Single-Pass, Depth-Selective Reading for Multi-Aspect Sentiment Analysis (Official Site)

🎤 ACL Conference Release

🎓 Journal Extension

🎤 ACL Conference Release

This repository contains the official code for our ACL 2026 (main conference, long paper, oral presentation) paper—Single-Pass, Depth-Selective Reading for Multi-Aspect Sentiment Analysis.

Updated on August 1, 2026 (Journal-extension code added.)

Updated on June 18, 2026 (ACL paper uploaded.)

Released on April 13, 2026

ACL DABS

DABS is a single-pass inference framework for Aspect-Term Sentiment Analysis in multi-aspect sentences. It encodes each sentence once to construct a reusable, depth-ordered substrate, and then performs aspect-conditioned readout without re-encoding. The framework consists of:

• DORA, which constructs a shared depth substrate via a single encoder pass.
• ACBS, which performs aspect-conditioned token localization and budget-aware depth selection.

Experiments on four ATSA benchmarks show that DABS achieves competitive performance while reducing end-to-end computation by up to 60% in multi-aspect settings (M > 2).

Repository Layout

.
├── data/                           # ACL ATSA data restored from data.tar.gz
│   └── semeval/                    # English and multilingual SemEval data
├── figures/                        # ACL paper figures
├── scripts/                        # ACL training, inference, and analysis entry points
├── src/                            # ACL DABS/DORA/ACBS implementation
│   ├── config/
│   ├── core/
│   └── utils/
├── journal/                        # isolated journal-extension code
│   ├── configs/                    # ASTE, multilingual ASTE, and ASQP protocols
│   ├── dabs_structured/            # Full DORA--QCBS structured model
│   │   ├── aste/                   # ASTE data, batching, and exact decoding
│   │   └── asqp/                   # ASQP data, batching, and exact decoding
│   ├── data/README.md              # official data sources and setup commands
│   ├── train_aste.py
│   ├── evaluate_aste.py
│   ├── train_asqp.py
│   ├── evaluate_asqp.py
│   └── requirements.txt            # separate journal environment
├── outputs/                        # generated checkpoints (git-ignored)
├── results/                        # generated summaries (git-ignored)
├── README.md
└── requirements.txt                # ACL environment

ACL Requirements

Install dependencies with:

pip install -r requirements.txt

Notes:

• Because this project was developed on an RTX 5090 GPU, some packages were early locally built versions. If exact builds are unavailable, please use approximately matching versions.
• Do not use transformers 5.x for the ACL code. Use the separate journal environment described below for the extension.

ACL Unified Data Loading

This section applies to the ACL conference implementation. The journal extension uses independent ASTE/ASQP readers and data roots documented under Journal Data and Official Downloads. We provide a unified entry point for all SemEval datasets in src/core/data.py. It covers the main English SemEval ATSA benchmarks together with multilingual Restaurant-16 variants:

• 2: Laptop-14
• 3: Restaurant-14
• 4: Restaurant-15
• 5: Restaurant-16
• 6: Restaurant-16-FR
• 7: Restaurant-16-RU
• 8: Restaurant-16-ES
• 9: Restaurant-16-DU
• 10: Restaurant-16-TU

The repository expects the processed JSON files under:

• data/semeval/Laptop_14/
• data/semeval/Restaurant_14/
• data/semeval/Restaurant_15/
• data/semeval/Restaurant_16/
• data/semeval/Restaurant_16_FR/
• data/semeval/Restaurant_16_RU/
• data/semeval/Restaurant_16_ES/
• data/semeval/Restaurant_16_DU/
• data/semeval/Restaurant_16_TU/

Download Data from Google Drive

The data/ directory is distributed as a compressed archive instead of being maintained directly in the GitHub repository.

• Dataset archive (data.tar.gz): Google Drive

After downloading data.tar.gz, place it at the project root and extract it with:

tar -xzf data.tar.gz

This restores the expected data/ directory used by src/core/data.py.

