@graffiticode/l0175 0.3.0 → 0.3.1

package/spec/examples/c1-t2-tm5-shorttext.gc

@@ -0,0 +1,20 @@
+target c1-t2 passage "The Blue Ribbon" type literary lines [
+  "Every spring, Tessa entered her drawing in the county fair."
+  "Every spring she came home with nothing."
+  "This year she almost did not enter at all."
+  "Her grandmother set a fresh pencil on the kitchen table."
+  "Tessa drew the same barn eleven times before she kept one."
+  "At the fair, the blue ribbon went to somebody else."
+  "Then a teacher asked to hang Tessa's drawing in the hall."
+  "Tessa walked home taller than she had the year before."
+]
+claims [
+  claim id "c1" status supported dimension summary subject "the fair" text "After the ribbon went to someone else, a teacher asked to hang Tessa's drawing in the hall, and she walked home proud." cites ["e6" "e7" "e8"] {}
+]
+evidence [
+  source id "e6" line 6 status directly-supports supports ["c1"] quote "At the fair, the blue ribbon went to somebody else." {},
+  source id "e7" line 7 status directly-supports supports ["c1"] quote "Then a teacher asked to hang Tessa's drawing in the hall." {},
+  source id "e8" line 8 status directly-supports supports ["c1"] quote "Tessa walked home taller than she had the year before." {}
+]
+outcomes [ outcome id "q1" type short-text task-model tm5 dimension summary subject "the fair" focus "c1" stem "Summarize what happens after the blue ribbon goes to somebody else. Use key details from the passage in your summary." {} ]
+{}..

package/spec/examples.md

@@ -1,135 +1,105 @@
 <!-- SPDX-License-Identifier: CC-BY-4.0 -->
 # L0175 RAG Training Examples
 
-_Revised: 2026-06-30_
+_Revised: 2026-08-28_
 
 Natural-language prompts for training a RAG model on L0175 — composing 5th-grade ELA
-assessment items (Smarter Balanced · Grade 5 · Claim 1) for learning targets **T4**
-(`target c1-t4`, Reasoning & Evidence, literary), **T11** (`target c1-t11`, Reasoning & Evidence,
-informational), **T9** (`target c1-t9`, Central Ideas, informational), **T8** (`target c1-t8`,
-Key Details, informational), and **T10** (`target c1-t10`, Word Meanings, informational). Each
-program declares its `target` first. Categories 1–6 below are literary (T4); Category 7 is
-informational R&E (T11); Category 8 is Central Ideas (T9); Category 9 is Key Details (T8);
-Category 10 is Word Meanings (T10). Category 11 is the **task-model coverage matrix** — at least
-one specific request for every task model in every target (task-model numbers are per-target).
-
-## Category 1: EBSR items (two-part selected response)
-
-1. Write an EBSR item about the main character's motivation in a story about a girl at a tide pool.
-2. Compose a two-part item: Part A asks what can be inferred about the narrator's feelings; Part B asks for the supporting line.
-3. Make an EBSR item about the setting and how it shapes the mood of the scene.
-4. Build an EBSR item about an event and why the character reacts the way she does.
-5. Write an EBSR item about the relationship between two characters, standard rl-3.
-6. Compose an EBSR item about the author's point of view, standard rl-6.
-7. Make an EBSR item about the theme of the passage.
-
-## Category 2: Hot Text items (select text)
-
-8. Make a hot-text item where students click the statement that best describes the character, then click the supporting sentences.
-9. Compose a hot-text item where Part B asks students to highlight every sentence that shows the character's focus.
-10. Write a hot-text item about the narrator's feelings toward another character.
-
-## Category 3: Short Text items (constructed response)
-
-11. Write a short-text item asking what inference can be made about the character, with evidence.
-12. Compose a short-text constructed-response item about the theme, hand-scored 0–2.
-13. Make a short-text item about how the setting affects the character, with a custom rubric.
-
-## Category 4: Authoring the inference graph
-
-Each prompt composes a complete item; the focus is how the claim/evidence graph is built.
-
-14. Compose an EBSR item about a girl named Mara at a tide pool: author a supported claim that she cares more about the tide pool than her family's picnic, and make that claim the question's focus.
-15. Compose a literary item with two questions about the tide-pool story — one built on a supported claim that Mara is absorbed by the pool, and a separate question built on a supported claim that her brother wants her attention.
