You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For NLQ "四川扬琴是否有代表性的曲目,这些曲目可能用到了哪些乐器?"(translated as "Does 'Sichuan Yangqin' have representative pieces, and what instruments might be used in these pieces?")--chatGPT cannot generate a perfect SPARQL
#41
Open
candlecao opened this issue
Jan 18, 2025
· 1 comment
As for this question, the expected entities of OWL are: {ctm:MusicType ctm:PieceWithPerformance mo:Instrument} {bf:instrument ctm:representativePiece ctm:samplePieceWithPerformance}
Actually the generated SPARQL is not incorrect. We ought not to owe the incompleteness to a malfunction of ChatGPT. Instead, it was due to that "have representative pieces" corresponds to both an objectProperty and a dataProperty:
however, GPT likely failed to incorporate :samplePieceWithPerformance in SPARQL because the literal meaning of that URI is less closely aligned with "have representative pieces" compared to :representativePiece
As for this question, the expected entities of OWL are:
{ctm:MusicType ctm:PieceWithPerformance mo:Instrument}
{bf:instrument ctm:representativePiece ctm:samplePieceWithPerformance}
Based on the extracted subgraph of ontology:
ChatGPT generated a version of SPARQL:
The text was updated successfully, but these errors were encountered: