@inproceedings{watson-etal-2026-frame2kg,
title = "{F}rame2{KG}: A Benchmark and Evaluation Toolkit for Interpretable Frame-to-Graph Generation",
author = "Watson, Lewis N. and
Strathearn, Carl and
Mitchell, Kenny and
Yu, Yanchao",
editor = "Piperidis, Stelios and
Bel, N{\'u}ria and
van den Heuvel, Henk and
Ide, Nancy and
Krek, Simon and
Toral, Antonio",
booktitle = "Proceedings of the Fifteenth Language Resources and Evaluation Conference",
month = may,
year = "2026",
address = "Palma de Mallorca, Spain",
publisher = "ELRA Language Resource Association",
url = "https://aclanthology.org/2026.lrec-1.854/",
doi = "10.63317/4ys6kofrzoc5",
pages = "10912--10926",
abstract = "Interpretable frame-to-knowledge-graph (Frame2KG) generation enables structured visual scene representation while supporting on-device inference to enhance privacy, improve interpretability, and minimise compute. We introduce Frame2KG-YC2, a synthetic, reproducible dataset derived from YouCook2 that pairs keyframes with schema-valid JSON knowledge graphs containing typed, spatially grounded entities and semantic predicates, alongside faithful textual paraphrases. Using this corpus, we fine-tune Qwen2.5-VL models (3B and 7B) with parameter-efficient LoRA adapters on attention layers (QKVO), with and without GateProj/Up/Down MLP projections. For evaluation and benchmarking, we propose a deterministic toolkit featuring two-stage node matching, an IoU gate followed by Hungarian assignment on blended spatial-semantic similarity, and comprehensive metrics spanning node/edge precision-recall-F1, matched-pair IoU, and structural validity. On a held-out test set, our models achieve Node F1{\ensuremath{\mu}} up to 0.621 and Edge F1{\ensuremath{\mu}} up to 0.208, with mean matched IoU of {\ensuremath{\approx}}0.61 and {\ensuremath{>}}98{\%} schema conformity. We show that MLP gating consistently improves predicate accuracy and spatial grounding, while post-training quantisation maintains accuracy and improves deployability on edge hardware. We release the dataset, code, adapters, and evaluation toolkit to establish an open, interpretable baseline for future temporal and multi-view extensions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="watson-etal-2026-frame2kg">
<titleInfo>
<title>Frame2KG: A Benchmark and Evaluation Toolkit for Interpretable Frame-to-Graph Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lewis</namePart>
<namePart type="given">N</namePart>
<namePart type="family">Watson</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carl</namePart>
<namePart type="family">Strathearn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kenny</namePart>
<namePart type="family">Mitchell</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanchao</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifteenth Language Resources and Evaluation Conference</title>
</titleInfo>
<name type="personal">
<namePart type="given">Stelios</namePart>
<namePart type="family">Piperidis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Núria</namePart>
<namePart type="family">Bel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Henk</namePart>
<namePart type="family">van den Heuvel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nancy</namePart>
<namePart type="family">Ide</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Simon</namePart>
<namePart type="family">Krek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Antonio</namePart>
<namePart type="family">Toral</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>ELRA Language Resource Association</publisher>
<place>
<placeTerm type="text">Palma de Mallorca, Spain</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Interpretable frame-to-knowledge-graph (Frame2KG) generation enables structured visual scene representation while supporting on-device inference to enhance privacy, improve interpretability, and minimise compute. We introduce Frame2KG-YC2, a synthetic, reproducible dataset derived from YouCook2 that pairs keyframes with schema-valid JSON knowledge graphs containing typed, spatially grounded entities and semantic predicates, alongside faithful textual paraphrases. Using this corpus, we fine-tune Qwen2.5-VL models (3B and 7B) with parameter-efficient LoRA adapters on attention layers (QKVO), with and without GateProj/Up/Down MLP projections. For evaluation and benchmarking, we propose a deterministic toolkit featuring two-stage node matching, an IoU gate followed by Hungarian assignment on blended spatial-semantic similarity, and comprehensive metrics spanning node/edge precision-recall-F1, matched-pair IoU, and structural validity. On a held-out test set, our models achieve Node F1\ensuremathμ up to 0.621 and Edge F1\ensuremathμ up to 0.208, with mean matched IoU of \ensuremath\approx0.61 and \ensuremath>98% schema conformity. We show that MLP gating consistently improves predicate accuracy and spatial grounding, while post-training quantisation maintains accuracy and improves deployability on edge hardware. We release the dataset, code, adapters, and evaluation toolkit to establish an open, interpretable baseline for future temporal and multi-view extensions.</abstract>
<identifier type="citekey">watson-etal-2026-frame2kg</identifier>
<identifier type="doi">10.63317/4ys6kofrzoc5</identifier>
<location>
<url>https://aclanthology.org/2026.lrec-1.854/</url>
</location>
<part>
<date>2026-05</date>
<extent unit="page">
<start>10912</start>
<end>10926</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Frame2KG: A Benchmark and Evaluation Toolkit for Interpretable Frame-to-Graph Generation
%A Watson, Lewis N.
%A Strathearn, Carl
%A Mitchell, Kenny
%A Yu, Yanchao
%Y Piperidis, Stelios
%Y Bel, Núria
%Y van den Heuvel, Henk
%Y Ide, Nancy
%Y Krek, Simon
%Y Toral, Antonio
%S Proceedings of the Fifteenth Language Resources and Evaluation Conference
%D 2026
%8 May
%I ELRA Language Resource Association
%C Palma de Mallorca, Spain
%F watson-etal-2026-frame2kg
%X Interpretable frame-to-knowledge-graph (Frame2KG) generation enables structured visual scene representation while supporting on-device inference to enhance privacy, improve interpretability, and minimise compute. We introduce Frame2KG-YC2, a synthetic, reproducible dataset derived from YouCook2 that pairs keyframes with schema-valid JSON knowledge graphs containing typed, spatially grounded entities and semantic predicates, alongside faithful textual paraphrases. Using this corpus, we fine-tune Qwen2.5-VL models (3B and 7B) with parameter-efficient LoRA adapters on attention layers (QKVO), with and without GateProj/Up/Down MLP projections. For evaluation and benchmarking, we propose a deterministic toolkit featuring two-stage node matching, an IoU gate followed by Hungarian assignment on blended spatial-semantic similarity, and comprehensive metrics spanning node/edge precision-recall-F1, matched-pair IoU, and structural validity. On a held-out test set, our models achieve Node F1\ensuremathμ up to 0.621 and Edge F1\ensuremathμ up to 0.208, with mean matched IoU of \ensuremath\approx0.61 and \ensuremath>98% schema conformity. We show that MLP gating consistently improves predicate accuracy and spatial grounding, while post-training quantisation maintains accuracy and improves deployability on edge hardware. We release the dataset, code, adapters, and evaluation toolkit to establish an open, interpretable baseline for future temporal and multi-view extensions.
%R 10.63317/4ys6kofrzoc5
%U https://aclanthology.org/2026.lrec-1.854/
%U https://doi.org/10.63317/4ys6kofrzoc5
%P 10912-10926
Markdown (Informal)
[Frame2KG: A Benchmark and Evaluation Toolkit for Interpretable Frame-to-Graph Generation](https://aclanthology.org/2026.lrec-1.854/) (Watson et al., LREC 2026)
ACL