eduKateSG · Why Science?
A search result becomes useful when you can explain why its meaning fits
Choose the closest question, or read the guide from the beginning. Each chapter keeps its explanation, example and next step visible.
A class is organising fictional observation notes. Several students wrote “size”, but one meant length, another meant mass, and a third meant the number of objects. A term search can return useful vocabulary, yet it cannot read the missing context in the students' minds. The Ontology Lookup Service, or OLS, makes ontology terms inspectable; scientific reasoning makes a proposed annotation defensible.
EMBL-EBI's OLS training describes searching, browsing and visualising ontology information. The official OLS4 repository identifies its public website and REST API. This article owns the term-inspection decision: how to move from source wording to a candidate whose definition can be checked. It complements eduKate's existing guides to Experimental Factor Ontology, measurement meaning and molecular-interaction vocabulary, which explain particular scientific domains.
Primary Science and PSLE Science learners can start with careful descriptions of ordinary objects. Secondary Science and computing students can extend that habit to data dictionaries and ontology identifiers. All source notes and proposed decisions in the worked activities are fictional, and generic labels used for explanation are not asserted to be real ontology terms. This is a reading and data-literacy guide, not clinical annotation advice or a promise that one service contains every possible concept.
Follow the connected evidence: Experimental Factor Ontology; Clinical Measurement Ontology; Molecular Interactions Controlled Vocabulary. For education and school-choice guidance, use the Education Hub; for possible work directions, continue to Careers by Subject Capability. These routes are separate from the specialist topic here.
| Layer | Question | Record | Boundary |
|---|---|---|---|
| Source record | What do we actually know? | Original text and context | Not a search query alone |
| Candidate | Which term might fit? | Identifier, definition and ontology | Not the first familiar label |
| Decision | Why was it selected? | Scope, alternatives and rationale | Not proof of the observation |
| History | What was used when? | Release, lookup date and reviewer | Not automatically the newest wording |
Inside this guide
Chapters 1–10 · Meaning and evidence
- 1. Did You Know? The best-looking label can be the wrong answer
- 2. Write the source question before opening the search box
- 3. OLS helps you inspect an ontology, not certify a dataset
- 4. Choose the ontology scope deliberately
- 5. Use several search phrases without changing the meaning
- 6. Read a candidate as a record, not a word
- 7. Identifiers keep similar names from collapsing together
- 8. Definitions reveal boundaries that synonyms may not
- 9. Relationships help interpretation but do not fill missing evidence
- 10. Review obsolete or superseded terms carefully
Chapters 11–20 · Comparison and learning
- 11. A worked term-selection decision
- 12. A no-match result can protect the record
- 13. Distinguish a lookup from a cross-ontology mapping
- 14. Preserve a lookup record that can be repeated
- 15. Programmatic lookup needs scientific checks too
- 16. A family activity on defining ordinary objects
- 17. A Secondary Science annotation review workshop
- 18. Connect term inspection to the wider eduKate ecosystem
- 19. Frequently asked questions about choosing terms
- 20. A final challenge: defend one annotation and one non-match
CHAPTER 1 OF 20 · Understand term lookup
1. Did You Know? The best-looking label can be the wrong answer
A label is a compact name. It may omit qualifications that appear in a definition or in the source context. Searching for a familiar word can therefore discover a candidate without establishing that the candidate fits. Treat lookup as the beginning of a comparison between meanings. This is an editorial review principle: the website supplies information, and the annotator still needs evidence for the decision.
Imagine a fictional note saying “leaf size increased”. A candidate referring specifically to leaf length might look attractive. But the note never tells us whether length, area or mass was observed. Selecting length would add information that the source does not contain. The honest next step is to inspect the original worksheet or ask for its measurement definition.
A useful search exercise has three possible outcomes: supported selection, rejection and unresolved meaning. All three can be good scientific work. Encourage learners to explain why a candidate was not chosen. A clear rejection can protect a dataset from a subtle mistake, while an unresolved entry tells the next reviewer exactly which detail would help the work continue.
CHAPTER 2 OF 20 · Understand term lookup
2. Write the source question before opening the search box
Begin with a short plain-language definition of the source variable. State what is being described, the relevant property and any necessary qualifiers. A search phrase is usually shorter than this definition, so preserve both. Otherwise the search process can gradually replace the original question with whatever result happens to be easiest to find.
