Vocabulary | What Is a Concept Merge? When Two Entries Should Become One

A concept merge is a deliberate change to a knowledge scheme that combines two or more concept records because, after careful review, they describe the same intended idea closely enough that separate identities no longer serve the scheme. Imagine a learning library containing ‘Equivalent Fractions’ and ‘Fractions with Equal Values’. If both entries use the same mathematical definition, include the same cases and answer the same learner question, retaining two independent concept records may divide useful resources and confuse readers.

This is vocabulary as knowledge-management technology. Labels shape what students find, which examples tutors select, how editors connect resources and what a search engine treats as the subject of a document. A sound merge can reduce duplication and give readers one consistent route to information. A mistaken merge can be far worse: it may collapse a requirement into a recommendation, combine different Science processes, erase history or redirect learners to the wrong task. The key is establishing conceptual identity rather than rewarding superficial similarity of words.

This guide explains when a merge is warranted, how to compare scope notes and counterexamples, how to choose a surviving concept identifier, and how to preserve old labels, mappings, resources and historic links. It uses original Mathematics, English and school-library cases, a practical migration worksheet and positive and negative tests. The technical references are W3C SKOS, W3C provenance and publicly documented thesaurus maintenance. The cases are fictional editorial exercises, not records of changes to real curricula.

Editorial note: All school collections, editorial decisions and quoted concept examples are fictional. The references establish knowledge-organisation principles, not equivalence between actual external education systems.

Contents: when a merge is justified and how to preserve meaning

Open 49 chapters and reviewed examples

1. The initial problem: two entries for one question
2. What precisely changes in a merge?
3. The W3C concept model
4. A documented real-world maintenance pattern
5. Merge versus rename
6. Merge versus synonym ring
7. Merge versus entity resolution
8. Merge versus semantic crosswalk
9. What qualifies two entries as duplicates?
10. Compare the scope notes side by side
11. An exact same-purpose test
12. Use counterexamples, not just agreement
13. A related concept may look nearly identical
14. The danger of obligation words
15. A merge can be appropriate after accidental duplication
16. Choose a surviving concept carefully
17. The surviving ID must preserve meaning
18. Retired IDs should not be reassigned
19. Retain older labels as appropriate
20. When a choice page is better than redirecting
21. Moving resources requires content review
22. Do not merge article content indiscriminately
23. Mappings and hierarchies may need repair
24. Exact-match mappings need extra caution
25. Search results can improve after a genuine merge
26. False merges can reduce search precision
27. An unchanged URL can conceal a changed subject
28. Respect old source citations
29. The merge decision needs an owner
30. The role of provenance
31. The Nomenclature history example
32. A merge can be provisional before publication
33. What to do when the evidence conflicts
34. A merge can require a new parent concept
35. A merge can reveal a mistaken earlier split
36. A small Mathematics merge case
37. A Mathematics non-merge case
38. An English non-merge case
39. A Science non-merge case
40. A school-workflow merge case
41. A user-intent collision test
42. A positive and a negative search test
43. A versioned migration register
44. A rollback must protect valid changes
45. How to verify the published outcome
46. Seven merge mistakes
47. Twelve diagnostic questions
48. Frequently asked questions
49. References and the next action

1. The initial problem: two entries for one question

A fictional Mathematics collection has separate subject cards called Equivalent Fractions and Fractions of the Same Value. Both define equality of fractional values, offer the same worked examples and lead to overlapping exercises.

One student opens only the first card and misses material filed under the second. Another assumes the cards must teach different mathematical ideas because the library gives them separate identities.

An editor can investigate whether one concept record should remain with the other name preserved as a search alternative. The aim is to remove false conceptual separation without deleting distinct teaching resources.

Back to contents

2. What precisely changes in a merge?

Two or more concept identities are consolidated into a surviving identity or a suitably defined replacement concept. Their labels, examples, relationships and document assignments need review.

That is more than deleting a duplicate title from a webpage. Existing links may refer to different IDs, and a mapping in another library may depend on the old record.

A genuine merge is complete only when the resulting scope is clear and the affected navigation and references behave appropriately under the scheme.

Back to contents

3. The W3C concept model

The W3C SKOS Reference defines concepts, labels, broader and narrower relationships, documentation notes and mapping properties.

A concept merge can be represented through revised concept records and appropriate historical or mapping information, but SKOS does not prescribe a universal automatic merge command that resolves all existing data.

Editors need a migration policy based on their scheme’s identifiers, current scope and actual resource relationships. The standard supplies useful building blocks, not an infallible synonym detector.

