%{search_type} search results

• 1. Information retrieval using the Boolean model
• 2. The dictionary and postings lists
• 3. Tolerant retrieval
• 4. Index construction
• 5. Index compression
• 6. Scoring and term weighting
• 7. Vector space retrieval
• 8. Evaluation in information retrieval
• 9. Relevance feedback and query expansion
• 10. XML retrieval
• 11. Probabilistic information retrieval
• 12. Language models for information retrieval
• 13. Text classification and Naive Bayes
• 14. Vector space classification
• 15. Support vector machines and kernel functions
• 16. Flat clustering
• 17. Hierarchical clustering
• 18. Dimensionality reduction and latent semantic indexing
• 19. Web search basics
• 20. Web crawling and indexes
• 21. Link analysis.
• (source: Nielsen Book Data)

Class-tested and coherent, this textbook teaches classical and web information retrieval, including web search and the related areas of text classification and text clustering from basic concepts. It gives an up-to-date treatment of all aspects of the design and implementation of systems for gathering, indexing, and searching documents; methods for evaluating systems; and an introduction to the use of machine learning methods on text collections. All the important ideas are explained using examples and figures, making it perfect for introductory courses in information retrieval for advanced undergraduates and graduate students in computer science. Based on feedback from extensive classroom experience, the book has been carefully structured in order to make teaching more natural and effective. Slides and additional exercises (with solutions for lecturers) are also available through the book's supporting website to help course instructors prepare their lectures.
(source: Nielsen Book Data)

• 1. Information retrieval using the Boolean model
• 2. The dictionary and postings lists
• 3. Tolerant retrieval
• 4. Index construction
• 5. Index compression
• 6. Scoring and term weighting
• 7. Vector space retrieval
• 8. Evaluation in information retrieval
• 9. Relevance feedback and query expansion
• 10. XML retrieval
• 11. Probabilistic information retrieval
• 12. Language models for information retrieval
• 13. Text classification and Naive Bayes
• 14. Vector space classification
• 15. Support vector machines and kernel functions
• 16. Flat clustering
• 17. Hierarchical clustering
• 18. Dimensionality reduction and latent semantic indexing
• 19. Web search basics
• 20. Web crawling and indexes
• 21. Link analysis.
• (source: Nielsen Book Data)

Class-tested and coherent, this textbook teaches classical and web information retrieval, including web search and the related areas of text classification and text clustering from basic concepts. It gives an up-to-date treatment of all aspects of the design and implementation of systems for gathering, indexing, and searching documents; methods for evaluating systems; and an introduction to the use of machine learning methods on text collections. All the important ideas are explained using examples and figures, making it perfect for introductory courses in information retrieval for advanced undergraduates and graduate students in computer science. Based on feedback from extensive classroom experience, the book has been carefully structured in order to make teaching more natural and effective. Slides and additional exercises (with solutions for lecturers) are also available through the book's supporting website to help course instructors prepare their lectures.
(source: Nielsen Book Data)

An introduction to statistical natural language processing (NLP). The text contains the theory and algorithms needed for building NLP tools. Topics covered include: mathematical and linguistic foundations; statistical methods; collocation finding; word sense disambiguation; and probalistic parsing.
(source: Nielsen Book Data)

• 1. Cutting the ergativity pie
• 2. Inuit (West Greenlandic)
• 3. Dyirbal
• 4. Concluding discussion.
• (source: Nielsen Book Data)

This volume considers and examines some of the phenomena that have led languages to be considered 'ergative'. Languages considered 'ergative' have only been sparsely studied, and many fundamental questions in their analysis seem at best incompletely answered. This volume fills that void by focusing on some of the basic issues: when ergativity should be analysed as syntactic or morphological; whether languages can be divided into two classes of syntactically and morphologically ergative languages, and if so where the division should be drawn; and whether ergative arguments are always core roles or not. Christopher Manning's codification of syntactic approaches to dealing with ergative languages is based on a hypothesis he terms the 'Inverse Grammatical Relations hypothesis'. This hypothesis adopts a framework that decouples prominence at the levels of grammatical relations and argument structure. The result is two notions of subject - grammatical subject and argument structure subject and a uniform analysis of syntactically ergative and Philippine languages. These language groups, the syntactically ergative and Philippine languages, allow an inverse mapping in the prominence of the two highest terms between argument structure and grammatical relations. This volume combines good scholarship with innovative ideas into an important work that will appeal to a wide range of linguists and scholars.
(source: Nielsen Book Data)

• Preface
• 1. Introduction
• 2. Predicate composition
• 3. Romance complex predicates
• 4. Serial verb constructions
• 5. Conclusions and prospects
• A. Glue language semantics
• B. Previous proposals
• Bibliography.
• (source: Nielsen Book Data)