-16. Compose an EBSR item whose focus inference is directly supported by two of the passage's lines — tag both sources directly-supports and cite them from the correct claim.
-17. Compose an EBSR item that includes a supports-wrong-claim source: real evidence that props up an 'anger' misreading rather than the correct inference, linked to both the correct claim and the anger distractor.
-18. Compose an EBSR item whose last two passage lines are tagged irrelevant so they can serve as Part B foils.
-
-## Category 5: Distractors by error type
-
-Each prompt composes a complete item; the focus is the error-typed distractor it features.
-
-19. Compose an EBSR item about Mara at the tide pool with a misreads-detail distractor that takes her silence as anger, with a rationale that names the error and targets the question.
-20. Compose an EBSR item about Mara at the tide pool with an erroneous-inference distractor that she dislikes being outdoors.
-21. Compose an EBSR item about Mara at the tide pool with a faulty-reasoning distractor that mistakes her whisper for fear.
-22. Compose an EBSR item whose distractors each carry a rationale explaining the student error they target and are tagged with the question id(s) they foil.
-
-## Category 6: Item-first composition
-
-23. Compose the questions first: write an EBSR question, a short-text question, and a hot-text question, each with its stem from the guideline catalog, then author foils targeting each.
-24. Compose one question per inference dimension the passage supports, then author a targeted foil set for each.
-25. Set a question's focus to a specific correct claim and write its stem, then author the distractors that target it.
-
-## Category 7: Informational items (target c1-t11)
-
-26. From an informational article about the history of bridge design, write an EBSR item about the relationships between the successive designs (dimension relationships-interactions, standard ri-3).
-27. Make an item about how the author uses evidence to support a point in a science article (dimension author-use-of-information, standard ri-8).
-28. Compose an EBSR item about the author's point of view in an informational passage about city planning (dimension point-of-view, standard ri-6).
-29. Write a short-text item asking what conclusion can be drawn about the author's opinion of renewable energy, with key evidence (dimension authors-opinion).
-30. Make a hot-text item about the author's purpose in an informational passage (dimension purpose).
-
-## Category 8: Central Ideas items (target c1-t9)
-
-A DIFFERENT skill from Reasoning & Evidence — the main idea, the key details that build it, and summary (DOK 2; standards ri-1+ri-2). Distractors are usually TRUE statements that just aren't central: too-narrow (a supporting detail), too-broad (an overgeneralization), insignificant (a minor detail), or misreads-detail.
-
-31. From an informational article about honeybees, write a multiple-choice item asking which sentence best states the main idea (dimension central-idea). Make the distractors a true supporting detail (too-narrow), an overgeneralization (too-broad), and a misread (misreads-detail).
-32. Make a multi-select item: choose the two sentences that should be included in a summary of the passage (dimension summary); `focus` lists the two correct claims, distractors are insignificant or too-broad statements.
-33. Compose an EBSR item where Part A asks for the main idea and Part B asks which detail best supports it (dimension central-idea, standard ri-2).
-34. Write a short-text item: "Determine the main idea of the passage. Explain using key details…" (dimension central-idea, DOK r-dok3).
-35. Write a multiple-choice "missing key detail" item: present a short summary in the stem and ask which key detail is missing; the focus claim is the missing detail.
-
-## Category 9: Key Details items (target c1-t8)
-
-A DIFFERENT model — the inference/conclusion is GIVEN in the stem and the student selects the supporting EVIDENCE (DOK 1–2; standards ri-1+ri-7; dimension supporting-evidence). Author ONE supported claim = the given inference (its `focus`), state it in the stem, and author `source`s as the options: directly-supports = correct evidence (with a `quote`), supports-wrong-claim/irrelevant = distractor evidence. No distractor claims.
-
-36. From an informational article about Roman aqueducts, write a multiple-choice item that states the conclusion "aqueducts brought water to distant cities" and asks which detail from the passage best supports it (dimension supporting-evidence).
-37. Make a multi-select item: which two details best support the stated conclusion? Select two answers (two directly-supports sources are the correct set).
-38. Make a single-part hot-text item: state the inference, then have students click the sentence(s) in the passage that support it (dimension supporting-evidence).
