Human Language Technology. Challenges for Computer Science and Linguistics: 9th Language and Technology Conference, LTC 2019, Poznan, Poland, May ... (Lecture Notes in Artificial Intelligence)

This book constitutes the refereed proceedings of the 9th Language and Technology Conference: Challenges for Computer Science and Linguistics, LTC 2019, held in Poznan, Poland, in May 2019. The 24 revised papers presented in this volume were carefully reviewed and selected from 67 submissions. The papers are categorized into the following topical sub-headings: Speech Processing; Language Resources and Tools; Computational Semantics; Emotions, Decisions and Opinions; Digital Humanities; Evaluation; and Legal Aspects. Preface Organization Contents Speech Processing A Set of Tools for Extraction and Manipulation of Speech Fundamental Frequency 1 Introduction 2 Automatic Close Copy Speech Synthesis 3 F0Manipulation and Speech Resynthesis 3.1 Pitch Contour Replacement 3.2 Pitch Contour Shift Using Praat Built-in Function 4 F0Extraction 5 Extraction of Prosodic Information 6 Drawing Waveform, Pitch and Annotation for Two-Channel Sounds 7 Summary and Conclusion References The Automatic Search for Sounding Segments of SPPAS: Application to Cheese! Corpus 1 Introduction 2 The Method to Search Automatically for IPUs 2.1 Algorithm and Settings 2.2 Optional Settings 2.3 Discussion 3 Cheese! Corpus 3.1 Description 3.2 The IPUs of Cheese! 4 Evaluation Method 5 Results 5.1 Quantitative Evaluation 5.2 Qualitative Evaluation 6 Conclusion References The Phonetic Grounding of Prosody: Analysis and Visualisation Tools 1 Introduction 2 An Online Tool: CRAFT 2.1 Motivation 2.2 Amplitude Modulation (AM) and Frequency Modulation (FM) 2.3 Amplitude and Frequency Demodulation 2.4 Operations and Transforms 3 The RFA Offline Extension Toolkit 3.1 Functional Specification 3.2 Standalone Offline Tools 3.3 Prosodic Comparison of Narrative Readings 3.4 Distance Mapping 4 Conclusion References Hybridised Deep Ensemble Learning for Tone Pattern Recognition 1 Introduction 1.1 Lightly Supervised Learning 1.2 Ensemble Learning 1.3 Problem Statement 2 The Ibibio Language 2.1 Syllable Structure 2.2 Tone System 3 Related Works 4 Hybridised Classification Framework 4.1 Speech Corpus Database 4.2 DNN Modelling 4.3 DNN Modelling 5 Results 5.1 SOM Training 5.2 DNN Model Training 5.3 DNN Validation 6 Conclusion and Future Research Direction References ANNPRO: A Desktop Module for Automatic Segmentation and Transcription 1 Introduction 2 The Tools 2.1 Annotation Pro 2.2 CLARIN-PL Align 3 CLARIN-PL Align Module for Annotation Pro: ANNPRO 3.1 Module Integration 3.2 Using the ANNPRO Module 4 Example Use Case 5 Discussion and Conclusions References Language Resources and Tools Analysis and Processing of the Uzbek Language on the Multi-language Modelled Computer Translator Technology 1 Introduction 2 Related Work 3 Methodology 3.1 The General Logical-Linguistic Model of Formation of a Word 3.2 Logical-Linguistic Models of Words on Types 3.3 Logical-Linguistic Models of Sentences 4 Experiments and Results 5 Conclusion and Future Work References Computer Application of Georgian Words 1 Introduction 2 Brief Information About Georgian Language and Alphabet 3 Software Architecture 4 Software Sections 5 Web Version 6 Conclusion References Community-Led Documentation of Nafsan (Erakor, Vanuatu) 1 Introduction 2 Technical and Procedural Training 3 Outcomes of the Community-Led Documentation 4 Benefits and Challenges 4.1 Benefits: The Engagement of Community Researchers Improves the Results of Language Documentation 4.2 Challenges: Sustainability 4.3 Benefits Outweigh the Challenges 5 Increased Potential for Applications of Language Technology to Less-Resourced Languages 6 Conclusion References Design and Development of Pipeline of Preprocessing Tools for Kazakh Language Texts 1 Introduction 2 Morphological Model of the Kazakh Language 3 The Structure of Pipeline of Preprocessing Tools 4 The Tool of Generation of the Word Forms 5 The Tool of Lemmatization