Sentence Similarity Search

This tool searches for semantically similar sentences in the Vedic Prose Corpus (VPC). It works on English machine translations of all VPC texts generated with Sebastian Nehrdich's Dharmamitra API. Therefore, your queries should resemble the style and vocabulary of these translations.

Example queries that work:

• "The stoma consists of 17 parts."
• "The gods drive away the cattle of the Asuras."
• "Cows are like Soma."
• "They dig a hole at the sacrificial ground."

What will not work well (or at all):

• Sanskrit text. This tool operates on English translations.
• Asking questions ("What is the meaning of the sacrifice?") or prompt instructions ("List all passages describing the agnistoma.") - this tool finds existing passages, it does not generate answers.
• Contemporary paraphrases or colloquial language
• Very short phrases or single words

Tips:

• Use complete, well-formed sentences.
• Try to match the register of Vedic translations.
• Try variations with different synonyms if initial results are poor.
• Similarity scores above 0.8 (strong matches) are marked with a star ⭐.
• Lower scores (0.6-0.8) may still contain relevant parallels worth exploring.

Technical details: The search uses all-MiniLM-L6-v2 finetuned with several thousand records (partly human judgments, partly prompt-generated).