Zum Inhalt springen

English:Search ranking, relevance and source evaluation - AI Year 5

Aus MOOCsWiki Staging
aiMOOC-Siegel aiMOOC

Search ranking, relevance and source evaluation - AI Year 5

QR-Code



Introduction

International library books on shelves as a context for finding and evaluating information
Finding a source is only a start. Compare relevance and evidence before accepting a claim. The six search-result cards in this lesson are invented.

Why is the first result not automatically the best evidence? Compare six fictional search cards, change a ranking signal and verify two real science claims. Keep popularity, relevance, source responsibility and supported conclusions separate while practising safe, teacher-supported information research.


Define the search need and inspect sources


Decide what the search is for

Our information need is how bees pollinate flowers. A useful source should explain the process at an appropriate level and support its claims. A very popular page about an unrelated topic does not meet that need. Search position, attractive design and click count are clues about a system's ordering, not proof that a particular statement is true.

A search system collects or indexes available material, matches a query, and orders candidates using selected signals and objectives. Different systems can use different mixtures of programmed rules and learned models. Our classroom ranking rules are written by us; this course does not train or reproduce a commercial search engine.

CodeAI: The Internet: How Search Works. Listen for the distinction between finding possible pages and ranking them. Use teacher support and the six fictional cards below. The recording explains general ideas; it is not a measurement of every search engine's current operation, a guarantee about a result's truth or evidence that our example scores are a real company's algorithm.


Inspect sources as well as snippets

Use five questions: who is responsible; what is the page's purpose; what evidence supports the claim; how relevant is it to this task; and does its date/context fit the claim? A named institution is worth investigating, not a magic truth label. A university page about economics can be irrelevant to bee biology. A blog can provide a useful observation without establishing a universal scientific rule.

A search snippet is a short extract, not the full argument or its context. Open an approved source, locate the relevant passage and check what it actually says. Distinguish a site's own claim about itself from independent evidence. If a result is labelled as advertising, recognise its commercial role; payment or an advertisement's presence is not scientific evidence. Our cards do not describe a particular engine's advertising policy.


Work a source-choice example

For how bees pollinate flowers, a current, relevant university explainer with identifiable expertise and references is a stronger starting point than an unrelated popular post. Explain why: it addresses the actual question, allows you to examine responsibility and provides evidence. Do not stop at its institution name or new date. Locate the scientific statement and compare it with an appropriate independent source.

Publication date matters differently for different needs. A stable definition of pollination may remain useful in an older institutional glossary; today's opening times or a current regulation need up-to-date confirmation. A recent date alone cannot repair an unsupported claim. If an actual page does not show a clear publication/update date, record that uncertainty instead of inventing one.

The real library photograph supplies an information-search context. It is not a screenshot of a search engine or evidence that the fictional pages below exist. You can complete the sorting experiment on paper without a personal search account or sharing browsing history.



Compare six results and ranking rules


Compare six fictional results

The six cards and their dates are invented teaching data, not real search results or endorsements of real institutions. R is an assigned relevance score from zero to three. E is an assigned evidence/support indicator from zero to three. P is a made-up popularity signal from zero to three, not an actual click total or a probability of truth. The ratings describe this exercise's stated assessment; inspect sources rather than treating a score as unquestionable fact.

Card Invented result description R E P
A Fictional university science explainer,2026; identified science author, clear references and a direct pollination explanation. 3 3 1
B Fictional public-museum guide,2025; named education team, helpful explanation and supporting sources. 3 2 1
C Fictional popular short clip,2026; dramatic bee claims, no supporting sources and many clicks. 2 0 3
D Fictional university economics page,1998; documented work on an unrelated topic. 0 2 2
E Fictional flower-shop page,2026; primarily selling products, with an unsupported claim that all flowers need bees. 1 0 3
F Fictional garden diary,2024; relevant local observations with limited scope and one supporting reference. 2 1 0

Try choosing two cards for the science task and justify each with three checks. A and B are strong initial candidates under the supplied descriptions: direct relevance, identifiable responsible teams and supporting sources. That is a starting selection, not a completed verification of every claim. F may contribute a limited observation if its scope is respected. D's institution does not make an unrelated subject relevant.


Apply an explicit evidence-focused toy rule

For Model One, compute score as twice R plus E. Sort from highest score to lowest. If scores tie, put the alphabetically earlier card ID first. This tie-break is an arbitrary display convention, not extra evidence of authority. Write the expected order before looking at the worked checks.

