Many NLP tasks have at their core a subtask of extracting the dependencies—who did what to whom—from natural language sentences. This task can be understood as the inverse of the problem solved in different ways by diverse human languages, namely, how to indicate the relationship between different parts of a sentence. Understanding how languages solve the problem can be extremely useful in both feature design and error analysis in the application of machine learning to NLP. Likewise, understanding cross-linguistic variation can be important for the design of MT systems and other...

Empirical methods are means to answering methodological questions of empirical sciences by statistical techniques. The methodological questions addressed in this book include the problems of validity, reliability, and significance. In the case of machine learning, these correspond to the questions of whether a model predicts what it purports to predict, whether a model's performance is consistent across replications, and whether a performance difference between two models is due to chance, respectively. The goal of this book is to answer these questions by concrete statistical tests that can...

The goal of text ranking is to generate an ordered list of texts retrieved from a corpus in response to a query. Although the most common formulation of text ranking is search, instances of the task can also be found in many natural language processing (NLP) applications.This book provides an overview of text ranking with neural network architectures known as transformers, of which BERT (Bidirectional Encoder Representations from Transformers) is the best-known example. The combination of transformers and self-supervised pretraining has been responsible for a paradigm shift in NLP,...

Meaning is a fundamental concept in Natural Language Processing (NLP), in the tasks of both Natural Language Understanding (NLU) and Natural Language Generation (NLG). This is because the aims of these fields are to build systems that understand what people mean when they speak or write, and that can produce linguistic strings that successfully express to people the intended content. In order for NLP to scale beyond partial, task-specific solutions, researchers in these fields must be informed by what is known about how humans use language to express and understand communicative intents. The...

Opportunity and Curiosity find similar rocks on Mars. One can generally understand this statement if one knows that Opportunity and Curiosity are instances of the class of Mars rovers, and recognizes that, as signalled by the word on, rocks are located on Mars. Two mental operations contribute to understanding: recognize how entities/concepts mentioned in a text interact and recall already known facts (which often themselves consist of relations between entities/concepts). Concept interactions one identifies in the text can be added to the repository of known facts, and aid the processing of...

Many applications within natural language processing involve performing text-to-text transformations, i.e., given a text in natural language as input, systems are required to produce a version of this text (e.g., a translation), also in natural language, as output. Automatically evaluating the output of such systems is an important component in developing text-to-text applications. Two approaches have been proposed for this problem: (i) to compare the system outputs against one or more reference outputs using string matching-based evaluation metrics and (ii) to build models based on human...

Labelling data is one of the most fundamental activities in science, and has underpinned practice, particularly in medicine, for decades, as well as research in corpus linguistics since at least the development of the Brown corpus. With the shift towards Machine Learning in Artificial Intelligence (AI), the creation of datasets to be used for training and evaluating AI systems, also known in AI as corpora, has become a central activity in the field as well. Early AI datasets were created on an ad-hoc basis to tackle specific problems. As larger and more reusable datasets were created,...

This book explores the cognitive plausibility of computational language models and why it’s an important factor in their development and evaluation. The authors present the idea that more can be learned about cognitive plausibility of computational language models by linking signals of cognitive processing load in humans to interpretability methods that allow for exploration of the hidden mechanisms of neural models. The book identifies limitations when applying the existing methodology for representational analyses to contextualized settings and critiques the current emphasis on form over...