Search vocabulary: separate aliases from different concepts

Which decision does a search vocabulary preserve?

A search vocabulary connects supplied expressions to defined concepts and makes unresolved meanings explicit. This fictional office-supply exercise produces an owned dictionary, subtype boundaries and clarification wording. It does not configure a search engine, return product results or prove that customers use these expressions.

A concept is the meaning being referred to. A canonical label is the chosen main name for that concept. An allowed alias is another expression the owner explicitly attaches to that same concept in V1. This strict identity rule is an exercise contract, not a universal definition of every synonym system or of natural language.

Preserve the literal supplied tokens: workdesk, writingdesk, sitstanddesk, diningtable, worklight and table. Their joined spelling is part of the input. Do not silently split words, normalize punctuation, infer translations or add common-sounding alternatives. An explanation in Azerbaijani does not establish an additional verified Azerbaijani alias list.

The catalog structure plan chooses a grouping axis and places products. This note makes a different decision: whether expressions refer to the same defined meaning. Sharing a category does not settle that question. No category tree, stock record, product identity or URL is changed.

How should the four owner-defined concepts remain distinct?

C1 is a writing work surface, canonically writingdesk, with workdesk as its allowed alternative. Both expressions refer to C1 under the supplied owner rule. The concept reference is neither a physical product ID nor a set of items retrieved from a store.

C2 is a height-adjustable writing work surface, with canonical label height-adjustablewritingdesk and alias sitstanddesk. It is expressly a strict subtype of C1. C3 is a meal surface, canonically diningtable. C4 is a light for the work surface, canonically desklamp. None of those definitions was discovered from a shopper or inferred by an AI model.

A strict subtype is not identical to its broader concept. Every C2 instance belongs to the supplied writing-work-surface type, but the exercise does not say every C1 instance is height adjustable. Keep C2 separate and do not turn sitstanddesk into a name for all C1 instances.

Likewise, C1 and C3 describe different supplied uses. The broad token table cannot be used to erase that difference. The search intent guide groups needs for appropriate page content; similar intent alone is not proof of lexical identity. A visitor looking for furniture does not make every furniture term interchangeable.

Which expressions need a question before a concept can be chosen?

The owner explicitly assigns two possible senses to worklight: C4 and an unspecified workshop light. The second concept has not been defined. Preserve that omission rather than creating a convenient C5 or assuming that all lights for work are desk lamps.

Save this clarification: “Do you mean a light for the work surface or a workshop light? The second concept has not been supplied.” This is prepared wording for a future meaning decision, not a message sent to a real customer. Until the owner resolves it, worklight is not an automatic C4 alias.

The token table is separately ambiguous between C1 and C3 without context. Ask: “Do you mean a writing work surface or a meal surface?” Keep HOLD instead of selecting the most convenient concept. HOLD means the semantic decision is unresolved; it does not mean a live engine returned no results.

The long-tail keyword guide addresses specific questions and content opportunities. Here no keyword frequency or customer demand is known. Do not broaden the dictionary simply to capture more queries. A reciprocal rule that conflates the narrower C2 with all C1 would change the supplied meanings rather than make them easier to find.

What do the handworked references actually show?

Look up each expression exactly as supplied in V1, then inspect only the explicit canonical-label or alias membership. workdesk and writingdesk refer to C1. sitstanddesk refers to C2, not an expanded list containing all C1. diningtable refers to C3. No hierarchy expansion or software matching rule is applied.

worklight and table have no selected concept reference because their meanings still need clarification. Do not borrow the result of a neighbouring term or choose a concept because its name shares a word. The two questions are part of the dictionary handoff, not defects to hide with a guessed answer.

Decide three claims. “The same category or intent proves the same concept” is rejected. “C2 is narrower, therefore all C1 is C2” is rejected. “Two expressions refer to C1, so a live search must return identical products” is unsupported. Attach the relevant definition or limitation to each decision.

Concept references are not search-result sets. This manual exercise does not inspect ranking, typo tolerance, plurals, recall, relevance or actual product matches. The negative keyword workflow has a different operational purpose: excluding unsuitable advertising queries while preserving useful ones. No advertising match syntax or exclusion list is copied into this vocabulary.

How far do the two sources support the owner rule?

The informative W3C SKOS Primer distinguishes preferred and alternative lexical labels on concepts from broader and narrower semantic relations. However, alternative labels can also represent near-synonyms, abbreviations and upward posting in some schemes. It does not universally require every alternative label to be strictly identical in meaning.

Our narrower same-concept rule comes from the fictional V1 owner contract. C2 is explicitly represented, so its identity remains separate here. We are not creating RDF, asserting SKOS conformance or using the Primer as linguistic proof that the invented workdesk assignment holds in every setting.

Algolia's documentation distinguishes regular interchangeable and one-way behaviours in its own system and warns about words with several meanings. It also states that synonym terms are not guaranteed to return identical result sets. These limitations are substantive: no ready-to-run synonym rule, current interface procedure or engine guarantee follows from this paper dictionary.

Neither source verifies the fictional owner's terminology, universal English–Azerbaijani equivalence or a real audience's vocabulary. Search quality and demand remain unmeasured. The URL structure guide concerns naming addresses; a canonical concept label here is not a canonical URL instruction. Keep the terminology decision separate from both page naming and search implementation.

How can the dictionary and unresolved questions be reopened?

On Windows, open Start, search for Notepad and launch the application. Create a new blank document headed Search vocabulary V1. Copy the four owner-defined concepts, their definitions, canonical labels and allowed alias memberships. Keep C1 through C4 beside the wording so another reader can distinguish labels from concepts.

Add the strict subtype relation, the separate C1 and C3 meanings and both ambiguity questions. Include the six manual references and the three claim decisions with their evidence. State that the artifact is a handworked concept plan, not an actual engine or customer test. Do not give the unspecified workshop light a newly invented concept ID.

Press Ctrl+S, choose Documents, enter search-vocabulary-V1.txt and select Save. Close Notepad. Open Documents in File Explorer and double-click search-vocabulary-V1.txt to reopen that named file. Confirm workdesk still refers to C1, sitstanddesk to C2 and both ambiguous expressions still retain HOLD and their questions.

Check that the file contains four defined concept IDs and no C5. Make sure sitstanddesk has not migrated into the broad C1 alias list. Read the Algolia result-set limitation again before using the note as a future technical handoff. Local readback confirms the saved text only; it is not an observed usability study or a configured search test.

A proposed fifteen-minute cognitive allowance gives four minutes to definitions, four to relation boundaries, four to questions and manual references and three to saving and reopening. It is a planning budget, not measured participant performance. In earlier public writing I discussed perfectionism delaying useful work; the application is finishing a small reproducible meaning note before expanding the task. The artificial intelligence guide provides broader context, but no model has been used to decide missing owner meanings here.