Our fictional class record says “length of the longest leaf on each plant, measured once on day seven”. That is more informative than “leaf size”. It identifies the object, dimension, selection rule and time. Even if no one term captures all these details, the annotation record can retain them in separate fields. A controlled term should not erase the source's useful specificity.
Before searching, underline details that a candidate must match and details that belong elsewhere in the record. The term may describe a quantity while the date and plant identifier remain metadata. This simple preparation keeps the review focused. It also gives students a practical reason to write clear scientific sentences: better source wording makes every later classification decision easier to justify.
CHAPTER 3 OF 20 · Understand term lookup
3. OLS helps you inspect an ontology, not certify a dataset
EMBL-EBI's training presents OLS as a way to navigate ontology information. Its repository documents programmatic access as well as a website. These are access capabilities, not a certificate that a user's source data are correct. Sources: EMBL-EBI OLS training and OLS4 repository. Keep that boundary clear when teaching learners how a professional information service can support their work.
For a fictional observation, selecting a term for length does not prove that the ruler reading was taken correctly. Nor does it establish which plant was measured or whether the value was transcribed accurately. Vocabulary review, measurement review and record review answer different questions. A complete analysis may need all three, but one cannot substitute for another.
Use a layered notebook: original evidence, candidate meaning, annotation decision and observation-quality checks. When a mistake appears, identify the layer that failed. A wrong term may require remapping; a wrong numerical value requires a source correction. This makes repair more precise and prevents a well-known service name from becoming an excuse to skip the remaining scientific checks.
CHAPTER 4 OF 20 · Understand term lookup
4. Choose the ontology scope deliberately
An ontology is designed around a scope. Similar words can appear in vocabularies serving different scientific purposes, and the same everyday phrase can be used differently in different records. A recommended lookup review considers the originating ontology alongside the term. Ask whether its domain and intended distinctions suit the question you are trying to describe.
Suppose our fictional project needs a measurement type. A term describing an experimental factor, a biological structure or a measuring method might be relevant elsewhere, but it may not describe the quantity in the result column. Read the existing eduKate CMO, EFO and molecular-interaction guides for examples of different vocabulary jobs. These related articles help identify the question before searching for the answer.
Write a scope note when several ontologies seem plausible. “This field records the measured property; the method is stored separately” may be enough to explain the choice. The note prevents later reviewers from interpreting a reasonable annotation as a statement about a different layer. Good scope selection reduces ambiguity without requiring the learner to memorise every ontology's complete contents.
CHAPTER 5 OF 20 · Understand term lookup
5. Use several search phrases without changing the meaning
Search is a discovery activity. Alternative wording can reveal candidates that the first phrase misses, but changing wording should not quietly change the scientific question. Keep a small search log: the original phrase, alternative spelling or synonym, ontology scope and useful candidates. A unsuccessful search is informative only when another reviewer can see what was actually tried.
For the fictional leaf record, try the explicitly defined property rather than the broad word “size”. If the source uses an abbreviation, expand it only when documentation establishes its meaning. An abbreviation can be ambiguous across projects. Do not select an expansion because it produces more results or because the first returned term sounds scientifically sophisticated.
Return to the source definition after each promising search. Does the candidate match the same entity, property and qualifiers? If not, reject it with a short reason. This is a practical way to prevent search drift. Students can work in pairs, with one person searching and the other holding the original definition, then switching roles to see which assumptions became visible.
CHAPTER 6 OF 20 · Review a candidate term
6. Read a candidate as a record, not a word
A recommended candidate record includes its label, identifier, definition, source ontology and any relevant status or relationships available in the selected release. Read these fields together. An isolated label is easy to copy but difficult to audit. An identifier makes the selection specific, while the definition explains the meaning that justified it.
In a fictional exercise, Candidate A has a familiar short label but concerns a whole organism. Candidate B has less familiar wording but clearly describes the specific structure in the source note. The second may be the better fit. The exercise is about reasoning from the definition, not a claim that such candidates exist in any particular live ontology.
Keep a brief extract or reference to the definition used, with the lookup date and release information where available. Do not copy a large ontology page into the notebook. The purpose is to preserve the decision's evidence, not to duplicate the source. A future reviewer should be able to identify the exact candidate and understand why it was accepted or rejected.
CHAPTER 7 OF 20 · Review a candidate term
7. Identifiers keep similar names from collapsing together
Human-readable labels make a catalogue understandable; identifiers distinguish the actual concepts being referenced. A label can be reused in another ontology or revised during maintenance. A recommended annotation record therefore keeps the full identifier and source ontology instead of storing only a copied display name. The reader should not have to guess which concept the author intended.
