Hybrid IndoBERT–BiLSTM automated essay scoring architecture with per-dimension classifiers

Design · All designs
Evidence · 2 claims (1 for, 1 mixed) · 1 study (1 design), q2 · 0 of 1 report an effect size · 2 claims rest on one study

Description

The IB-BiLSTM architecture "combines IndoBERT as a contextual encoder with a two -layer Bidirectional LSTM (BiLSTM) as a sequential classifier." IndoBERT produces token-level embeddings of shape (batch × 128 × 768) fed to the BiLSTM, followed by Dense layers (256 and 128 units, ReLU) and a four-unit softmax for score levels 1–4. Six independent classification models were built, one per FRISCO dimension, rather than a single multi-label classifier, and all IndoBERT parameters were trainable for joint fine-tuning.

Design Implications

Context

Requirements

• Pretrained IndoBERT (indobert-base-p1) with WordPiece tokenization; balanced class-weighted loss; hyperparameter tuning via 5-fold CV on macro-F1

Constraints

• Evaluated on a small dataset (83 training / 23 test students); the article reports that accuracy alone is insufficient for ordinal AES evaluation because high-accuracy variants can show majority-class collapse

Target Learners

• Indonesian high school physics students

Learning Goals

• Automated multidimensional assessment of critical thinking in physics essays

Claims

• Aes Frisco Qwk Varies By Dimension [+M]
• Svm Outperforms Bilstm Reason Dimension [~M]

Related Designs

Examples

-

Key Sources

• Firdausi, H.; Wasis; Sholikah, R. W.; Lemantara, J.; Rosyadi, F. D.; Ginardi, R. V. H.; Pradityo, M. H. A.; Hakim, A. R. (2026). Educational Innovation through Automated Essay Scoring: A Multidimensional Framework for Evaluating Critical Thinking in High School Physics Essays. International Journal of Intelligent Engineering and Systems, Vol.19, No.8. https://doi.org/10.22266/ijies2026.0831.09

Page metadata

Field	Value
type	design
id	indobert-bilstm-hybrid-aes-architecture
title	Hybrid IndoBERT–BiLSTM automated essay scoring architecture with per-dimension classifiers
description	The IB-BiLSTM architecture "combines IndoBERT as a contextual encoder with a two -layer Bidirectional LSTM (BiLSTM) as a sequential classifier." IndoBERT produces token-level embeddings of shape (batch × 128 × 768) fe...
status	review
generated	by process:settle-candidates on 2026-10-10
trust	unverified — no one has confirmed this page's content
sources	1 — firdausi-2026

Frontmatter is the page's machine-readable record — see the Schema & Guide.