179 questions
No questions match those filters.
← All questions
Retrieval & EmbeddingsMedium· Cross-lingual Retrieval· BM25· Bi-encoders
Users query in Swahili, your corpus is mostly English,...
This is one of the questions in the full AI/ML interview bank. Pro unlocks all 1789 questions; Premium includes the same bank plus the highest daily Practice limit.
See plansBM25 scores are based on IDF-weighted overlap; if there is no overlap, the score is zero, and the ranking is arbitrary. Tuning hyperparameters or adding synonyms cannot solve this fundamental lack of lexical intersection. A multilingual bi-encoder maps both languages into a shared semantic space, allowing for retrieval based on meaning rather than exact word matches.