Our fictional class invents local cards called “length” and “length”. One concerns a paper strip; the other concerns a selected leaf measurement. Their identical display labels do not make them identical concepts. Assign clearly local card identifiers and retain their definitions. When professional ontology terms are used, use the actual supplied identifier rather than constructing one from a label.
Check identifiers independently from spelling. A beautifully written label beside the wrong identifier is still a wrong annotation. When transferring records between files, compare both fields and preserve the original source text. This makes it easier to detect accidental mismatches introduced by copying, sorting or lookup formulas. The principle is simple enough for a classroom and valuable enough for a serious data workflow.
CHAPTER 8 OF 20 · Review a candidate term
8. Definitions reveal boundaries that synonyms may not
Synonyms can help people discover terms, but a matched synonym still needs definition review. A synonym may be broad, narrow or otherwise qualified depending on how the ontology records it. Do not treat every alternative name as an unrestricted equivalence claim. Inspect the available synonym information and the underlying meaning before deciding what the match supports.
For a fictional example, an everyday word might refer to several dimensions while a scientific term concerns only one. A search engine can still return that term because the word is useful for discovery. The annotator must decide whether the source actually supports the narrower meaning. Search relevance and annotation validity remain separate questions.
Ask learners to write one sentence explaining the candidate boundary and one counterexample outside it. If they cannot produce a counterexample, they may be relying on the label rather than understanding the definition. This activity makes scientific reading active. It invites the learner to explore where a concept stops, which often reveals more than repeating the preferred name or memorising a list of synonyms.
CHAPTER 9 OF 20 · Review a candidate term
9. Relationships help interpretation but do not fill missing evidence
Ontology relationships can make a candidate easier to understand by showing surrounding concepts. EMBL-EBI's OLS training includes visualisation and browsing. Source: OLS training. However, the existence of a nearby or more specific concept does not establish that the source record belongs there. The hierarchy supplies context for review; the source supplies evidence for selection.
Imagine a fictional broad category with three more specific kinds beneath it. The source says only that an observation belongs to the broad category. Selecting one of the three because it appears in the diagram would invent detail. A broader supported annotation is preferable, with the unresolved distinction kept visible for later review.
Use neighbouring concepts as a comparison set. What additional information would justify each candidate? Write those requirements down and check the source. This changes browsing from passive exploration into a useful decision tool. Students can practise with ordinary classification cards before specialist ontologies, learning that a more detailed branch is an option only when the observations support the extra distinction.
CHAPTER 10 OF 20 · Review a candidate term
10. Review obsolete or superseded terms carefully
Scientific vocabularies change. A term recorded in an older dataset may be marked obsolete or otherwise changed in a newer release. A recommended review preserves the historical identifier and meaning while recording any proposed update separately. Do not silently replace an old term with the first active-looking result, because the scientific interpretation may change with it.
For a fictional example, a broad local class is later split into two narrower classes. Historical notes do not contain the distinguishing detail. Neither newer class can be assigned confidently to every old record. Retain the older interpretation for historical analysis and mark the update unresolved where appropriate. The existence of replacements does not create missing source evidence.
When a release supplies replacement or consideration information, inspect its intended meaning rather than treating every pointer as an automatic conversion rule. Document the decision and reviewer. A clean historical record helps others reconstruct the work as it was performed. A separate current annotation can support new analysis without pretending that the original authors had information they never recorded.
CHAPTER 11 OF 20 · Build an annotation decision
11. A worked term-selection decision
Let's return to the fictional leaf project. The original note is “longest leaf length on day seven”, with a worksheet confirming that centimetres were used. The field's job is to describe the measured quantity. The plant identifier, observation date and ruler method are stored separately. This preparation gives the lookup a clear target and prevents one term from being asked to carry the entire experiment.
Review three fictional candidate descriptions: a general plant trait, a leaf-length quantity and a leaf-area quantity. The first is too broad for the available detail; the third concerns a different property. The second is the best fit within this invented comparison. No actual ontology identifier is assigned here because the exercise teaches the decision structure rather than asserting a live term lookup result.
The decision note should state what was selected, what evidence supported it and why the alternatives were rejected. It should also preserve the local selection rule “longest leaf”, even if that detail is outside the selected quantity concept. A useful annotation improves consistency while retaining source context. The outcome is an inspectable decision, not merely a cell containing a controlled label.
CHAPTER 12 OF 20 · Build an annotation decision
12. A no-match result can protect the record
Sometimes no candidate fits. The source may be vague, the chosen ontology may be unsuitable, or a concept may genuinely be absent from the available release. A no-match outcome is not proof that the observation is meaningless. It is a reason to preserve the original description and explain what remains unresolved instead of forcing a convenient term.
In a fictional note, “looks healthier” has no defined observation rule. Searching many ontologies cannot recover a measurement that was never specified. The productive response is to ask what was observed: leaf colour, count, height or another documented feature. If the record cannot be clarified, retain the wording as qualitative source text and avoid treating it as a specific numerical trait.
Recommended unresolved fields include the source phrase, attempted scope, candidates considered and the detail needed to decide. For a professional vocabulary contribution, follow the selected project's current process and provide evidence for the requested concept. For a student exercise, a clearly labelled local definition may be enough. In either setting, honesty about uncertainty supports future improvement better than a confident invented match.
CHAPTER 13 OF 20 · Build an annotation decision
13. Distinguish a lookup from a cross-ontology mapping
Finding a term and asserting a relationship between terms are different tasks. OLS helps inspect ontology information. If a source uses a term from one vocabulary and the target system uses another, a mapping decision needs its own evidence. A similar label or a shared search result does not automatically establish exact equivalence between the concepts.
Use a fictional example with two local vocabularies. One class defines “large” as above a fixed length; another defines it relative to that day's group average. Their labels match, but their decision rules differ. A lookup can reveal both definitions. It cannot justify merging them as equivalent. Keep the mapping relation and rationale separate from the term-inspection record.
This boundary helps organise a larger workflow. First inspect the source and target concepts. Then review any proposed mapping, including its provider, relation and scope. Finally decide whether the mapping is appropriate for the intended analysis. Each step should remain visible so a later reviewer can identify whether a disagreement concerns the terms themselves, the relationship asserted or its application to the dataset.
CHAPTER 14 OF 20 · Build an annotation decision
14. Preserve a lookup record that can be repeated
A useful lookup log includes source wording, ontology scope, search phrase, candidate identifier, release information, date and decision. These are editorial recommendations for reproducible review, not an official OLS data-entry form. The record need only be detailed enough to explain choices that materially affect interpretation. A small project may use a simple table.
Imagine another student reviews the same variable a month later and finds different candidates. The log helps determine whether the difference comes from a revised source definition, a changed release, a different search scope or a new judgement. Without it, the disagreement can look arbitrary. With it, the group can compare specific evidence and update the record deliberately.
Keep original decisions when they are revised. Add the new rationale and indicate which annotation is used in the current analysis. Do not erase history merely to make the table look uniform. A transparent revision trail demonstrates learning and makes reuse safer: readers can understand both the current decision and the earlier result that may still appear in a published report.
CHAPTER 15 OF 20 · Build an annotation decision
15. Programmatic lookup needs scientific checks too
The OLS4 repository documents a REST API, and the OLS search guidance describes programmatic search fields. Sources: OLS4 repository and OLS API guidance. An API can make repeated lookup practical, but automation does not remove the need to inspect meaning. A script that consistently chooses the wrong candidate can scale an error very efficiently.
Recommended automated review separates candidate retrieval from final acceptance. Save the original input, returned candidates and the rule used to select or flag a result. Validate the rule on difficult examples before applying it widely. Include ambiguous phrases, missing definitions and identifiers from another ontology. The expected behaviour may be a review flag rather than an automatic match.
Students interested in coding can build a local simulation with invented term cards before using a live service. The challenge is to produce a clear decision trail, not merely a fast answer. Check the current service documentation for actual endpoints and limits rather than copying a remembered URL. This article does not prescribe an API contract or claim that every deployment behaves identically.
CHAPTER 16 OF 20 · Practise scientific reading
16. A family activity on defining ordinary objects
For younger learners, begin with paper cards rather than ontology software. Ask a child to describe a button, leaf drawing or paper strip using a property that another person can observe. Keep the activity safe and avoid collecting personal or health information. The aim is to practise clear description and classification, not to introduce professional terminology before it helps.
Give the child three candidate definitions, all fictional: “longer than the reference strip”, “heavier than the reference object” and “contains more pieces than the reference set”. Ask which fits a length comparison and why the others do not. Then remove the reference strip from the instructions. Discuss how the missing detail changes what can be decided.
Parents looking for Primary Science tuition or PSLE Science practice can use this activity to strengthen explanation. Praise the reasoning that supports a selection and the observation that makes a selection impossible. Both are useful scientific moves. A child who says “I need to know what size means here” is asking a precise question that can improve the whole exercise.