Back to contents

4. A documented real-world maintenance pattern

The Nomenclature linked-data documentation describes how a thesaurus records changes to concepts, including deprecation and merge events.

In its implementation, a deprecated source concept can carry a replacement relationship and appropriate historic information, while labels from the source may become alternatives under a surviving concept.

This illustrates one documented implementation, not a requirement that every school or WordPress library adopt exactly the same graph structure.

Back to contents

5. Merge versus rename

An entry changes its preferred label from ‘Secondary One Maths’ to ‘Secondary 1 Mathematics’ without changing what it means. That is a label revision.

A merge concerns more than one concept record. If those records were genuinely duplicates, the editor may be able to consolidate them and retain familiar words as alternative labels.

Do not create or retire IDs when an ordinary spelling or display change would be enough. Identifiers and labels perform different jobs.

Back to contents

6. Merge versus synonym ring

A synonym ring groups expressions that can serve the same retrieval purpose. It may allow readers to find all resources without combining the underlying concept records.

A concept merge changes the governed scheme’s identities and classifications. Sometimes a ring is enough; sometimes duplicate conceptual records still need an editorial decision.

Use the Synonym Ring guide for search-term equivalence and its limits. One search route is not automatically one catalogue identity.

Back to contents

7. Merge versus entity resolution

Entity resolution asks whether records refer to the same underlying thing at the selected identity level. A concept merge is a policy or editorial change that consolidates entries after resolving their intended meanings.

Two documents about Equivalent Fractions can be distinct teaching resources even when both refer to the same subject concept. Resolving their shared subject does not justify merging their article identities.

The Entity Resolution guide helps distinguish identity matching from similarity and mere association.

Back to contents

8. Merge versus semantic crosswalk

A crosswalk can relate two concepts in different vocabularies as exact, close, broader, narrower or related. The two schemes may retain their independent identities.

A merge is a decision inside a governed set of records to consolidate what was previously stored separately. Cross-system correspondence does not automatically authorise one institution to delete or replace another’s concept.

Keep the boundary between local governance and external mapping. A related topic is not necessarily a merge candidate.

Back to contents

9. What qualifies two entries as duplicates?

Compare their definitions, scope notes, intended reader questions, included examples, exclusions and actual document assignments. Do they describe the same concept for the scheme’s purpose?

Identical titles do not settle the issue; two different topics can share a label. Different titles also do not prove difference; ordinary paraphrases can name the same subject.

The decision should be justified by meaning and use, not by an algorithm that counts shared letters.

Back to contents

10. Compare the scope notes side by side

One concept includes every type of fraction operation, while the other is restricted to recognising equal values. Their titles may sound similar but their boundaries differ.

A genuine merge would widen or narrow one of the scopes unless the editor first resolves the mismatch. That could be an invalid change to how historic resources are classified.

The Scope Note guide explains how positive, negative and boundary examples reveal those differences.

Back to contents

11. An exact same-purpose test

Write one question each concept entry claims to answer. If both are ‘How can I recognise that two fractions represent the same value?’, the records may be candidates.

Now ask whether one entry includes a distinct teaching objective, such as performing fraction addition, that the other excludes. If so, an apparent duplicate may reflect a broader/narrower relationship instead.

A merge should simplify a real reader’s route without deleting a meaningful choice.

Back to contents

12. Use counterexamples, not just agreement

An editor observes that both records include the example 1/2 = 2/4. One shared example is weak evidence of identical scope.

Introduce cases involving equivalent negative fractions, mixed numbers or operations under the scheme’s actual level definition. Compare how each entry classifies them.

A counterexample showing one record includes cases the other excludes is a reason to investigate further before merging.

Back to contents

13. A related concept may look nearly identical

Evaporation and Boiling both concern liquid changing to gas, and educational pages may use many of the same words.

Yet their processes and conditions differ. Merging them under one concept solely to simplify the index would harm scientific precision.

Prefer distinct concept identities with an appropriate broader or related link. The same principle applies to inference and prediction in English reading.

Back to contents

14. The danger of obligation words

Required and Recommended may appear together in study notices. They both concern how a person should approach an activity, but they imply different expectations.

Combining them into one status or subject concept can cause real errors if a learner treats optional practice as mandatory or an actual requirement as merely advice.

Treat distinctions affecting permission, eligibility or obligation as high-impact boundaries. Similarity in context is not sufficient grounds for a merge.

Back to contents

15. A merge can be appropriate after accidental duplication

A staff member creates a second subject record because the preferred name was spelled differently. The records are later found to have identical scopes and resources.

