Research map / Architectures and efficiency / RNNs and normalization
Working Memory-Driven Neural Networks with a Novel Knowledge Enhancement Paradigm for Implicit Discourse Relation Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add mutual attention matrix to BiLSTM (BMAN) (ablation) | BiLSTM · PDTB four-way | F1 (%): 36.54 -> 42.21 | Helped |
| add bidirectional encoding (LSTM to BiLSTM) (ablation) | LSTM · PDTB four-way | F1 (%): 36.41 -> 36.54 | No clear effect |
| add WordNet knowledge enhancement (implicit + explicit) to BMAN (ablation) | BMAN · PDTB four-way | F1 (%): 42.21 -> 47.90 | Helped |
| add knowledge enhancement to BMAN (ablation) | BMAN · PDTB four-way (pre-trained embeddings) | accuracy (%): 55.92 -> 57.25 | Helped |
| use pre-trained GloVe vs random embeddings (ablation) | KANN · PDTB four-way | accuracy (%): 55.35 -> 57.25 | Helped |
| add external knowledge, with random embeddings (ablation) | BMAN with pre-trained embeddings · PDTB four-way | accuracy (%): 54.80 -> 55.35 | Helped |
| KANN vs best prior on binary Comparison | Lei2018 · PDTB binary Comparison | F1 (%): 43.24 -> 43.92 | Helped |
| KANN vs best prior on binary Expansion | Lei2018 · PDTB binary Expansion | F1 (%): 72.88 -> 73.45 | Helped |
| KANN vs Lan2017 on binary Contingency | Lan2017 · PDTB binary Contingency | F1 (%): 58.96 -> 57.67 | Hurt |
| KANN vs Lan2017 on binary Temporal | Lan2017 · PDTB binary Temporal | F1 (%): 38.50 -> 36.33 | Hurt |
| KANN vs Lan2017 on four-way | Lan2017 · PDTB four-way | F1 (%): 47.80 -> 47.90 | No clear effect |
| use mean-pooling instead of max-pooling for importance vectors (ablation) | KANN · PDTB | performance | Helped |
Closest papers by meaning
- Linguistic Properties Matter for Implicit Discourse Relation Recognition: Combining Semantic Interaction, Topic Continuity and Attribution (AAAI 2018 · similarity 0.89)
- Implicit Discourse Relation Classification via Multi-Task Neural Networks (AAAI 2016 · similarity 0.89)
- Improving Implicit Discourse Relation Recognition with Natural Language Explanations from LLMs (AAAI 2026 · similarity 0.86)
- A Label Dependence-Aware Sequence Generation Model for Multi-Level Implicit Discourse Relation Recognition (AAAI 2022 · similarity 0.86)
- Learning to Memorize Entailment and Discourse Relations for Persona-Consistent Dialogues (AAAI 2023 · similarity 0.85)
- A Framework for Inference Inspired by Human Memory Mechanisms (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.