Download Released Full-Model Checkpoints

The released full-model checkpoints are also provided as a compressed archive:

• Checkpoint archive (full_model.tar.gz): Google Drive

After downloading full_model.tar.gz, place it at the project root and extract it with:

tar -xzf full_model.tar.gz

This restores outputs/full_model/. You can then run batch inference over the released checkpoints with:

python scripts/run_full_model_inference.py --device cuda:0

Run ACL DABS

Single run:

DATASET_CHOICE=3 RANDOM_SEED=42 python scripts/train.py --dual-layer

Batch runs for the full model on the four benchmarks:

python scripts/run_full_model_batch.py --datasets 2 3 4 5 --seeds 42 123 456

Outputs are written to:

• outputs/full_model/...
• results/full_model_batch_<timestamp>/...

Reuse vs Non-Reuse Comparison

To compare standard aspect-wise evaluation against the reuse path on a given checkpoint, run:

python scripts/compare_reuse_non_reuse_eval.py outputs/full_model/Restaurant-14/seed_42 --dataset-choice 3

You can also specify a JSON output path if you want to save the comparison report:

python scripts/compare_reuse_non_reuse_eval.py \
  outputs/full_model/Restaurant-14/seed_42 \
  --dataset-choice 3 \
  --json results/reuse_vs_non_reuse_res16_seed42.json

ACL Evaluation Protocol

These benchmarks do not provide a standard development split. Following the protocol in the paper, the best checkpoint within the training budget is selected on the test split by macro-F1.

---

🎓 Journal Extension

The journal extension studies whether the same single-encoder, reusable DABS principle transfers from supplied-aspect ATSA to latent structured affective extraction. It supports English ASTE, multilingual ASTE, and ASQP through one Full DORA--QCBS implementation. The extension is isolated under journal/: it does not import or modify the ACL src/ and scripts/ implementation.

Journal Environment

Use a separate environment because the journal implementation has a newer software stack:

python -m venv journal/.venv
source journal/.venv/bin/activate
pip install -r journal/requirements.txt

Journal Data and Official Downloads

The journal datasets are not redistributed here. Obtain them from their upstream releases and place them under journal/data/:

Task	Datasets used here	Upstream release
English ASTE	14lap, 14res, 15res, 16res	SemEval-Triplet-data / ASTE-Data-V2-EMNLP2020
Polish ASTE	pl_hotels, fold 1	Polish-ASTE-Datasets
Catalan/Basque ASTE	ca, eu	ASTE-Transformer multilingual conversions
ASQP	rest15, rest16	ABSA-QUAD

Ready-to-run clone and copy commands, including the Polish fold mapping valid.txt to dev.txt, are provided in journal/data/README.md. Please follow the licences and terms of the respective upstream datasets.

Run the Journal Extension

English ASTE:

python -m journal.train_aste \
  --config journal/configs/aste_en.json \
  --dataset 16res \
  --seed 42

Multilingual ASTE uses the same trainer with mDeBERTa:

python -m journal.train_aste \
  --config journal/configs/aste_multilingual.json \
  --dataset ca \
  --seed 42

ASQP:

python -m journal.train_asqp \
  --config journal/configs/asqp.json \
  --dataset rest16 \
  --seed 42

See journal/README.md for evaluation commands, the five fixed seeds, output locations, and smoke checks.

Citation

If you find our code useful, feel free to ⭐ star this repository. If you use the ACL work in your research, please cite:

@inproceedings{xia2026single,
  title={Single-Pass, Depth-Selective Reading for Multi-Aspect Sentiment Analysis},
  author={Xia, Yan and Pan, Zhuangzhuang and Kamsin, Amirrudin and Chan, Chee Seng},
  booktitle={Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages={14638--14656},
  year={2026},
  doi={10.18653/v1/2026.acl-long.667},
  url={https://aclanthology.org/2026.acl-long.667/}
}