If no legitimate distinction is lost, merging can preserve one concept identity and make the unused spelling an alternative label.

Document why the identity decision was made so a future editor does not recreate the duplicate when they encounter the older form.

Back to contents

16. Choose a surviving concept carefully

A common strategy is to retain a stable, widely referenced identifier whose scope already matches the intended final concept. The other entry can be deprecated or redirected under a documented policy.

But age alone does not make an ID appropriate. If both old concepts had misleading definitions, a new replacement identity might be safer.

Select the identity based on scope continuity, existing links and governance, not a simplistic rule such as ‘always keep the lowest number’.

Back to contents

17. The surviving ID must preserve meaning

Imagine the source concept covers every reading strategy while the target covers only bridging inference. Merging source into target would make a broad historic ID appear narrow.

That is not an ordinary duplicate consolidation; it changes the intended subject. A parent relationship, split or separate mapping may be more accurate.

Ask whether users who saved the old concept ID would still be directed to a representation of the same intended idea after the merge.

Back to contents

18. Retired IDs should not be reassigned

When a concept ID is retired after an authorised merge, keep enough history to explain its former use and approved replacement.

Do not reuse that identifier for an unrelated subject. Historic records might still contain it, and reuse could silently alter what those records mean.

The Concept Identifier guide explains why identity needs continuity beyond changing labels.

Back to contents

19. Retain older labels as appropriate

Readers may continue searching the older preferred label long after a merge. A reviewed alternative label can preserve the route to the surviving concept.

The label should not be retained as an unrestricted synonym if it historically covered a different scope. A scope explanation or choice page may be required.

Correct search behaviour matters more than a neat record with the maximum possible aliases.

Back to contents

20. When a choice page is better than redirecting

If the former entry covers several related but non-identical concepts, a one-to-one redirect can mislead users.

Rather than force a merge, a short choice page can direct readers to the right successor topic. The Disambiguation Page guide explains that navigational problem.

Use the route that preserves the earlier reader intent. A convenient redirect is not itself proof of exact conceptual equivalence.

Back to contents

21. Moving resources requires content review

A merged concept record may inherit article-to-subject links from both originals. Some links may have been correct under one old scope but wrong under the new one.

Inspect the resource’s actual teaching purpose, not only the title string. A worksheet containing the phrase equivalent fractions may focus chiefly on another skill.

Reclassification is an editorial decision that benefits from positive and negative example checks.

Back to contents

22. Do not merge article content indiscriminately

Two subject records may consolidate while their teaching articles remain different. One article can be introductory; another can contain challenging proof, original diagrams or practice questions.

Deleting or joining whole posts solely because their subject concepts merged can destroy useful material and established citations.

Preserve distinct learning outcomes and only edit duplicate article content after a separate reader- and quality-focused review.

Back to contents

23. Mappings and hierarchies may need repair

Each old concept may have broader, narrower or related links. After the merge, copying every relation automatically can produce contradictions.

A narrower child of one original may be genuinely appropriate; a broad-match mapping from the other may be incompatible with the new scope.

Review each high-impact relationship with its source definition. The merged record should not become an accumulation of unexamined arrows.

Back to contents

24. Exact-match mappings need extra caution

SKOS skos:exactMatch is transitive. An erroneous exact correspondence can propagate through several schemes.

If one merged concept inherited an invalid exact mapping, the error can affect many downstream search results. Review strong mappings before and after consolidation.

See the SKOS mapping specification for the formal relations and their differing properties.

Back to contents

25. Search results can improve after a genuine merge

A reader searching either of two equivalent labels might now reach one coherent subject collection, with relevant resources previously scattered across duplicate entries.

This can improve discoverability when the search and navigation systems use the updated aliases and ID mappings correctly.

Do not claim that a merge automatically raises search ranking or reader success. The actual retrieval implementation must be tested.

Back to contents

26. False merges can reduce search precision

If two distinct subjects are consolidated, the result page can become crowded with material that does not answer the reader’s question.

A pupil seeking word recall practice may receive paragraph-writing activities that serve a different immediate task.

Use the Relevance Judgement guide to test whether the combined collection still serves specific information needs.

Back to contents

27. An unchanged URL can conceal a changed subject

An editor preserves the surviving concept’s URL while significantly broadening its scope. Historic articles linking to it now direct readers to a category different from the original meaning.

Check those references and decide whether they remain truthful. Keeping a permalink technically live is not the same as preserving its intended destination.