Card Twice R plus E Model One position
A 9 1
B 8 2
F 5 3
C 4 4
D 2 5
E 2 6

For A, the calculation is two times three plus three, nine. D and E both score two, so D comes first by the stated ID rule. That does not prove D is the more reliable source for the biology question. A useful numerical ordering still requires a source review and an understanding of what the indicators mean.


Change the objective, then change one signal in a controlled comparison

Model Two uses R plus P instead. With the same cards, query and fictional user context, the order becomes C, A, B, E, D, F. This rule gives more influence to the assigned popularity signal. C now comes first despite having no supporting sources in its description. The change demonstrates why a high position or many clicks cannot prove authority.

Now hold Model Two and every other card value fixed, but change only C's P from three to zero. Recompute the order: A, B, E, C, D, F. C, D and F now tie at two, so their IDs determine the display order. Recording what was held constant makes this a controlled comparison of one signal, not a story that quietly changed the query, weights and data at once.


Watch for feedback and limits

A system that uses interaction signals can be affected by what people choose to view. If visible items attract more clicks and those clicks influence later ordering, a feedback effect may occur. It is not proof that any one engine always behaves that way, nor that the most-clicked claim becomes scientifically correct. Users can seek another source, refine the question and use available controls rather than equating a personalised suggestion with an independent fact.

Our table contains only six candidates and three simplified signals. It has no actual personal profile, live click tracking, learned ranking model or assessment of all relevant pages. New sources, missing dates and different reading-accessibility needs would require more information and new checks. Passing this toy calculation does not certify real-world search quality.



Verify claims transfer and keep evidence


Verify two science claims using real institutional sources

The following sources are real and separate from our six fictional cards. First check the claim that pollination involves moving pollen from a flower's anther to a stigma. The University of Maryland Extension Agriculture Terms & Definitions glossary, adapted from the USDA, gives that core definition. The Royal Horticultural Society's How plants reproduce guidance also explains the transfer. The anther is the pollen-producing flower part; the stigma receives pollen in this process.

Second test the claim that every flower needs bees for pollination. RHS guidance explains insect pollination and wind pollination, including grasses as wind-pollinated plants. That supplies a counterexample to the universal claim. Bees are important pollinators in many contexts, but they are not the only possible agents. Pollination and the later process of fertilisation should not be treated as identical just because a short search snippet puts them together.

Claim to check Real source evidence Supported conclusion
Pollination involves pollen moving from anther to stigma University of Maryland Extension institutional glossary; RHS reproduction guidance Supported at this core definitional level
Every flower needs bees for pollination RHS explains wind-pollinated plants such as grasses The universal claim is false; retain the stated scope

Use the exact source links in OER and Wikipedia below. Note the institution, responsible source and the relevant passage; compare scope and wording. The Maryland document is an institutional glossary adapted from the USDA, not a newly published field experiment. The RHS page identifies its advice team. Their agreement on a stable definition is useful, but it is not evidence that we personally conducted a pollination experiment.


Write a short evidence note

State the information need, choose two suitable starting sources and justify each with at least three checks. For actual verification, record the two claims, the real institutions/passages consulted, the conclusion and any remaining limitation. Distinguish fictional card scores from real citations. Do not invent a quotation, a publication date or an author because it would make the note look more complete.

Correct the misconception explicitly: high click count measures a form of popularity, not necessarily expertise, relevance or truth. A top result can be useful, but its position is not the reason a scientific claim is supported. The evidence must address that claim. Avoid repeating a universal statement from a shop or clip merely because it is confidently worded.


Transfer to a different information need

Suppose the new task is to find an age-appropriate original poem about bees for a class reading. The science explanation may help background knowledge, but it does not automatically meet the new creative-reading need. Reassess relevance, reading level, purpose and appropriate rights or school-approved use. Do not keep the old biology scores while pretending the objective has not changed.

For a practical local or time-sensitive question, you would also need actual current local evidence. Our invented cards cannot answer one. A safe fallback is to ask the teacher or an appropriate responsible person for help checking the source rather than clicking an unknown download or supplying private information to a suspicious page.


Keep a test log that admits what remains unknown

Retain the six-card table, Model One/Two orders, the D/E tie explanation and the controlled change to C's popularity. Add the two real claim checks, a limitation and one genuinely untested future case, such as a new result with no date or unclear responsible author. Mark it untested until you actually inspect it; do not silently assign confidence from its position.

