CTM: when you prompt GPT to generate SPARQL, if the natural language question is expressed in Chinese, it usually defaults to add @zh sign and related code, then the desired result cannot be retrieved

[candlecao]

e.g.,
NLQ: 《越南闲愁》涉及到什么音乐类型(乐种)、乐器、民族？

PREFIX ctm: <https://lib.ccmusic.edu.cn/ontologies/chinese_traditional_music#>
PREFIX dbo: <http://dbpedia.org/ontology/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?musicTypeLabel ?instrumentLabel ?ethnicGroupLabel
WHERE {
  # Link the resource "越南闲愁" with SpecialIndependentResource
  ?resource rdfs:label "越南闲愁"@zh ;
            ctm:relatesMusicType ?musicType ;
            ctm:relatesInstrument ?instrument ;
            ctm:relatesEthnicGroup ?ethnicGroup .

  # Get labels for music type, instrument, and ethnic group
  ?musicType rdfs:label ?musicTypeLabel .
  ?instrument rdfs:label ?instrumentLabel .
  ?ethnicGroup rdfs:label ?ethnicGroupLabel .
  
  # Language filter for Chinese labels
  FILTER (lang(?musicTypeLabel) = "zh")
  FILTER (lang(?instrumentLabel) = "zh")
  FILTER (lang(?ethnicGroupLabel) = "zh")
}

while the expected SPARQL is:

define input:inference 'urn:owl.ccmusicrules'
PREFIX ctm: <https://lib.ccmusic.edu.cn/ontologies/chinese_traditional_music#>
PREFIX dbo: <http://dbpedia.org/ontology/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?musicTypeLabel ?instrumentLabel ?ethnicGroupLabel
WHERE {
  # Link the resource "越南闲愁" with SpecialIndependentResource
  ?resource rdfs:label "越南闲愁" ;
            ctm:relatesMusicType ?musicType ;
            ctm:relatesInstrument ?instrument ;
            ctm:relatesEthnicGroup ?ethnicGroup .

  # Get labels for music type, instrument, and ethnic group
  ?musicType rdfs:label ?musicTypeLabel .
  ?instrument rdfs:label ?instrumentLabel .
  ?ethnicGroup rdfs:label ?ethnicGroupLabel .
}

[candlecao]

We can add labels with @zh in the future