A historic explanation or a more precise new link can be appropriate when the meaning changed.

Back to contents

28. Respect old source citations

A cited resource may have used an earlier subject label that was correct for its time. The merge should not silently change a quotation or citation to pretend the source used today’s preferred term.

Maintain the original evidence and add a current navigation or terminology note where helpful.

This protects provenance and makes the history of the concept intelligible to research readers.

Back to contents

29. The merge decision needs an owner

Someone should have authority under the local scheme to approve the consolidation, record why it is justified and coordinate updates to dependent records.

A proposed match from a language model or search engine can be reviewed, but it is not automatically an approved governance decision.

Keep Proposed, Approved, Implemented and Verified as distinct states when the change affects published content.

Back to contents

30. The role of provenance

A merge record should identify original concept IDs, the surviving ID, compared scope evidence, reviewer, decision date and affected classifications.

The W3C PROV-O specification provides a way to represent entities, activities and responsibility in structured provenance data.

A small library may use a simpler change log. The principle is the same: a later editor should be able to understand what was combined and why.

Back to contents

31. The Nomenclature history example

The Nomenclature documentation describes a merge event that records changes to the target and removal of the source concept from the active scheme.

It also retains information for deprecated concepts and historic changes, helping users trace earlier terms after the consolidation.

Treat this as a specific implementation example. The standard for a different project may require another representation or additional fields.

Back to contents

32. A merge can be provisional before publication

Two records may look equivalent but lack clear definitions. Editors can mark the candidate pair Needs Evidence instead of immediately removing one.

A review might request a positive example, an excluded neighbour and the original source scheme’s intended scope.

An unresolved status protects the collection from a false certainty that would be hard to reverse once many pages link to the merged record.

Back to contents

33. What to do when the evidence conflicts

One concept note includes vocabulary use in reading, while the other excludes reading and focuses on writing. The titles look close, but the definitions conflict.

Do not simply select whichever note has a more recent date. Ask whether one reflects an approved scope change, an error or a legitimate separate concept.

The right outcome may be keeping both, creating a broader parent or documenting a qualified relation rather than merging.

Back to contents

34. A merge can require a new parent concept

Two records overlap heavily but neither accurately covers the other’s full meaning. Instead of combining them into one existing ID, editors may introduce a broader concept and retain the two narrower entries.

This changes the hierarchy without erasing useful differences. The new parent can provide a shared browsing route.

Do not call the result a simple duplicate merge if the distinct child scopes remain active. Describe the actual structural operation.

Back to contents

35. A merge can reveal a mistaken earlier split

A library previously divided a subject into two categories that users and editors cannot reliably distinguish. A later review finds that both now have the same intended scope.

A carefully documented merge can repair that false division while preserving any genuinely unique articles attached to either entry.

The reversal should explain why the earlier distinction failed rather than suggest that categories must never be adjusted after publication.

Back to contents

36. A small Mathematics merge case

Record A is Equivalent Fractions. Record B is Fractions of Equal Value. Both use the same definition and include the same kinds of examples under the local subject scheme.

The editor may merge the concept records, keep a stable approved ID and preserve the other name as an alternative label. Specific exercise pages remain separate when they have different learning purposes.

Test searches for both names and one neighbouring topic such as addition of fractions. The first two should find the intended subject without falsely treating every fraction operation as equivalent-fractions work.

Back to contents

37. A Mathematics non-merge case

Record A is Comparing Fractions. Record B is Equivalent Fractions. Their content overlaps because establishing equivalence can support comparison.

However, one topic asks learners to order values, while the other concerns representations of equal values. They have different central learning outcomes.

Keep the concepts separate and link them through a suitable relation or lesson route. A shared worked example does not prove conceptual identity.

Back to contents

38. An English non-merge case

One index entry teaches reference chains: how pronouns and descriptions point to continuing referents. Another teaches bridging inference: how a reader connects distinct but related entities or events.

Both concern coherence in reading, yet one addresses identity tracking and the other supplies missing relationships. Their scope notes should preserve the distinction.

The Reference Chain and Bridging Inference guides demonstrate why near neighbours need not be merged.

Back to contents

39. A Science non-merge case

A school catalogue contains Evaporation and Boiling under a broader subject about changes of state. Both can describe liquid becoming gas.

The difference in conditions and process matters to scientific understanding. Treating the terms as exact aliases would mislead learners.

Preserve separate subject identities with an appropriate broader connection or comparison resource. The simpler catalogue would not be a more accurate catalogue.

