| Description |
1 online resource (145 p.) |
| Series |
Signals and Communication Technology |
|
Signals and communication technology.
|
| Contents |
Intro -- Preface -- Contents -- Probabilistic Linguistic Knowledge and Token-Level Text Augmentation -- 1 Introduction -- 2 Related Works -- 3 Augmentation Methods -- 3.1 Text Augmentation Techniques -- 3.2 N-Gram Language Model -- 4 Experimental Settings -- 4.1 Task and Data -- 4.2 Classification Models -- 4.3 Training Details -- 4.4 Augmentation Details -- 5 Main Experiments -- 5.1 Chinese: LCQMC -- 5.2 English: QQQD -- 5.3 Interim Summary -- 6 Supplementary Experiments -- 6.1 Comparison of Texts Augmented by REDA and REDANG -- 6.2 Effect of Transformer |
|
6.3 Effect of Single Augmentation Technique -- 7 Discussion and Conclusion -- Appendix -- A. Text Restoration Experiments -- B. Ablation Experiments on LCQMC -- References -- Scaling Up Paraphrase Generation Datasets with Machine Translation and Semantic Similarity Filtering -- 1 Introduction -- 2 Related Work -- 3 Dataset Creation -- 3.1 Translation and Semantic-Similarity-Based Filtering -- 4 Evaluation -- 4.1 Paraphrase Generation -- 4.2 Data Augmentation -- 5 Conclusion -- References -- Generative Byte-Level Models for Restoring Spaces, Punctuation, and Capitalization in Multiple Languages |
|
1 Introduction -- 2 Related Work -- 2.1 Space Restoration/Word Segmentation -- 2.2 Restoration of Capitalization and Punctuation -- 2.3 Restoration of Spaces, Capitalization, and Punctuation -- 2.4 Text Normalization/Diacritization with Byte-Level Transformers -- 3 Byte-Level Transformer Architecture -- 3.1 Encoder-Decoder Overview -- 3.2 Byte-Level Transformers -- 3.2.1 Self-Attention -- 3.3 ByT5 -- 4 Experiments -- 4.1 Languages -- 4.2 Datasets -- 4.3 Evaluation Metrics -- 4.4 Models -- 5 Fine-Tuned ByT5 Models -- 5.1 Architecture -- 5.2 Preparation of Training Data -- 5.3 Training |
|
5.4 Post-Processing of Outputs -- 5.4.1 Chunking -- 5.4.2 Unpredictability of Model Outputs -- 5.4.3 Character Matching Method -- 6 Results -- 6.1 Overall Performance -- 6.2 Mid-Token Capitalization and Punctuation -- 6.3 Precision vs. Recall -- 6.4 Effect of Model Size -- 6.5 Considerations for Non-English Languages -- 7 Conclusion and Future Work -- References -- Hierarchical Multi-task Learning with Articulatory Attributes for Cross-Lingual Phoneme Recognition -- 1 Introduction -- 2 Phoneme Recognition Architecture -- 2.1 Hybrid Transformer Acoustic Model |
|
2.2 Hierarchical Multi-task Classification of Phonemes and Attributes -- 3 Evaluation -- 3.1 Datasets -- 3.2 Training -- 4 Discussion -- 4.1 Common Voice -- 4.2 UCLA Phonetic Corpus -- 4.3 Combined Analysis -- 4.4 Phoneme Errors -- 5 Conclusion -- References -- Comparison of Error Correction and Extraction Approaches -- 1 Introduction -- 2 Prior Research -- 3 Benchmark Data -- 4 Approaches -- 5 Experimental Setup -- 6 Results -- 7 Conclusions and Further Work -- References -- Learning Affective Responses to Music from Social Media Discourse -- 1 Introduction -- 2 Related Work |
| Summary |
This book unveils the most advanced techniques and innovative applications in the natural language processing (NLP) field. It uncovers the secrets to enhancing language understanding, and presents practical solutions to different NLP tasks, as text augmentation, paraphrase generation, and restoring spaces and punctuation in multiple languages. It unlocks the potential of hierarchical multi-task learning for cross-lingual phoneme recognition, and allows readers to explore more real-world applications such as error correction, aggregating industrial security findings as well as predicting music emotion values from social media conversations. "Practical Solutions for Diverse Real-World NLP Applications" is the suitable guidebook for researchers, students, and practitioners as it paves the way for them by delivering invaluable insights and knowledge. This book: Delves into cutting-edge research topics in NLP, covering diverse subjects in the field; Provides practical implementations and methodologies for various topics, enabling readers to apply the techniques in their own research; Offers an in-depth exploration of each topic, providing comprehensive insights and understanding for readers seeking a deeper understanding of NLP techniques used in real-world applications |
| Notes |
Description based upon print version of record |
|
2.1 Acoustic Features |
|
Includes index |
|
Online resource; title from PDF title page (SpringerLink, viewed January 22, 2024) |
| Subject |
Natural language processing (Computer science)
|
| Form |
Electronic book
|
| Author |
Abbas, Mourad
|
| ISBN |
9783031442605 |
|
3031442601 |
|