Use fictional or teacher-approved non-personal work, no pupil names or private browsing profiles. This AI Year 5 route is editorial enrichment aligned with England's KS 2 search and data-evaluation expectations. A worksheet result shows reasoning about supplied evidence; it is not a trained search engine, a guarantee of online safety or a certification that every future result is reliable.



Interactive Tasks


Quiz: Test Your Knowledge

Which pair is the strongest starting selection for the science task under the fictional descriptions? (A and B) (!C and E because they are most popular) (!D and E because they tie in one model) (!Only D because every university page is relevant)



What does a high click count by itself establish? (Popularity under the stated measure) (!Scientific truth) (!Relevant expertise for every question) (!A correct publication date)



What is card A’s Model One score? (Nine) (!Seven) (!Four) (!Three)



Why does D precede E when both have Model One score two? (The stated alphabetic ID tie-break puts D first.) (!D has been proved scientifically correct.) (!E has no right to appear in a list.) (!The score guarantees a universal source-quality order.)



Which card leads Model Two with the original data? (C) (!A) (!B) (!F)



What happens when only C’s popularity signal changes from three to zero in Model Two? (The order becomes A B E C D F.) (!All other source claims become automatically true.) (!A trained model is created from the six cards.) (!The biology query changes to a poem request automatically.)



Which real source supports the anther-to-stigma definition? (The University of Maryland Extension institutional glossary) (!The invented shop card alone) (!A click count without a source) (!A fictional card ID with no text)



What counters the claim that every flower needs bees? (RHS explains wind-pollinated plants such as grasses.) (!All pages with new dates must be correct.) (!A popular clip contains no references.) (!The library photograph contains books.)



Why can an older glossary still help with this task? (A stable definition can remain useful when its source and scope are checked.) (!Old material is always better than newer evidence.) (!Publication dates never matter for any query.) (!The glossary proves today’s local opening times.)



What must change for a new poem-reading information need? (Reassess relevance purpose reading level and appropriate use.) (!Keep the old scores as if the task were unchanged.) (!Assume the top science explainer is already an original poem.) (!Choose only the result with the highest click signal.)





Memory Game

Query The information need expressed as a search request
Relevance How well a source addresses the actual task
Evidence Support that can be checked for a particular claim
Ranking Ordering candidates using selected signals and objectives
Popularity An interaction or attention measure rather than a truth guarantee
Tie-break A stated convention for ordering equal scores





Drag and Drop

A high position in results Inspect the source rather than assuming the claim is true
An unrelated authoritative institution Check relevance to this question as well as expertise
Two equal toy scores Apply the stated display tie-break without inventing extra authority
A different information need Reassess which signals and sources matter
A universal scientific claim Seek relevant evidence and a valid counterexample if one exists





Crossword Puzzle

QUERY What names the search request?
SOURCE What should you inspect beyond the snippet?
RANKING What orders candidates using chosen signals?
EVIDENCE What supports a checkable claim?
RELEVANCE What concerns fit to the actual information need?
CONTEXT What surrounding situation helps interpret a source?





LearningApps

Optional extra practice. Choose an appropriate activity with your teacher. If the external collection is unavailable, use the native activities and independent tasks in this course.


Cloze Text

Complete the explanation.
The information request is a

. Fit to that need is

. A claim needs checkable

. Ordering candidates is

. Many clicks can measure

, not guarantee truth. Equal scores use the stated

. Card

leads Model One. Card

leads the original Model Two. RHS gives

pollination as a counterexample to every flower needs bees. An unresolved source question can go to the

.



Open Tasks


Easy

  1. State the information need: Explain what how bees pollinate flowers is asking for.
  2. Separate popularity and authority: Give a reason why many clicks do not prove a science claim.
  3. Find the responsible source: Identify who is behind an approved page or mark uncertainty.
  4. Explain the tie-break: Show why D comes before E without claiming extra scientific quality.


Standard

  1. Compare six fictional results: Choose two for the science task and justify each using three checks.
  2. Recompute the controlled comparison: Hold Model Two fixed and change only C’s popularity, recording the new order.
  3. Verify two actual claims: Use the real institutional sources, record passages and respect scope.
  4. Keep an evidence note: Include the toy ranking, test result, real claim checks, limitation, untested case and human fallback.


