@inproceedings{guo-etal-2026-exgql,
title = "{E}x{GQL}: Self-Improving Natural Language-to-Graph Query Generation via Experience-Guided Learning",
author = {Guo, Penghong and
Du, Boxin and
Zhou, Houliang and
Gupta, Chaman and
Gu, Yina and
K{\"u}llian, Rayssa and
Jiang, Shen},
editor = "Kamalloo, Ehsan and
Gontier, Nicolas and
Rozanov, Nikolai and
Hashemzadeh, Maryam and
Arora, Akhil",
booktitle = "Proceedings of the Second Workshop for {RE}search on Agent Language Models ({REALM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.realm-1.58/",
pages = "846--855",
ISBN = "979-8-89176-485-9",
abstract = "Natural language-to-graph query generation (NL2GQL) requires schema-aware reasoning and multi-hop traversal planning, posing graph-specific challenges beyond the extensively studied Text-to-SQL setting. Existing NL2GQL systems typically solve each query in isolation, limiting their ability to learn from prior executions. We present ExGQL, an experience-guided framework for self-improving NL2GQL. ExGQL combines hierarchical memory, experience evolution, and capability synthesis to distill execution feedback into reusable reasoning strategies and convert validated strategies into executable capabilities. We further introduce a Failure-Directed Curriculum (FDC) for a separate practice phase, in which recurring failure patterns guide the selection of tasks that address the system{'}s current weaknesses. We evaluate ExGQL on NL2GQL-Bench, a benchmark of 400 expert-annotated graph query tasks over a production-scale graph database. ExGQL achieves an average accuracy of 0.869 versus 0.662 for the strongest evaluated baseline (+31{\%} relative improvement), while reducing average query-generation time by 28{\%}. In a separate practice phase, FDC reaches comparable accuracy of 0.861 with 60{\%} fewer practice episodes and 36{\%} lower runtime than fixed-order practice. These results show that NL2GQL systems can improve by reusing execution-derived experience, while failure-directed practice makes experience acquisition more efficient."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="guo-etal-2026-exgql">
<titleInfo>
<title>ExGQL: Self-Improving Natural Language-to-Graph Query Generation via Experience-Guided Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Penghong</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Boxin</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Houliang</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chaman</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yina</namePart>
<namePart type="family">Gu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rayssa</namePart>
<namePart type="family">Küllian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shen</namePart>
<namePart type="family">Jiang</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>Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ehsan</namePart>
<namePart type="family">Kamalloo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicolas</namePart>
<namePart type="family">Gontier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikolai</namePart>
<namePart type="family">Rozanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maryam</namePart>
<namePart type="family">Hashemzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Akhil</namePart>
<namePart type="family">Arora</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">979-8-89176-485-9</identifier>
</relatedItem>
<abstract>Natural language-to-graph query generation (NL2GQL) requires schema-aware reasoning and multi-hop traversal planning, posing graph-specific challenges beyond the extensively studied Text-to-SQL setting. Existing NL2GQL systems typically solve each query in isolation, limiting their ability to learn from prior executions. We present ExGQL, an experience-guided framework for self-improving NL2GQL. ExGQL combines hierarchical memory, experience evolution, and capability synthesis to distill execution feedback into reusable reasoning strategies and convert validated strategies into executable capabilities. We further introduce a Failure-Directed Curriculum (FDC) for a separate practice phase, in which recurring failure patterns guide the selection of tasks that address the system’s current weaknesses. We evaluate ExGQL on NL2GQL-Bench, a benchmark of 400 expert-annotated graph query tasks over a production-scale graph database. ExGQL achieves an average accuracy of 0.869 versus 0.662 for the strongest evaluated baseline (+31% relative improvement), while reducing average query-generation time by 28%. In a separate practice phase, FDC reaches comparable accuracy of 0.861 with 60% fewer practice episodes and 36% lower runtime than fixed-order practice. These results show that NL2GQL systems can improve by reusing execution-derived experience, while failure-directed practice makes experience acquisition more efficient.</abstract>
<identifier type="citekey">guo-etal-2026-exgql</identifier>
<location>
<url>https://aclanthology.org/2026.realm-1.58/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>846</start>
<end>855</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ExGQL: Self-Improving Natural Language-to-Graph Query Generation via Experience-Guided Learning
%A Guo, Penghong
%A Du, Boxin
%A Zhou, Houliang
%A Gupta, Chaman
%A Gu, Yina
%A Küllian, Rayssa
%A Jiang, Shen
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Rozanov, Nikolai
%Y Hashemzadeh, Maryam
%Y Arora, Akhil
%S Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-485-9
%F guo-etal-2026-exgql
%X Natural language-to-graph query generation (NL2GQL) requires schema-aware reasoning and multi-hop traversal planning, posing graph-specific challenges beyond the extensively studied Text-to-SQL setting. Existing NL2GQL systems typically solve each query in isolation, limiting their ability to learn from prior executions. We present ExGQL, an experience-guided framework for self-improving NL2GQL. ExGQL combines hierarchical memory, experience evolution, and capability synthesis to distill execution feedback into reusable reasoning strategies and convert validated strategies into executable capabilities. We further introduce a Failure-Directed Curriculum (FDC) for a separate practice phase, in which recurring failure patterns guide the selection of tasks that address the system’s current weaknesses. We evaluate ExGQL on NL2GQL-Bench, a benchmark of 400 expert-annotated graph query tasks over a production-scale graph database. ExGQL achieves an average accuracy of 0.869 versus 0.662 for the strongest evaluated baseline (+31% relative improvement), while reducing average query-generation time by 28%. In a separate practice phase, FDC reaches comparable accuracy of 0.861 with 60% fewer practice episodes and 36% lower runtime than fixed-order practice. These results show that NL2GQL systems can improve by reusing execution-derived experience, while failure-directed practice makes experience acquisition more efficient.
%U https://aclanthology.org/2026.realm-1.58/
%P 846-855
Markdown (Informal)
[ExGQL: Self-Improving Natural Language-to-Graph Query Generation via Experience-Guided Learning](https://aclanthology.org/2026.realm-1.58/) (Guo et al., REALM 2026)
ACL