Annotating Training Data for Conditional Semantic Textual Similarity Measurement using Large Language Models