-
-## Category 10: Word Meanings items (target c1-t10)
-
-A DIFFERENT model — the question asks for the MEANING of a targeted word/phrase in context, so the options are meanings (DOK 1–2; standard ri-4 + the L-4 family). Author a top-level `words` list: a `word` (the targeted word, with line/quote for context) holding `meanings` — `status correct` (the answer) + `status distractor` (error-type other-meaning/misinterprets/wrong-context + rationale). The outcome's `focus` names the word; state the word + its sentence in the stem.
-
-39. From an informational article, write a multiple-choice item asking what the word "aqueduct" most likely means as used in the passage (dimension word-meaning, standard l-4a — context). Distractors: another meaning that ignores context (other-meaning), a misread (misinterprets), and a wrong-context meaning.
-40. Make a multi-select item: "What does the word 'channel' most likely mean? Choose two answers." (two correct meanings, dimension word-meaning).
-41. Write a roots/affixes item: ask what the root in a word means, standard l-4b (dimension word-meaning).
-42. Make a click-the-word item (hot-text): "Read the dictionary entry … Click the word in the paragraph that matches this definition." Author the candidate `word`s — the correct one is the outcome's `focus` with the `line` of its paragraph, plus a few distractor candidate words from that same paragraph (`text` only, no `meanings`). The compiler shows the paragraph and makes those authored candidates clickable; the focus word is correct. (Author only the correct word and every content word in the paragraph becomes a choice.) Keep the passage out of the stem.
-
-## Category 11: Task-model coverage — one request per task model, per target
-
-Task-model NUMBERS are **per-target and collide** — TM3 is Short Text in T4/T11, EBSR in T9, and
-Hot Text in T8/T10 — so each request below names **both** the number and the item type (the way an
-author actually expresses the intent). This is the full author's-intent matrix: at least one
-specific request for every task model in every target, resolved against that target's table.
-
-**Target 4 (`c1-t4`) — Reasoning & Evidence, literary (RL):**
-
-43. For `c1-t4`, author a **Task Model 1 (EBSR)** item about a boy who hides his fear during a thunderstorm — Part A infers what he feels, Part B gives the supporting line (dimension character, standard rl-1/rl-3).
-44. For `c1-t4`, author a **Task Model 2 (Hot Text)** item whose Part A is the best statement about the narrator's point of view in a first-person camp diary, then Part B highlights the supporting sentences (dimension point-of-view, standard rl-6).
-45. For `c1-t4`, author a **Task Model 3 (Short Text)** constructed-response item asking what the story's theme is and to explain it with evidence from the text (dimension theme, standard rl-2, DOK r-dok3).
-
-**Target 11 (`c1-t11`) — Reasoning & Evidence, informational (RI):**
-
-46. For `c1-t11`, author a **Task Model 1 (EBSR)** item about how two bridge designs solve the same problem — Part A the relationship between them, Part B the supporting detail (dimension relationships-interactions, standard ri-3).
-47. For `c1-t11`, author a **Task Model 2 (Hot Text)** item whose Part A is the best statement of the author's purpose in an article on recycling, then Part B highlights the supporting sentences (dimension purpose, standard ri-8).
-48. For `c1-t11`, author a **Task Model 3 (Short Text)** item asking what conclusion can be drawn about the author's opinion of solar power, explained with key evidence (dimension authors-opinion, DOK r-dok3).
-
-**Target 9 (`c1-t9`) — Central Ideas, informational (RI-1/RI-2):**
-
-49. For `c1-t9`, author a **Task Model 1 (Multiple Choice)** item asking which sentence best states the central idea of an article about coral reefs (dimension central-idea, standard ri-2); distractors too-narrow, too-broad, insignificant.
-50. For `c1-t9`, author a **Task Model 2 (Multi-Select)** item asking which two sentences belong in a summary of a passage about the water cycle (dimension summary); `focus` is the two correct claims.
-51. For `c1-t9`, author a **Task Model 3 (EBSR)** item — Part A asks for the central idea of an article on honeybee colonies, Part B asks which detail best supports it (dimension central-idea, standard ri-2). *(NB: in `c1-t9`, TM3 is EBSR — not Short Text as in T4/T11.)*
-52. For `c1-t9`, author a **Task Model 4 (Hot Text)** single-part item: click the sentence(s) that best show the main idea of an article about volcanoes (dimension central-idea, standard ri-2).
-53. For `c1-t9`, author a **Task Model 5 (Short Text)** item asking the student to determine the central idea and explain it with key details from the passage (dimension central-idea, DOK r-dok3).