of the Word Forms 6 The Reduction of Homonymy 7 Pipeline Implementation 7.1 Word Generator 7.2 Tokenizer 7.3 Spelling Corrector 7.4 Morphological Analyzer 7.5 Dependency Parser 8 Conclusion References Thai Named Entity Corpus Annotation Scheme and Self Verification by BiLSTM-CNN-CRF Abstract 1 Introduction 2 Related Work 3 Corpus 3.1 Structure of the Corpus 3.2 Corpus Challenges 4 Corpus Revision Methodology 4.1 Approaches to Clean the Corpus 4.2 Named Entity Recognition Model 5 Experiment 5.1 Pre-processing 5.2 Experiment Setup 6 Result and Discussion 7 Combined Corpus 8 Conclusion Acknowledgements References Computational Semantics Analogies Between Short Sentences: A Semantico-Formal Approach 1 Goal of the Present Article 2 Semantic Analogies 2.1 Vector Arithmetic 2.2 Extension to Sets of Words 3 Formal Analogies 3.1 Traces 3.2 LCS Edit Distance and Parallel Traversal of Traces 4 Analogies Between Sentences 4.1 Distance Between Words and Sentences 4.2 Analogy-Compatible Decomposition of a Quadruple of Strings 4.3 Semantico-Formal Resolution of Analogies between Sentences 5 Extension: Computation of Analogies Between Two Words and a Bigram 5.1 Candidate Solution Where D Is One Word 5.2 Candidate Solution Where D Is a Bigram 5.3 Selection of the Best Candidate 6 Experiments and Released Datasets 6.1 Used Datasets 6.2 Released Datasets 6.3 Interesting Phenomena Observed 6.4 Difficulties of Assessment 7 Conclusion References Effective Development and Deployment of Domain and Application Conceptualization in Wordnet-Based Ontologies Using Ontology Repository Tool 1 Introduction 2 Development and Deployment of Information Languages 3 General, Domain and Application Conceptualization in Wordnet-Based Ontologies 3.1 General and Domain Conceptualizations in Wordnet-Based Ontologies 3.2 Application Conceptualizations in Wordnet-Based Ontologies 4 Ontology Repository Tool in Development and Deployment of Wordnet-Based Ontologies 4.1 Ontology Editor 4.2 Ontology Repository 5 Deployment of Wordnet-Based Ontologies 5.1 Development and Deployment of Wordnet-Based Ontology in Repository of e-learning Content 5.2 Development and Deployment of Wordnet-Based Ontology in Hatch System 6 Conclusions References Emotions, Decisions and Opinions Speech Prosody Extraction for Ibibio Emotions Analysis and Classification 1 Introduction 2 Ibibio Vowel, Consonant and Tone Systems 2.1 Vowel System 2.2 Consonant System 2.3 Tone System 3 Ibibio and Emotions 4 Methods 4.1 Classification Approach 4.2 Adopted Classifiers 5 Results 5.1 Acoustic-Emotion Correlates 5.2 Classification Results 6 Conclusion and Research Perspective References Multilingual and Language-Agnostic Recognition of Emotions, Valence and Arousal in Large-Scale Multi-domain Text Reviews 1 Introduction 2 Data Annotation 3 Data Transformation 3.1 Scores Distribution 4 Experiments 4.1 Word Embeddings 4.2 Multilingual and Language-Agnostic Text Representations 4.3 Evaluation Procedure 5 Monolingual Classification Results (Polish) 6 Multilingual and Language-Agnostic Regression Results 7 Conclusions 8 Further Work and Data Availability References Construction and Evaluation of Sentiment Datasets for Low-Resource Languages: The Case of Uzbek 1 Introduction 2 Related Work 3 Methodology 3.1 Data Collection 3.2 Pre-processing 3.3 Annotation 3.4 Translation 4 Experiments and Results 5 Conclusion and Future Work References Using Book Dialogues to Extract Emotions from Texts 1 Introduction 1.1 Sentiment Analysis and Emotion Detection 1.2 Related Work 1.3 Emotion Models: The Wheel and the Hourglass 1.4 Challenges in Analysis of Sentiment and Emotions 1.5 Research Goals 2 Methodology 2.1 Processing Pipeline Overview 2.2 Datasets 2.3 Emotion Representation 2.4 Lexicon-Based Classifiers 2.5 Neural Models 2.6 Evaluation Dataset 3 Evaluation Results 4 Conclusion References Digital Humanities NLP Tools for Lexical Structure Studies of the Literary Output of a Writer. Case Study: Literary Works of Tadeusz Boy-Żeleński and Julia Hartwig 1 Introduction 2 Linguistic