A Subject Has More Than One Side

A report on repairing timber bridges in coastal Japan during the 1980s has several perfectly reasonable subjects. A carpenter may care about timber. A transport engineer may care about bridges. A researcher may want the place or period.

The report does not change when a different person approaches it. What changes is the distinction they need.

This is why S. R. Ranganathan matters to ontology work. His approach to classification made room for a subject to be described through several dimensions, then combined according to a defined method. The result was more deliberate than attaching a handful of keywords and hoping for the best.

Start with the person using the library

Ranganathan’s The Five Laws of Library Science appeared in 1931. Its concerns include use, the relationship between readers and books, the reader’s time, and the library’s capacity to grow. Those concerns make classification part of a service rather than an achievement judged only by the elegance of its categories. Ranganathan, The Five Laws of Library Science, 1931

That starting point can be surprisingly easy to lose in a modern ontology workshop. People may spend a morning debating the exact position of “inspection” in a hierarchy without asking what the maintenance planner needs to retrieve before arranging tomorrow’s work.

A reader-centred question makes the disagreement more useful. Does the planner need to distinguish routine inspections from inspections triggered by damage? Must the system show the equipment involved, the required qualifications, or the date by which the work must happen? The relevant distinctions follow from an actual task.

No historical librarian could have specified every modern maintenance workflow. The continuity is in judging the organization of knowledge by what it enables someone to do.

A subject assembled from parts

Ranganathan’s Colon Classification was published in 1933. A contemporary review describes schedules of divisions grouped by function or characteristic, with a subject’s class mark formed by combining the appropriate divisions in a specified order. The colon was among the devices used to connect them. Contemporary review of Colon Classification in Nature

The plain idea is that a complex subject need not be anticipated as one indivisible category. Its components can be identified and combined.

In the developed theory, Ranganathan’s familiar categories are Personality, Matter, Energy, Space, and Time. The vocabulary can sound stranger than the task. “Energy,” for example, concerns action in this classificatory setting; it does not mean that every subject must be described in joules. The theory evolved, and the later five-category formulation should not be read backward as though every edition had already settled it. ISKO’s research account of facet analysis

Take the bridge report. “Timber” identifies material. “Repair” identifies an activity. Japan identifies place, and the 1980s identify time. The bridge is the principal thing being discussed. These dimensions can support different searches without requiring five copies of the report.

A facet is a grouping based on one characteristic. Keeping that characteristic stable is the important part. A list mixing steel, Scotland, urgent, and inspection does not become a useful facet because it appears beside a set of checkboxes.

The method developed through people

Ranganathan worked within a profession already debating classification. W. C. Berwick Sayers was an important teacher in his library training. Later, the British Classification Research Group extended and reconsidered faceted methods. Jack Mills and colleagues developed the second edition of the Bliss Bibliographic Classification within that broader tradition.

Vanda Broughton’s account of BC2 describes its debt to Ranganathan and the group’s modifications, including a richer range of categories. The method did not require every domain to be forced into an unchanged list of five. Broughton, account of the Bliss Bibliographic Classification, second edition

That history is useful because it shows a technical practice being tested and revised. A classification method can remain recognizably continuous while its categories change.

Modern ontology teams encounter the same need. A library might describe the form of a publication. A materials database might need manufacturing process and grain structure. Borrowing a useful method does not mean importing every category from the previous application.

What the search form cannot tell you

Many people first meet facets in a shopping or library search interface. They select a material, date, location, or format to narrow the results. The interface makes the method feel obvious.

Behind it, the choices still need definitions. Does the date refer to publication, the period described, the date of an observation, or the latest revision? A report published in 2024 may describe work performed in 1985. One date filter cannot answer both questions unless the distinction is preserved.

A professional ontology can go further than a faceted search interface. It can say that an inspection concerns a specific bridge, used a particular method, and produced a report. Facets help describe and retrieve the report. Explicit relationships help explain the events and objects behind it.

Neither arrangement should be asked to do the other’s work invisibly.

Suppose a user selects timber, repair, and Japan. The system returns a report that mentions timber in a bibliography, describes repairs to steel structures, and lists a Japanese publisher. Every word is present. The requested subject is not.

The model needs to distinguish what the report is about from words that merely occur in it. It also needs enough connection among the selected dimensions to avoid assembling a false match.

That leaves a question which will become central to early information retrieval. How can the people designing an index tell whether it actually finds what a reader needs? A carefully constructed vocabulary still has to face the search.