@inproceedings{cui-etal-2026-llm,
title = "{LLM} Knows, Speaks Not: Extracting Decoder-State Representations for Column Type Annotation",
author = "Cui, Lingxi and
Zhang, Jianfeng and
Li, Huan and
Chen, Ke and
Shou, Lidan",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1107/",
pages = "23427--23441",
ISBN = "979-8-89176-462-0",
abstract = "Column Type Annotation (CTA) enriches tables with semantic column types that are often missing or incomplete in real-world data, aiding multiple downstream table tasks. Yet CTA remains challenging because sparse or noisy column values often require table-level context for disambiguation. Prior work either flattens tables into encoder inputs, losing structure and semantics, or follows a description-then-encoding pipeline that separates encoding and adds generation cost. We propose D2R, a decoder-based framework that obtains column representations directly from the hidden state that initiates column description generation. Specifically, D2R first encodes the table context once as a shared prefix KV cache, providing table-level context for all target columns. For each column, we append its sampled values and a description trigger token, then extract the trigger{'}s hidden state as the column embedding, preserving semantics without producing text. A decoder-attached classifier, trained via contrastive learning aligned with the CTA objective, predicts column types directly from this embedding. Across four CTA benchmarks, D2R achieves the best Micro-F1, outperforming TURL, a strong table PLM, by up to 12.57 points, while retaining efficient inference. The code is available at https://github.com/IvesCheung/cta."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cui-etal-2026-llm">
<titleInfo>
<title>LLM Knows, Speaks Not: Extracting Decoder-State Representations for Column Type Annotation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lingxi</namePart>
<namePart type="family">Cui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianfeng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huan</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ke</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lidan</namePart>
<namePart type="family">Shou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Column Type Annotation (CTA) enriches tables with semantic column types that are often missing or incomplete in real-world data, aiding multiple downstream table tasks. Yet CTA remains challenging because sparse or noisy column values often require table-level context for disambiguation. Prior work either flattens tables into encoder inputs, losing structure and semantics, or follows a description-then-encoding pipeline that separates encoding and adds generation cost. We propose D2R, a decoder-based framework that obtains column representations directly from the hidden state that initiates column description generation. Specifically, D2R first encodes the table context once as a shared prefix KV cache, providing table-level context for all target columns. For each column, we append its sampled values and a description trigger token, then extract the trigger’s hidden state as the column embedding, preserving semantics without producing text. A decoder-attached classifier, trained via contrastive learning aligned with the CTA objective, predicts column types directly from this embedding. Across four CTA benchmarks, D2R achieves the best Micro-F1, outperforming TURL, a strong table PLM, by up to 12.57 points, while retaining efficient inference. The code is available at https://github.com/IvesCheung/cta.</abstract>
<identifier type="citekey">cui-etal-2026-llm</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1107/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>23427</start>
<end>23441</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LLM Knows, Speaks Not: Extracting Decoder-State Representations for Column Type Annotation
%A Cui, Lingxi
%A Zhang, Jianfeng
%A Li, Huan
%A Chen, Ke
%A Shou, Lidan
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F cui-etal-2026-llm
%X Column Type Annotation (CTA) enriches tables with semantic column types that are often missing or incomplete in real-world data, aiding multiple downstream table tasks. Yet CTA remains challenging because sparse or noisy column values often require table-level context for disambiguation. Prior work either flattens tables into encoder inputs, losing structure and semantics, or follows a description-then-encoding pipeline that separates encoding and adds generation cost. We propose D2R, a decoder-based framework that obtains column representations directly from the hidden state that initiates column description generation. Specifically, D2R first encodes the table context once as a shared prefix KV cache, providing table-level context for all target columns. For each column, we append its sampled values and a description trigger token, then extract the trigger’s hidden state as the column embedding, preserving semantics without producing text. A decoder-attached classifier, trained via contrastive learning aligned with the CTA objective, predicts column types directly from this embedding. Across four CTA benchmarks, D2R achieves the best Micro-F1, outperforming TURL, a strong table PLM, by up to 12.57 points, while retaining efficient inference. The code is available at https://github.com/IvesCheung/cta.
%U https://aclanthology.org/2026.findings-emnlp.1107/
%P 23427-23441
Markdown (Informal)
[LLM Knows, Speaks Not: Extracting Decoder-State Representations for Column Type Annotation](https://aclanthology.org/2026.findings-emnlp.1107/) (Cui et al., Findings 2026)
ACL