Data 2.1 Tadeusz Boy-Żeleński Corpus 2.2 Julia Hartwig Corpus 3 AMU Tools Used in Studies of Literary Outcome of an Author 3.1 System POLEX 3.2 Text SubCorpora Creator (TSCC) 3.3 Named Entity Recognizer and Entity Linker 4 Other Tools 4.1 NLTK Libraries 4.2 Korpusomat 5 Applications to Study Literary Outcome of an Author 5.1 Tagging and Lemmatizing Texts 5.2 Use of NLTK Libraries for Obtaining Frequency Lists of Adjectival Forms and Lexemes for the Poetry and Prose of Julia Hartwig 5.3 Quantitative Studies of Adjectives in the Corpora of Tadeusz Boy-Żeleński and Julia Hartwig 5.4 Study of Saturation of Text Corpora with Adjectival Forms Using TSCC 5.5 Use of the Named Entity Recognition and Entity Linking Systems to Study Literary Output 6 Concluding Remarks Appendix References Neural Nets in Detecting Word Level Metaphors in Polish 1 Introduction 2 Related Work 3 Data Description 4 Experiment Description 4.1 LSTM Based Solution 4.2 BERT Based Solution 5 Results 6 Conclusion References Frame-Based Annotation in the Corpus of Synesthetic Metaphors 1 Introduction 2 Method 2.1 Synamet – A Polish Corpus of Synesthetic Metaphor 2.2 Frames in Synamet 2.3 Annotation Procedure 3 Results 3.1 Source and Target Frames in Strong Synesthetic Metaphors 3.2 Source and Target Frames in Weak Synesthetic Metaphors 3.3 Frame Activators 4 Discussion 5 Conclusion References Evaluation Evaluating Natural Language Processing tools for Polish during PolEval 2019 1 Introduction 2 Task 1: Recognition and Normalization of Temporal Expressions 2.1 Problem Statement 2.2 Task Description 2.3 Training Data 2.4 Evaluation 2.5 Results 3 Task 2: Lemmatization of Proper Names and Multi-word Phrases 3.1 Problem Statement 3.2 Task Description 3.3 Training Data 3.4 Evaluation 4 Task 3: Entity Linking 4.1 Problem Statement 4.2 Task Description 4.3 Training Data 4.4 Evaluation 5 Task 4: Machine Translation 5.1 Problem Statement 5.2 Task Description 5.3 Training Data 5.4 Evaluation and Results 6 Task 5: Automatic Speech Recognition 6.1 Problem Statement 6.2 Task Description 6.3 Training Data 6.4 Evaluation 6.5 Results 7 Task 6: Automatic Cyberbullying Detection 7.1 Problem Statement 7.2 Task Description 7.3 Training Data 7.4 Evaluation 7.5 Task 6: Results 8 Conclusions and Future Plans References Open Challenge for Correcting Errors of Speech Recognition Systems 1 Introduction 2 Related Work 2.1 Shared Tasks 2.2 ASR Error Correction Systems 3 Dataset 4 Evaluation 5 Baseline Models 5.1 Brill Tagger 5.2 Transformer Tagger 5.3 Flair Tagger 5.4 Baseline Results References Evaluation of Basic Modules for Isolated Spelling Error Correction in Polish Texts 1 Introduction 2 Problems of Spelling Correction for Polish 3 Methods 3.1 Baseline Methods 3.2 Vector Distance 3.3 Recurrent Neural Networks 4 Experimental Setup 5 Results 6 Conclusion References Assessment of Document Similarity Visualisation Methods 1 Introduction 2 Vectorization Methods 2.1 TF-IDF 2.2 fastText 2.3 ELMo 2.4 BERT 2.5 Method Summary 3 Reduction Methods 3.1 PCA 3.2 T-SNE 3.3 UMAP 4 Evaluation Methods 4.1 Closest Match (CM) 4.2 KNN 4.3 ARI 4.4 Internal Similarity (INT-SIM) 4.5 External Dissimilarity (EXT-DIS) 5 Experiments 5.1 Datasets 5.2 Results 6 Conclusion References Legal Aspects Legal Regime of the Language Resources in the Context of the European Language Technology Development 1 Introduction 2 Models for Language Resource Creation 2.1 The Usage of the LR Materials Protected by Copyright and Related Rights 2.2 The Usage of the LR Materials that Contain Personal Data 3 Criteria and Legal Mechanisms for Language Resource Database Protection 3.1 Language Resource Database Characteristics 3.2 Legal Mechanisms for Database Protection 4 Conclusion References Correction to: ANNPRO: A Desktop Module for Automatic Segmentation and Transcription Correction to: Chapter “ANNPRO: A Desktop Module for Automatic Segmentation and Transcription” in: Z. Vetulani et al. (Eds.): Human Language Technology, LNAI 13212, https://doi.org/10.1007/978-3-031-05328-3_5 Author Index