2026-07-01 04:00 UTCOriginal source2 min readUpdated: 2026-07-01 08:11 UTC

Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text

This study evaluates the robustness of Bangla event detection systems in real-world noisy conditions by introducing a benchmark of 9,979 annotated sentences across 40 event subtypes, including clean text, ASR transcripts, and orthographically corrupted text. Results show a trade-off: encoder models excel on clean text but degrade under noise, while decoder-only LLMs are more robust. Instruction tuning with annotation guidelines improves baseline performance on noise but not consistently. Model scaling boosts decoder robustness, and mixed training narrows the robustness gap, especially benefiting encoders.

SourcearXiv Computational LinguisticsAuthor: Tanvir Ahmed Sijan, S. M Golam Rifat, Nayeemul Islam, Md. Musfique Anwar

Article intelligence

EngineersAdvanced

Key points

Introduced a Bangla event detection benchmark with 9,979 sentences across 40 event subtypes, covering clean text, ASR transcripts, and corrupted text.
Encoder models perform better on clean text but degrade significantly under noise; decoder-only LLMs are more robust.
Embedding annotation guidelines in instruction tuning improves noise baseline but yields inconsistent degradation reductions.
Model scaling consistently improves decoder robustness, while mixed training benefits encoders and narrows the robustness gap.

Why it matters

This matters because introduced a Bangla event detection benchmark with 9,979 sentences across 40 event subtypes, covering clean text, ASR transcripts, and corrupted text.

Technical impact

May affect model selection, inference cost, product capability, and evaluation benchmarks.

This panel is AI-generated and reviewed for accuracy.

-->

[Submitted on 29 Jun 2026]

Title:Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text

View a PDF of the paper titled Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text, by Tanvir Ahmed Sijan and 3 other authors

View PDF HTML (experimental)

Abstract:Event detection (ED) systems are typically evaluated on clean, curated text, leaving their robustness to real-world noise largely unexplored, particularly for low-resource languages such as Bangla. We introduce a generalized Bangla news event ontology and a benchmark comprising 9,979 annotated sentences across 40 event subtypes, spanning clean news text, real-world Automatic Speech Recognition (ASR) transcripts, and orthographically corrupted text. We systematically evaluate fine-tuned encoder-only models (BanglaBERT and XLM-R) alongside instruction-tuned decoder-only large language models (Llama 3 and Gemma 3). Our results reveal a clear architectural trade-off: encoder models achieve higher performance on clean text but degrade substantially under noise, whereas decoder-only LLMs are markedly more robust, particularly when event triggers are corrupted. We further show that embedding annotation guidelines during instruction tuning establishes a higher performance baseline on noisy text but yields inconsistent reductions in performance degradation across noisy conditions. Finally, model scaling consistently improves the robustness of decoder-only LLMs, while combined training on clean and noisy data serves as an effective regularization strategy that disproportionately benefits encoder architectures, significantly narrowing the robustness gap.

Comments: 17 pages, 8 figures

Subjects:

Computation and Language (cs.CL)

Cite as: arXiv:2606.30914 [cs.CL]

(or arXiv:2606.30914v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2606.30914

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Tanvir Ahmed Sijan [view email] [v1] Mon, 29 Jun 2026 21:03:32 UTC (2,793 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text, by Tanvir Ahmed Sijan and 3 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CL

new | recent | 2026-06

Change to browse by:

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)