Named Entities for Computational Linguistics

One of the challenges brought on by the digital revolution of the recent decades is the mechanism by which information carried by texts can be extracted in order to access its contents. The processing of named entities remains a very active area of research, which plays a central role in natural language processing technologies and their applications. Named entity recognition, a tool used in information extraction tasks, focuses on recognizing small pieces of information in order to extract information on a larger scale. The authors use written text and examples in French and English to present the necessary elements for the readers to familiarize themselves with the main concepts related to named entities and to discover the problems associated with them, as well as the methods available in practice for solving these issues.

Autorentext
Damien Nouvel is Associate Professor at the National Institute of Oriental Languages And Civilizations (Inalto) in Paris, France. Maud Ehrmann is a Research Scientist at EPFL (École polytechnique fédérale de Lausanne) in Geneva, Switzerland. Sophie Rosset is a Senior Researcher at the French National Centre for Scientific Research (CNRS) in Paris, France.

Inhalt
Introduction ix

Chapter 1. Named Entities for Accessing Information 1

1.1. Research program history 2

1.1.1. Understanding documents: an ambitious task 2

1.1.2. Detecting basic elements: named entities 3

1.1.3. Trend: a return to slot filling 7

1.2. Task using named entities as a basic representation 9

1.3. Conclusion 10

Chapter 2. Named Entities, Referential Units 11

2.1. Issues with the named entity concept 12

2.1.1. A heterogeneous set 12

2.1.2. Existing defining formulas 17

2.1.3. An NLP object 21

2.2. The notions of meaning and reference 22

2.2.1. What is the reference? 22

2.2.2. What is meaning? 24

2.3. Proper names 27

2.3.1. The traditional criteria for defining a proper name 28

2.3.2. Meaning and referential function of proper names 30

2.3.3. The referential load of proper names 34

2.4. Definite descriptions 35

2.4.1. What is a definite description? 35

2.4.2. The meaning of definite descriptions 38

2.4.3. Complete and incomplete definite descriptions 39

2.5. The meaning and referential functioning of named entities 41

2.5.1. Reference to a particular 42

2.5.2. Referential autonomy 44

2.5.3. A natural heterogeneity 45

2.6. Conclusion 46

Chapter 3. Resources Associated with Named Entities 47

3.1. Typologies: general and specialist domains 48

3.1.1. The notion of category 48

3.1.2. Typology development 49

3.1.3. Typologies beyond evaluation campaigns 53

3.1.4. Other uses of typologies 54

3.1.5. Illustrated comparison 57

3.1.6. Issues to consider regarding entities 57

3.2. Corpora 59

3.2.1. Introduction . 59

3.2.2. Corpora and named entities 60

3.2.3. Conclusion 65

3.3. Lexicons and knowledge databases 65

3.3.1. Lexical databases 66

3.3.2. Knowledge databases 72

3.4. Conclusion 75

Chapter 4. Recognizing Named Entities 77

4.1. Detection and classification of named entities 78

4.2. Indicators for named entity recognition 79

4.2.1. Describing word morphology 79

4.2.2. Using lexical databases 81

4.2.3. Contextual clues 83

4.2.4. Conclusion 85

4.3. Rule-based techniques 85

4.4. Data-driven and machine-learning systems 88

4.4.1. Majority class models 91

4.4.2. Contextual models (HMM) 92

4.4.3. Multiple feature models (Softmax and MaxEnt) 93

4.4.4. Conditional Random Fields (CRFs) 95

4.5. Unsupervised enrichment of supervised methods 95

4.6. Conclusion 96

Chapter 5. Linking Named Entities to References 99

5.1. Knowledge bases 100

5.2. Formalizing polysemy in named entity mentions 102

5.3. Stages in the named entity linking process 103

5.3.1. Detecting mentions of named entities 103

5.3.2. Selecting candidates for each mention 103

5.3.3. Entity disambiguation 104

5.3.4. Entity linking 106

5.4. System performance 106

5.4.1. Practical application: DBpedia Spotlight 107

5.4.2. Future prospects 108

Chapter 6. Evaluating Named Entity Recognition 111

6.1. Classic measurements: precision, recall and F-measures 112

6.2. Measures using error counts 115

6.3. Evaluating associated tasks 120

6.3.1. Detecting entities and mentions 121

6.3.2. Entity detection and linking 122

6.4. Evaluating preprocessing technologies 126

6.5. Conclusion 128

Conclusion 131

Appendices 137

Appendix 1. Glossary 139

Appendix 2. Named Entities: Research Programs 141

Appendix 3. Summary of Available Corpora 147 Appendix 4. Annotation Formats 151</...