CHAPTER 17 OF 20 · Practise scientific reading
17. A Secondary Science annotation review workshop
Prepare six fictional variables for a group workshop: final length, length change, mass, count, colour description and an ambiguous “size” field. Give each group a small local vocabulary with definitions and clearly local identifiers. Ask them to annotate supported variables and leave ambiguous ones unresolved. No live scientific term is required for the basic lesson.
Have a second group review the decisions without speaking to the first. They should inspect the source definitions and rationale, not guess the first group's intentions. Record disagreements by type: different interpretation of the source, different reading of a candidate, missing evidence or an accidental identifier mismatch. This makes feedback actionable and prevents all disagreements from being treated as careless work.
Advanced learners can repeat part of the exercise using a chosen ontology and OLS, following its current documentation. Keep the local exercise and real annotation records separate. The learning outcome is a defensible decision process: identify the field's job, inspect meaning, preserve evidence and explain uncertainty. That process transfers across Biology, Chemistry, Physics and data literacy without promising a particular examination or admission result.
CHAPTER 18 OF 20 · Practise scientific reading
18. Connect term inspection to the wider eduKate ecosystem
The recently published CMO article asks what quantity a number represents. EFO concerns experimental variables, while the molecular-interaction guide examines vocabulary for interaction evidence. This OLS guide offers the practical inspection step that can support those domain questions. Use the related pages when a term's purpose is unclear, rather than trying to solve every scientific distinction in one search box.
Reading and Science learning reinforce each other here. A learner needs to identify qualifiers, distinguish a broad statement from a precise definition and explain why a piece of evidence supports a choice. Mathematics becomes relevant when the definition includes a quantity, denominator or interval. Computing becomes relevant when decisions must remain traceable across many records.
For broader planning, continue through the Science Learning Hub and Education Hub. Use the career capability page to explore how clear reasoning and data organisation relate to possible work interests. School choice and formal qualifications should be checked through the relevant current official information. Completion of a lookup exercise is a useful learning experience; it does not itself establish admission eligibility or a guaranteed professional pathway.
CHAPTER 19 OF 20 · Practise scientific reading
19. Frequently asked questions about choosing terms
Should I always take the first result? No. Search ranking is a discovery aid, and the source definition still needs to match. Is an exact-looking label enough? No. Check the identifier, ontology and meaning. Can a broader term be acceptable? Yes, when it faithfully represents what the source supports and its limitations are recorded rather than hidden.
What if the source wording is unclear? Preserve it and identify the missing detail. Do not let a candidate definition rewrite the evidence. What if two reviewers disagree? Compare their source interpretations and candidate boundaries, then document the reason for the final decision. A disagreement can reveal useful ambiguity even when neither reviewer made a copying mistake.
Do I need to browse every related term? Use the surrounding concepts that clarify the decision, and keep the review proportionate to the task. What should go into a report? State the ontology and release used, explain material mapping choices and retain unresolved cases. A concise but inspectable method is more useful than a claim that terminology was standardised without showing how the decisions were made.
CHAPTER 20 OF 20 · Practise scientific reading
20. A final challenge: defend one annotation and one non-match
Choose two fictional source records. One should have a clear definition; the other should lack a critical qualifier. For the first, select a candidate from a local vocabulary or a documented ontology and explain why its definition fits. For the second, show why the available evidence does not support a precise selection. Keep the two outcomes equally visible.
Ask a partner to challenge both decisions. Can they identify the exact source and candidate? Can they see what information would change the unresolved outcome? Revise the log if the answer is no. The exercise is complete when another reader can understand the reasoning without relying on private conversation or the annotator's memory.
That is an optimistic answer to “Why Science?” Scientific learning gives us the confidence to inspect a meaning, make a supported choice and recognise when more information is needed. OLS provides a useful window onto vocabulary; the learner supplies careful reading and evidence-based judgement. Begin with one well-described record, and let clarity grow through a decision another person can check.
Back to contents · Previous chapter · Continue to the Science Learning Hub
Sources and scope
Official and primary sources checked on 15 October 2026: EMBL-EBI: OLS training; OLS4: official repository; OLS: search API guidance; EMBL-EBI: ontologies and data management. Project descriptions above are attributed to their named sources. Worked datasets are fictional; record designs, review checklists and teaching activities are editorial recommendations, not official ontology requirements. No search-volume, clinical, admission or career-outcome claim is made.