Advanced

  1. Compare ranking objectives: Explain why Model Two can promote an unsupported popular source.
  2. Transfer to the poem task: Change the relevance criteria and justify appropriate reading/use choices.
  3. Examine possible feedback: Explain how interaction signals can influence visibility without making claims true.
  4. Evaluate a source’s date: Contrast stable definitions with a genuinely time-sensitive query without inventing dates.



Learning Check

  1. Required source-choice worked case: Explain why a current university explainer can be a stronger start than an unrelated popular post.
  2. Model One scores and boundary tie: State the evidence-focused order and the equal-score rule.
  3. Model Two and one-signal change: Explain both changed orders while holding the stated context fixed.
  4. Two independently checked claims: Identify the real sources and supported scope.
  5. Three checks per selected source: Write an evidence-based justification rather than a popularity claim.
  6. Transfer and evidence limits: Explain a new query and a safe unresolved-case route.


Worked checks and assessment guidance

  1. Use task relevance, identified responsibility and supporting evidence, not institution name or date alone. Under our fictional descriptions, A and B are strong starting candidates. Actually inspect their claimed evidence; the real verification below uses separate named institutional pages rather than invented search-card URLs.
  2. Twice R plus E gives A 9, B 8, C 4, D 2, E 2, F 5. The order is A, B, F, C, D, E. D precedes E solely because alphabetic card ID is the declared tie-break. Equal numeric scores and a display convention do not prove equal or superior scientific authority.
  3. R plus P gives C 5, A 4, B 4, E 4, D 2, F 2, so the order is C, A, B, E, D, F. With only C’s P changed to zero, C falls to 2 and the order becomes A, B, E, C, D, F. The query and other values remain fixed. These are hand-written toy rules, not a commercial learned ranking algorithm.
  4. The University of Maryland Extension glossary adapted from the USDA and RHS reproduction guidance support pollen transfer from anther to stigma. RHS also describes wind-pollinated plants, including grasses, which counters every flower needs bees. Bees remain important in many contexts. The course did not conduct an experiment or establish every species’ pollination method.
  5. Check responsibility/expertise, purpose, relevant supporting passage, relevance and date/context. Use at least three for each of two sources. A stable glossary definition can remain useful, while a present-day local fact needs current evidence. A recent date, domain label or many clicks cannot independently certify a claim.
  6. An age-appropriate poem request changes the objective: reassess genre, reading level, purpose and approved use rather than keeping biology scores unchanged. Retain the toy data, version, orders, tie, controlled change, two real claim checks, limitation and untested future result. Ask a teacher for help with uncertainty; the worksheet is not a trained engine or a guarantee about all future results.


Learning evidence

Keep six fictional cards, the two original ranking orders, a tie explanation and the controlled C-popularity change. Select two suitable starting sources and justify each with three checks. Record two actual institutional claim verifications, their scope, a limitation, one untested future result and a human fallback. No private pupil browsing data, trained search engine, universal safety guarantee or invented real search-card URL is involved.


OER and Wikipedia

DfE: KS 2 Computing supplies search/data-evaluation context; this separate AI Year 5 sequence is editorial. Google: ranking results is the provider’s description of signals/context, not an independent guarantee of truth or a disclosure of our toy formula.

University of Maryland Extension: Agriculture Terms & Definitions, adapted from the USDA, was checked for the core pollination definition. RHS Advice Team: How plants reproduce was checked for that definition and insect/wind pollination. Six cards, their dates/scores and all exercises are original fiction, separate from these real references. Reviewed 14 November 2026.

International books at Kent Library.jpg, Roc0ast3r, CC0. Unchanged real-library photograph used as an information-search context, not a screenshot or evidence for the invented results.

https://www.youtube.com/watch?v=LVV_93mBfSU — CodeAI: The Internet: How Search Works. Original publisher recording embedded unchanged.



Linked learning areas


aiMOOC Projects



Schulfach+




aiMOOCs



aiMOOC Projekte









MOOCwiki · Deutsch

Nach dem Lernen ist vor dem Lernen

Entdecke direkt den nächsten Lernkurs. Weitere Inhalte erscheinen, wenn Du weiter nach unten scrollst.

Zur MOOCwiki-Hauptseite

Mediathek

Suche verfeinern

Dein Lernplan bestimmt, was oben steht. Alle anderen Inhalte bleiben erreichbar. Seitenaufrufe beeinflussen die Auswahl nicht.

Inhalte werden geladen ...

Mediathek wird aus dem Wiki geladen ...