Back to contents

40. A school-workflow merge case

Two status labels, Review Finished and Checks Complete, may be found to have identical authorised conditions within a fictional project. The group might choose one consistent label.

Before merging, verify that both really require the same factual, editorial and procedural checks. If one status includes approval and the other does not, they are not duplicates.

An operational definition determines whether the label change is safe for publication decisions.

Back to contents

41. A user-intent collision test

Write the top question readers expect each concept to answer. If both are truly the same and point to the same scope, that supports consolidation.

Then find a near-neighbour question that must not appear as an exact synonym. Use this to test whether the merge has overbroadened the category.

A strong merge reduces needless choice while preserving useful distinctions. It should not simply shorten the menu.

Back to contents

42. A positive and a negative search test

After merging the two equivalent-fractions concepts, search with both old and new labels and check whether the intended resource is accessible.

Then search for a distinct related concept such as percent conversion, and verify that the system has not misclassified it as an exact equivalent.

These tests provide specific evidence of search behaviour. The existence of a merged record alone is not proof that the site works.

Back to contents

43. A versioned migration register

Use columns for Retired Concept ID, Surviving Concept ID, Compared Scope, Approved Relation, Labels Preserved, Resources Reassigned, Review Date and Test Result.

Separate records that were moved from those merely linked or left untouched. A rollback or later audit should be able to identify the exact affected entries.

Do not collapse a completed migration into a one-line note saying ‘duplicates fixed’ when several relationships and historic references were changed.

Back to contents

44. A rollback must protect valid changes

A merge may later be found wrong because one source concept had a meaning the review missed. The team needs a way to separate the records again without deleting legitimate new work.

Preserve the old definitions, ID mappings and affected-resource log where the governance policy permits. A targeted correction is often safer than restoring an entire older database snapshot.

The Change Impact Analysis guide explains how to inspect dependencies before and after such a revision.

Back to contents

45. How to verify the published outcome

Check the surviving concept’s scope note, preferred label and alternatives. Verify that retired identifiers point to an accurate explanation or authorised replacement.

Inspect resources whose classification changed, the internal links, search aliases and crosswalks. Note which tests were actually performed and any unresolved limitations.

Do not report that an entire knowledge library is deduplicated because four local queries worked. State the observed scope of verification.

Back to contents

46. Seven merge mistakes

One: merging records from a title match alone. Two: using an old broad ID for a narrower target without noting the scope change. Three: copying every mapping without review.

Four: deleting unique teaching articles because their subject concepts merge. Five: forgetting old search labels or bookmarks. Six: treating a local navigation merge as official curriculum equivalence. Seven: reporting success before checking the live reader routes.

Each error calls for a focused repair grounded in what identity or relationship was actually mistaken.

Back to contents

47. Twelve diagnostic questions

1. Does same label prove same concept? No. 2. Can different labels represent one concept? Yes. 3. Is a merge always a deletion of an article? No. 4. Can an alias avoid a duplicate entry? Often.

5. Can two topics overlap without being identical? Yes. 6. Do stable IDs matter? Yes. 7. Should old records remain traceable? Generally yes under a responsible scheme. 8. Can strong crosswalks need review? Yes.

9. Does merging guarantee better search? No. 10. Can a candidate remain unresolved? Yes. 11. Is rollback different from deleting history? Yes. 12. What is the goal? One accurate conceptual identity where the apparent distinction was false.

Back to contents

48. Frequently asked questions

Is a concept merge the same as a concept split? No. A merge consolidates identities, while a split creates separately scoped concepts from a broader or ambiguous earlier record. What about synonym labels? They may be preserved without creating separate concepts.

Can unique articles survive the merge? Yes. Content identity and concept identity are different. Does a merge create official equivalence across curricula? No.

Why learn this as vocabulary? It teaches how names, definitions and identifiers can be governed so that different ways of asking lead to the same idea without losing necessary distinctions.

Back to contents

49. References and the next action

The W3C SKOS Reference supports the management of concept labels, notes and mapping relationships. The Nomenclature documentation demonstrates a real scheme’s merge and deprecation history, while W3C PROV-O supplies the vocabulary for tracing changes and their sources.

The school resource examples and proposed merge decisions here are fictional teaching cases. They are not an audit of any live WordPress collection or a verified equivalence between actual curricula.

Choose two similar subject entries. Write the reader question, definition, examples and exclusions for each. Try to find a meaningful counterexample that separates them. If none appears under the intended scheme, propose a reviewed merge, preserve older access routes and test one correct and one deliberately wrong search.

Back to contents