Jian-Yun Nie
How search behaves differently when the query is a conversation in a non-English language — and how language models are changing that picture.
Nie spent most of his career on cross-lingual and multilingual information retrieval — work that was a niche specialty when the field was largely English-centric, and that turned out to be directly relevant once LLM-based search products started shipping to non-English markets. His ConvGQR research on conversational generative query reformulation extends the same thread: when a multi-turn dialogue happens in a less-resourced language, the rewriting step that makes retrieval possible has to do more work, and many of the established assumptions break down. For ai100, which audits AI engines in five languages, this is the literature that maps which assumptions in standard IR don't survive the language switch.