Skip to content

QUESTION_FORMATS.md

Latest commit

 

History

History
332 lines (262 loc) · 7.99 KB

QUESTION_FORMATS.md

File metadata and controls

332 lines (262 loc) · 7.99 KB
Raw

Question Import Formats

The manual uploader accepts either:

{ "items": [ ... ] }

or a raw JSON array:

[ ... ]

Use strict JSON only. No Markdown fences, comments, or trailing commas. The app accepts both snake_case and camelCase field names, but these snake_case formats match the generator prompts.

Valores Del Se

Required item shape:

{
  "sentence": "Se venden pisos en este barrio.",
  "difficulty": 2,
  "se_value": "Pasiva refleja",
  "se_function": "Marca de pasiva",
  "accepted_functions": ["Marca de pasiva", "Sin funcion sintactica propia"],
  "verbal_structure": "Verbo simple",
  "periphrasis_type": "No aplica",
  "phrase_type": "Oracion simple pasiva refleja",
  "explanation": "El verbo concuerda con el sujeto paciente 'pisos'."
}

Allowed difficulty: 1, 2, 3.

Allowed se_value:

[
  "Reflexivo",
  "Reciproco",
  "Sustitucion (le -> se)",
  "Pronominal",
  "Pasiva refleja",
  "Impersonal",
  "Accidental (dativo de interes)"
]

Allowed se_function and accepted_functions values:

[
  "CD",
  "CI",
  "Sin funcion sintactica propia",
  "Morfema verbal",
  "Marca de pasiva",
  "Marca de impersonal"
]

Allowed verbal_structure:

["Verbo simple", "Perifrasis verbal", "Locucion verbal", "Dos verbos"]

Allowed periphrasis_type:

[
  "No aplica",
  "Modal obligativa",
  "Modal de posibilidad",
  "Aspectual ingresiva",
  "Aspectual incoativa",
  "Aspectual durativa",
  "Aspectual terminativa",
  "Aspectual reiterativa",
  "Aspectual resultativa",
  "Aspectual habitual"
]

Perifrasis

Important: this section is for the exercise category "construcciones verbales". It must include a mixed bank of:

  • true verbal periphrases
  • verbal locutions
  • cases with two independent verbs

Do not generate only true verbal periphrases unless explicitly asked.

Required item shape:

{
  "sentence": "Debes entregar el informe antes del viernes.",
  "difficulty": 2,
  "verbal_structure": "Perifrasis verbal",
  "periphrasis_type": "Modal obligativa",
  "phrase_type": "Oracion simple predicativa",
  "explanation": "'Deber + infinitivo' expresa obligacion."
}

Allowed difficulty: 1, 2, 3.

Allowed verbal_structure:

["Perifrasis verbal", "Locucion verbal", "Dos verbos"]

Target distribution when generating a balanced batch:

{
  "Perifrasis verbal": "40%",
  "Locucion verbal": "30%",
  "Dos verbos": "30%"
}

For exactly 20 items, use exactly:

{
  "Perifrasis verbal": 8,
  "Locucion verbal": 6,
  "Dos verbos": 6
}

Allowed periphrasis_type:

[
  "No aplica",
  "Modal obligativa",
  "Modal de posibilidad",
  "Aspectual ingresiva",
  "Aspectual incoativa",
  "Aspectual durativa",
  "Aspectual terminativa",
  "Aspectual reiterativa",
  "Aspectual resultativa",
  "Aspectual habitual"
]

Rule: if verbal_structure is "Perifrasis verbal", periphrasis_type must not be "No aplica". If verbal_structure is "Locucion verbal" or "Dos verbos", periphrasis_type must be exactly "No aplica".

Copy-paste prompt for 20 difficulty-2 items:

Generate exactly 20 high-quality items for the Spanish "perifrasis" exercise category at difficulty 2.
Important: "perifrasis exercise category" means a mixed drill about verbal constructions, not 20 true periphrases.

Return only strict JSON:
{ "items": [ ... ] }

Required distribution:
- 8 items with "verbal_structure": "Perifrasis verbal"
- 6 items with "verbal_structure": "Locucion verbal"
- 6 items with "verbal_structure": "Dos verbos"

For "Perifrasis verbal", choose a real periphrasis type from the allowed list and do not use "No aplica".
For "Locucion verbal" and "Dos verbos", set "periphrasis_type" exactly to "No aplica".

