signal exactly what each term is doing in the string. In the basic version, a comma precedes an entity segment, a semicolon precedes a property segment, a colon precedes a process segment, a hyphen marks a qualifying sub-segment, and a “greater-than” sign marks a narrower term. These markers preserve the semantic relationships between concepts, so the meaning of a heading does not collapse when the terms are rearranged.

The permutation that gives POPSI its name 🔗

The word “permuted” is the final piece. Once the formalised, modulated string exists, POPSI generates multiple entries by rotating the sought terms so the subject can be approached from each significant access point. This produces two kinds of output: entries for organising classification, which keep the full context, and entries for associative classification, which create the alphabetical access points a searcher actually uses. A reader looking up Lungs, Tuberculosis, or Chemotherapy will each land on the same document with its context intact.

Vocabulary control through the Classaurus 🔗

A subject indexing language is only as good as its control of vocabulary. POPSI’s answer is the Classaurus, a vocabulary-control device that combines features of a classification scheme and a thesaurus. It stores terms in their hierarchical relationships, like a classification, while also recording synonyms and related terms, like a thesaurus. The standardisation and modulation steps both draw on it, which is how POPSI keeps its headings consistent across a whole collection without being tied to any external classification number.

Why POPSI mattered 🔗

POPSI’s lasting contribution was to show that a rigorous, theory-driven indexing language could be built on Ranganathan’s postulates without depending on a classification scheme at all. It solved the disappearing-chain problem and offered a systematic way to handle multi-dimensional, complex subjects. Although systems like PRECIS, developed by Derek Austin for the British National Bibliography, became more widely adopted internationally, POPSI remains a landmark in the theory of subject analysis and a regular feature of Library and Information Science curricula. It demonstrates how careful attention to the deep structure of language can be turned into a practical tool for finding information.

What do you think? If POPSI was theoretically so strong, why do you think it never achieved the wide practical adoption that systems like PRECIS did? And in an age of full-text search and AI-driven retrieval, does the postulate-based discipline of POPSI still have something to teach us about organising knowledge?

References

1. https://www.librarianshipstudies.com/2017/05/popsi.html
2. https://www.academia.edu/78618589/Two_Decades_of_POPSI_1969_1988_A_Literature_Review
3. https://www.britannica.com/biography/Shiyali-Ramamrita-Ranganathan
4. https://www.srels.org/index.php/sjim/article/view/50472
5. https://egyankosh.ac.in/bitstream/123456789/35771/5/Unit-11.pdf
6. https://egyankosh.ac.in/bitstream/123456789/33116/1/Unit-17.pdf
7. https://slideshare.net/PAQUIAAIZEL/indexing-popsi