-
-**Target 8 (`c1-t8`) — Key Details, informational (RI-1/RI-7):**
-
-54. For `c1-t8`, author a **Task Model 1 (Multiple Choice)** item that states the conclusion "aqueducts carried water to distant cities" and asks which detail best supports it (dimension supporting-evidence, standard ri-7); the options are sources, not statements.
-55. For `c1-t8`, author a **Task Model 2 (Multi-Select)** item asking which two details best support a stated conclusion about migrating birds (the two directly-supports sources are the correct set).
-56. For `c1-t8`, author a **Task Model 3 (Hot Text)** single-part item: state the inference, then have students click the sentence(s) in the passage that support it (dimension supporting-evidence).
-
-**Target 10 (`c1-t10`) — Word Meanings, informational (RI-4/L-4):**
-
-57. For `c1-t10`, author a **Task Model 1 (Multiple Choice)** item asking what "aqueduct" most likely means as used in the passage (dimension word-meaning, standard l-4a — context); distractors other-meaning, misinterprets, wrong-context.
-58. For `c1-t10`, author a **Task Model 2 (Multi-Select)** item: "What does the word 'channel' most likely mean? Choose two answers." (two correct meanings, dimension word-meaning).
-59. For `c1-t10`, author a **Task Model 3 (Hot Text)** click-the-word item: click the word in the paragraph that matches a given definition; the focus word is correct plus a few distractor candidate words from the same paragraph (standard l-4a). Keep the passage out of the stem.
+assessment items (Smarter Balanced · Grade 5 · Claim 1). They are written the way a fifth-grade
+teacher would ask for an item, and they stay away from specific passage content so the model
+learns the task shape, not a topic. Each program declares its `target` first.
+
+This is a covering set: every task model of every target appears at least once, and across the
+task models each target's dimensions are each covered at least once. Standard is a function of
+dimension, so covering dimensions covers the standards too — except in Word Meanings, where the
+one dimension spans four L-4 strategies (context, roots and affixes, dictionary, word
+relationships), which are covered instead. The dimension is carried by the skill the prompt
+describes, not a tag.
+
+Item type maps to a task-model number within each target, and the numbers collide across targets —
+a short-text item is tm3 in T4/T11 but tm5 in T2/T9; a hot-text item is tm2 in T4/T11, tm4 in T2/T9, and
+tm3 in T1/T8/T10 — so each prompt names its target and task model.
+
+## Target 4 (`c1-t4`) — Reasoning & Evidence, literary
+
+Students read a story and reason about it — a character's feelings or motives, how two characters
+relate, the setting, an event, the narrator's point of view or feelings, the theme, or the topic —
+and back it up with evidence from the text.
+
+1. c1-t4 tm1 — Make a two-part item for a story: in Part A students figure out what the main character is feeling or why they act the way they do, and in Part B they pick the line that best backs it up.
+2. c1-t4 tm1 — Make a two-part item for a story about an important event: in Part A students figure out why it happens or how a character reacts to it, and in Part B they pick the line that best supports their answer.
+3. c1-t4 tm1 — Make a two-part item for a story about two characters: in Part A students figure out how the two feel about each other, and in Part B they pick the line that best shows it.
+4. c1-t4 tm2 — Make an item for a story where students first click the sentence that best tells the narrator's point of view, then click the sentence or two that support it.
+5. c1-t4 tm2 — Make an item for a story where students first click the sentence that best shows how the setting shapes the mood, then click the sentences that support it.
+6. c1-t4 tm2 — Make an item for a story where students first click the sentence that best shows how the narrator feels about another character, then click the sentences that support it.
+7. c1-t4 tm3 — Make a written-response item for a story: ask students what the theme is and have them explain it using details from the text.
+8. c1-t4 tm3 — Make a written-response item for a story: ask students what the story is mostly about and have them support it with details from the text.
+
+## Target 11 (`c1-t11`) — Reasoning & Evidence, informational
+
+The same kind of reasoning as Target 4, but for an informational passage — how ideas or events
+connect, how the author uses information to support a point, the author's point of view or
+purpose, or the author's opinion — always supported with evidence.
+
+9. c1-t11 tm1 — Make a two-part item for an informational passage: in Part A students figure out how two ideas or events are connected, and in Part B they choose the detail that best supports it.
+10. c1-t11 tm1 — Make a two-part item for an informational passage: in Part A students figure out the author's point of view, and in Part B they choose the detail that best supports it.
