Now 5582 visitors
Today:0 Yesterday:0
Total: 149559
216S 51P 5R
2026-10-22, Week 43 |
| Member Login |
| TACT Journal Homepage |
| Call for Paper |
| Committee |
| Find My Paper |
| Author Homepage |
| Statistics & History |
| Welcome Message |
| Paper Procedure |
| FAQ |
| Paper Archives |
| Outstanding Papers |
| Author Homepage |
| - Paper Procedure |
| - Journal Procedure |
| - Presentation Tips |
|
| Program & Proceedings |
| Program with Papers |
| Plenary Session |
| Tutorial Session |
|
| Presentation Platform |
| Hotel & Travel Info |
| About Korea |
| Accommodation |
| Transportation |
| VISA |
| Other Infomation |
|
| Photo Gallery |
| Scheduler Login |
| Seminar |
| Archives Login |
|
| Sponsors |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Save Q&A |
 |
* Edit or answer any Q&A by selecting Q&A number Hyper Link below + Write button (Save)
|
ICACT20260145 Question.1
Questioner: Lee19711936516@163.com 2026-02-09 오후 4:31:39 |
ICACT20260145 Answer.1
Answer by Auhor m11407505@mail.ntust.edu.tw 2026-02-09 오후 4:31:39
Chrome  Click!! |
| What is the core reason why metaheuristic algorithms (such as NSGA-II) showed a particularly significant improvement in the ROUGE-2 metric compared to LM based baseline models in the extraction summarization task of this study? Why do these algorithms perform better in capturing two word combination content? |
The superior performance of meta-heuristic algorithms on ROUGE-2 can be attributed to their global optimization nature. ROUGE-2 measures bigram overlap, which depends not only on the importance of individual sentences but also on whether related content units are jointly selected. LM-only baselines typically score sentences independently based on semantic similarity, without explicitly modeling inter-sentence interactions. In contrast, the proposed meta-heuristic framework optimizes sentence combinations under coverage and redundancy constraints, which naturally favors the preservation of adjacent content units. As a result, the algorithm is more effective at retaining meaningful bigram structures, leading to higher ROUGE-2 scores. |
ICACT20260145 Question.2
Questioner: Lee19711936516@163.com 2026-02-09 오후 4:32:07 |
ICACT20260145 Answer.2
Answer by Auhor m11407505@mail.ntust.edu.tw 2026-02-09 오후 4:32:07
Chrome Click!! |
| What are the key factors that enable the two-stage pipeline of "meta heuristic candidate generation+LM assisted rearrangement" proposed in the study to maintain performance advantages in cross domain generalization (such as from news corpus CNN/DailyMail to technology abstract corpus SciTLDR-AIC)? How does this framework adapt to the task characteristics of different corpora? |
The proposed framework exhibits strong cross-domain generalization because its optimization objectives are largely domain-agnostic. Instead of learning dataset-specific patterns, the method formulates extractive summarization as a multi-objective search problem that optimizes coverage, redundancy, and length constraints—criteria that remain consistent across different summarization domains. Moreover, the two-stage design explicitly separates structural candidate generation from semantic refinement. In Stage 1, NSGA-II explores sentence combinations based on global summary-level properties, which are insensitive to domain-specific language distributions. In Stage 2, pre-trained language models are used only as auxiliary semantic scorers to re-rank candidates, rather than as primary decision-makers. This modular design allows the framework to adapt to different datasets without retraining or feature redesign, explaining why stable performance is maintained across news and scientific summarization tasks. |
ICACT20260145 Question.3
Questioner: zhaozx@dsp.ac.cn 2026-02-09 오후 4:32:24 |
ICACT20260145 Answer.3
Answer by Auhor m11407505@mail.ntust.edu.tw 2026-02-09 오후 4:32:24
Chrome Click!! |
| Could you elaborate on the specific resource-limited scenarios (e.g., edge devices, real-time text summarization systems) that your work primarily targets, and what are the key computational/energy constraints in these scenarios that make large PLMs infeasible? |
This work targets resource-constrained summarization scenarios where both computational resources and domain-specific training data are limited. A representative example is edge-device applications such as e-book readers with built-in automatic summarization functions. In such products, documents span a wide range of topics and genres, making it impractical to fine-tune a separate language model for each domain. Moreover, e-book readers and similar edge devices typically rely on low-power CPUs and battery-based operation, where repeated PLM inference would introduce unacceptable latency and energy consumption.
In addition, many real-world applications involve emerging or specialized domains where annotated summarization data is scarce or unavailable. In these settings, PLM-based methods often suffer from degraded performance due to domain mismatch and the lack of fine-tuning data. In contrast, the proposed framework does not rely on domain-specific supervision. Instead, it formulates extractive summarization as a multi-objective optimization problem that focuses on structural properties such as coverage, redundancy control, and length constraints, which remain stable across domains. The two-stage design further enhances adaptability by separating sentence selection from semantic refinement: NSGA-II generates structurally valid candidate summaries, while pre-trained language models are used only as auxiliary semantic scorers. As a result, the framework maintains stable performance across different datasets and application scenarios, making it particularly suitable for low-resource, low-power, and data-scarce environments. |
| |
