@inproceedings{liu-liang-2026-ncl,
title = "{NCL} at {S}em{E}val-2026 Task 8: Deterministic Small-{LLM} {RAG} with Relation Classification",
author = "Liu, Zehao and
Liang, Huizhi",
editor = "Kochmar, Ekaterina and
Ghosh, Debanjan and
North, Kai and
Komachi, Mamoru",
booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
month = jul,
year = "2026",
address = "San Diego, California, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.semeval-1.191/",
pages = "1477--1481",
ISBN = "979-8-89176-414-9",
abstract = "We present NCL{'}s system for SemEval-2026 Task 8B, the generation track for multi-turn retrieval-augmented dialogues. Our submission follows a compact and reproducible RAG pipeline: (1) global and local question rewriting with LLM-based multi-turn relation control, (2) passage reranking with BGE-M3, (3) context-level answerability filtering with strict binary LLM judgments ({``}yes''/{``}no''), and (4) deterministic inference with a small-LLM (Qwen2.5-1.5B-Instruct) plus post-generation quality fallback (cleaning, bad-answer gate, one stricter retry, then an IDK fallback).On the official test set, our system achieved a harmonic mean score of 0.5973 (RB{\$}{\{}agg{\}}{\$} 0.4993, RL{\$}F{\$} 0.7235, RB{\$}{\{}llm{\}}{\$} 0.6105), ranking 19th out of 26 teams on the leaderboard."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-liang-2026-ncl">
<titleInfo>
<title>NCL at SemEval-2026 Task 8: Deterministic Small-LLM RAG with Relation Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zehao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huizhi</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 20th International Workshop on Semantic Evaluation (2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Kochmar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Debanjan</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">North</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mamoru</namePart>
<namePart type="family">Komachi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-414-9</identifier>
</relatedItem>
<abstract>We present NCL’s system for SemEval-2026 Task 8B, the generation track for multi-turn retrieval-augmented dialogues. Our submission follows a compact and reproducible RAG pipeline: (1) global and local question rewriting with LLM-based multi-turn relation control, (2) passage reranking with BGE-M3, (3) context-level answerability filtering with strict binary LLM judgments (“yes”/“no”), and (4) deterministic inference with a small-LLM (Qwen2.5-1.5B-Instruct) plus post-generation quality fallback (cleaning, bad-answer gate, one stricter retry, then an IDK fallback).On the official test set, our system achieved a harmonic mean score of 0.5973 (RB${agg}$ 0.4993, RL$F$ 0.7235, RB${llm}$ 0.6105), ranking 19th out of 26 teams on the leaderboard.</abstract>
<identifier type="citekey">liu-liang-2026-ncl</identifier>
<location>
<url>https://aclanthology.org/2026.semeval-1.191/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>1477</start>
<end>1481</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T NCL at SemEval-2026 Task 8: Deterministic Small-LLM RAG with Relation Classification
%A Liu, Zehao
%A Liang, Huizhi
%Y Kochmar, Ekaterina
%Y Ghosh, Debanjan
%Y North, Kai
%Y Komachi, Mamoru
%S Proceedings of the 20th International Workshop on Semantic Evaluation (2026)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, USA
%@ 979-8-89176-414-9
%F liu-liang-2026-ncl
%X We present NCL’s system for SemEval-2026 Task 8B, the generation track for multi-turn retrieval-augmented dialogues. Our submission follows a compact and reproducible RAG pipeline: (1) global and local question rewriting with LLM-based multi-turn relation control, (2) passage reranking with BGE-M3, (3) context-level answerability filtering with strict binary LLM judgments (“yes”/“no”), and (4) deterministic inference with a small-LLM (Qwen2.5-1.5B-Instruct) plus post-generation quality fallback (cleaning, bad-answer gate, one stricter retry, then an IDK fallback).On the official test set, our system achieved a harmonic mean score of 0.5973 (RB${agg}$ 0.4993, RL$F$ 0.7235, RB${llm}$ 0.6105), ranking 19th out of 26 teams on the leaderboard.
%U https://aclanthology.org/2026.semeval-1.191/
%P 1477-1481Markdown (Informal)
[NCL at SemEval-2026 Task 8: Deterministic Small-LLM RAG with Relation Classification](https://aclanthology.org/2026.semeval-1.191/) (Liu & Liang, SemEval 2026)
ACL