+11. c1-t11 tm2 — Make an item for an informational passage where students click the sentence that best states the author's purpose, then click the sentences that support it.
+12. c1-t11 tm2 — Make an item for an informational passage where students click the sentence that best shows how the author uses facts to support a point, then click the sentences that support it.
+13. c1-t11 tm3 — Make a written-response item for an informational passage: ask students what conclusion they can draw about the author's opinion and have them back it up with evidence from the text.
+
+## Target 2 (`c1-t2`) — Central Ideas, literary
+
+Students find the theme of a story or poem, the key details and events that build it, or summarize
+part of it. The theme is a lesson or message, not a plot event. Wrong answers are usually true but
+not central. Summaries here always cover a section or a key event, never the whole story.
+
+14. c1-t2 tm1 — Make an item for a story that asks which sentence best tells the theme.
+15. c1-t2 tm2 — Make an item for a story where students choose the two sentences that belong in a summary of it.
+16. c1-t2 tm3 — Make a two-part item for a story: in Part A students pick the theme, and in Part B they choose the sentence that best supports it.
+17. c1-t2 tm4 — Make an item for a story where students click the one or two sentences that best show the theme.
+18. c1-t2 tm5 — Make a written-response item for a story: ask students to summarize what happens after a key moment, using details from the text.
+
+## Target 9 (`c1-t9`) — Central Ideas, informational
+
+Students find the main idea of an informational passage, the key details that build it, or a
+summary. Wrong answers are usually true but not central — a small supporting detail, an
+overgeneralization, or a misread. T2 is its literary twin; the two are the only targets that use all five task models.
+
+19. c1-t9 tm1 — Make an item for an informational passage that asks which sentence best states the main idea.
+20. c1-t9 tm1 — Make an item that shows students a short summary of an informational passage with one key detail left out and asks which detail is missing.
+21. c1-t9 tm2 — Make an item for an informational passage where students choose the two sentences that belong in a summary.
+22. c1-t9 tm3 — Make a two-part item for an informational passage: in Part A students pick the main idea, and in Part B they choose the key detail that best supports it.
+23. c1-t9 tm4 — Make an item for an informational passage where students click the sentence or two that best show the main idea.
+24. c1-t9 tm5 — Make a written-response item for an informational passage: ask students to figure out the main idea and explain it using key details from the text.
+
+## Target 1 (`c1-t1`) — Key Details, literary
+
+The same job as Target 8, but on a story instead of an article. Give students a conclusion about
+the story and have them find the line or lines that support it. Wrong answers are other lines that
+tell about the story without backing up that particular conclusion.
+
+25. c1-t1 tm1 — Give students a conclusion about a story and make an item asking which line from the story best supports it.
+26. c1-t1 tm2 — Give students a conclusion about a story and make an item asking which two lines best support it.
+27. c1-t1 tm3 — Give students a conclusion about a story and make an item where they click the sentence or two that support it.
+
+## Target 8 (`c1-t8`) — Key Details, informational
+
+The conclusion is handed to the students; their job is to find the evidence. Give them an
+inference about an informational passage and have them pick the detail(s) that support it. The
+wrong answers are other details that don't actually support it.
+
+28. c1-t8 tm1 — Give students a conclusion about an informational passage and make an item asking which detail from the passage best supports it.
+29. c1-t8 tm2 — Give students a conclusion about an informational passage and make an item asking which two details best support it.
+30. c1-t8 tm3 — Give students a conclusion about an informational passage and make an item where they click the sentence or two that support it.
+
+## Target 10 (`c1-t10`) — Word Meanings, informational
+
+Students figure out what a word or phrase means the way it is used in an informational passage.
+The answer choices are meanings; wrong ones come from another meaning of the word, a misread, or
+the wrong context. Items lean on context, but can also use a word's roots and affixes, a dictionary
+entry, or word relationships like synonyms and antonyms.
+
+31. c1-t10 tm1 — Make an item that asks what a key word most likely means the way it is used in an informational passage.
+32. c1-t10 tm1 — Make an item that asks what a word means by using its Greek or Latin root or prefix.
+33. c1-t10 tm2 — Make an item that asks students to choose the two answers that best give the meaning of a key word as it is used in an informational passage.
+34. c1-t10 tm2 — Make an item that asks students to choose the two words that mean about the same as a key word from the passage.
+35. c1-t10 tm3 — Give students a short dictionary definition and make an item where they click the word in the paragraph that matches it.