Every item must have:
- "sentence"
- "difficulty": 2
- "verbal_structure"
- "periphrasis_type"
- "phrase_type"
- "explanation"

Allowed verbal_structure values:
["Perifrasis verbal", "Locucion verbal", "Dos verbos"]

Allowed periphrasis_type values:
[
  "No aplica",
  "Modal obligativa",
  "Modal de posibilidad",
  "Aspectual ingresiva",
  "Aspectual incoativa",
  "Aspectual durativa",
  "Aspectual terminativa",
  "Aspectual reiterativa",
  "Aspectual resultativa",
  "Aspectual habitual"
]

Quality rules:
- Use natural Spanish sentences.
- Avoid duplicated verbs, sentence frames, and near-duplicates.
- Make difficulty 2: medium length, clear but not trivial.
- Include both infinitive, gerund, and participle periphrasis patterns where appropriate.
- Include locutions that are genuinely lexicalized, not true periphrases.
- Include "Dos verbos" cases where the two verbs are independent predicates, coordinated, juxtaposed, or in a subordinate construction.
- No Markdown, no prose, no comments, no trailing commas.

Morfologia

Required item shape:

{
  "word": "desordenados",
  "difficulty": 2,
  "word_type": "Adjetivo",
  "lexeme": "orden",
  "accepted_lexemes": ["orden"],
  "morphemes": ["des", "ad", "o", "s"],
  "morpheme_types": [
    "Prefijo derivativo",
    "Sufijo derivativo",
    "Morfema flexivo nominal (genero)",
    "Morfema flexivo nominal (numero)"
  ],
  "analysis_type": "Adjetivo con derivacion y flexion",
  "explanation": "Prefijo des- + lexema orden + sufijo -ad- + flexivos -o y -s."
}

Allowed difficulty: 1, 2, 3.

Allowed word_type:

["Sustantivo", "Adjetivo", "Verbo", "Adverbio"]

Allowed morpheme_types:

[
  "Prefijo derivativo",
  "Interfijo",
  "Sufijo derivativo",
  "Morfema flexivo nominal (genero)",
  "Morfema flexivo nominal (numero)",
  "Vocal tematica",
  "Morfema flexivo verbal (tiempo/modo/aspecto)",
  "Morfema flexivo verbal (persona/numero)"
]

Rules:

  • accepted_lexemes must include the exact lexeme.
  • morphemes must not include the lexeme.
  • morphemes and morpheme_types must have the same length and matching order.
  • Do not put decorative boundary hyphens in morphemes.
  • The app validates that the word rebuilds as prefixes + lexeme + remaining morphemes.
  • For word_type: "Verbo", use only personal conjugated forms, not infinitives, gerunds, or participles.

Sintaxis

Required item shape:

{
  "phrase": "Aunque llovia, salimos temprano.",
  "difficulty": 2,
  "correction": "**Analisis:** oracion compuesta por subordinacion adverbial concesiva.\n\n- **Subordinada:** \"Aunque llovia\"\n- **Principal:** \"salimos temprano\""
}

Allowed difficulty: 1, 2, 3.

Rules:

  • correction supports Markdown formatting.
  • Keep the JSON strict: escape line breaks inside correction as \n.
  • The uploader also accepts sentence or frase instead of phrase, but phrase is the preferred key.

Derivative

Required item shape:

{
  "function": "f(x)=x^3-5x^2+2x",
  "derivative": "f'(x)=3x^2-10x+2"
}

Rules:

  • Use LaTeX-compatible math strings.
  • The app renders both function and derivative with KaTeX.
  • The uploader also accepts function_to_deriv, functionToDeriv, functionText, or funcion instead of function.
  • The uploader also accepts answer or derivada instead of derivative.
  • The correction box is just the derivative value.

General Sintaxis Sentence Generator

This is the older sentence-generation response shape used by the syntax workspace code:

{
  "sentence": "La alumna resolvio el problema con calma.",
  "tokens": ["La", "alumna", "resolvio", "el", "problema", "con", "calma", "."],
  "targetFeatures": ["Oracion simple", "CD", "CC modo"],
  "appliedPolicies": {
    "punctuationSimplified": true
  }
}

The manual uploader added in the React app is for the three active sections: valores del se, perifrasis, and morfologia.

Prompt Template For ChatGPT

Ask:

Generate 20 high-quality items for [valores del se/perifrasis/morfologia].
Return only strict JSON in this exact shape:
{ "items": [ ... ] }
Use only the allowed labels listed below.
Every item must be grammatically correct, non-duplicated, and internally consistent.
No Markdown, no prose, no comments.
[paste the relevant section format and allowed label lists]