This book provides a simple but precise framework for describing complex predicates and related constructions, and applies it principally to the analysis of complex predicates in Romance, and certain serial verb constructions in Tariana and Miskitu. The authors argue for replacing the projection architecture of LFG with a notion of differential information spreading within a unified feature structure. Another important feature is the use of the conception of argument-structure in Chris Manning's Ergativity to facilitate the description of how complex predicates are assembled. In both of these aspects the result is a framework that preserves the descriptive parsimony of LFG while taking on key ideas from HPSG.
(source: Nielsen Book Data)

Humans, through deep and expressive conversations, have perfected the art of exchanging information about the world around them seamlessly. But even with the latest NLP methods, chatbots struggle in being informative. In this dissertation, I describe my work on building neural systems for informative conversations. First, I describe Chirpy Cardinal, our Alexa Prize 2020 Socialbot, that was deployed to tens of thousands of users across the US, and served as a test-bed for an initial system for informative conversations. While we used state-of-the-art models that improved over prior work, they fell short of expectations when deployed in the real-world setting. In particular, our system had two components: a retriever to find conversationally relevant passages from a large corpus (like Wikipedia) and a language generator to weave it into the dialogue with conversational-sounding utterances, and these two components were unable to cohesively work together. Second, inspired by linguistics literature on conversations, I analyze human-human informative conversations and identify various strategies for acknowledgement, presentation, transition and detail-selection. I also present a case study, where I improve acknowledgements by using conditional mutual information to select better chatbot utterances. Third, I explore the possibility of learning these strategies from data by jointly training a neural retriever and a neural generator such that they work together cohesively. To train them, we need to know which passages are relevant to the conversation, but the abundant conversational data available for training is not annotated for relevant passages! Our method, HINDSIGHT, uses a posterior retriever to find relevant passages during training. The posterior retriever is jointly trained alongside the original retriever and the generator using the evidence lower bound (ELBo). We find that HINDSIGHT has better inductive biases than existing methods - at inference, the retriever finds more relevant passages and the generator is more grounded in the retrieved passages, resulting in better end-to-end performance. Together, these projects provide a strong practical motivation, rich linguistic guidance and an effective training method for our aim of building neural systems to have deep and topically broad conversations

Neural language models (LMs) have become the workhorse of most natural language processing tasks and systems today. Yet, they are not perfect, and the two most important challenges in improving them further are (1) their lack of interpretability, and (2) their inability to generalize consistently, both in- and out-of-distribution. In this dissertation, I first describe my work on studying these LMs via black-box analysis, in order to understand how their predictions change in response to strategic changes in inputs. This makes model predictions more transparent by highlighting the features of the input that the model relies on. Then, I describe my work on Generalization through Memorization -- exploiting the notion of similarity between examples by using data saved in an external memory and retrieving nearest neighbors from it. This approach improves existing LM and machine translation models in terms of both in- and out-of-domain generalization, without any added training costs. Beyond improving generalization, memorization also makes model predictions more interpretable

Advances in Deep Learning have enabled more fluent and flexible Natural Language Generation (NLG). While these neural generative systems achieved early success in machine translation, they encounter problems — such as repetition, incoherence, and uncontrollability — when applied to more open-ended tasks such as abstractive summarization, story generation and chitchat dialogue. Furthermore, open-ended neural generative models tend to be evaluated by crowdworkers in carefully-controlled environments; it is less well-understood how they behave in realistic environments with real-life users. This thesis analyzes and improves neural generative systems performing several open-ended tasks; in the case of dialogue, the systems are evaluated in their full social context. First, for abstractive summarization, I present a pointer-generator model to improve copying accuracy and a coverage mechanism to reduce repetition in the generated summaries. Next, for chitchat dialogue, I present a large-scale detailed human evaluation to reveal the relationship between bot behaviors (such as repetition, specificity, staying on topic, and question-asking) and human quality judgments, and show that by controlling these bot behaviors, we can improve user experience. Third, for story generation, I characterize the effect of large-scale pretraining, and of the decoding algorithm, on several syntactic, semantic, structural, and stylistic aspects of the generated text. Lastly, I present a study of a neural generative chitchat model in deployment as part of the Alexa Prize, talking to real, intrinsically-motivated users. By analysing bot-user interactions, I identify the bot's main error types, and how they relate to user dissatisfaction. Furthermore, I demonstrate a semi-supervised method to learn from dissatisfaction and thus improve the dialogue system

Modern natural language processing (NLP) systems have achieved outstanding performance on benchmark datasets, in large part due to the stunning rise of deep learning. These research advances have led to great improvements in production systems for tasks like machine translation, speech recognition, and question answering. However, these NLP systems still often fail catastrophically when given inputs from different sources or inputs that have been adversarially perturbed. This lack of robustness exposes troubling gaps in current models' language understanding capabilities, and creates problems when NLP systems are deployed to real users. In this thesis, I will argue that many different aspects of the current deep learning paradigm for building NLP systems can be significantly improved to ensure greater robustness. In the first half of this thesis, I will build models that are robust to adversarially chosen perturbations. State-of-the-art models that achieve high average accuracy make surprising errors on inputs that have been slightly perturbed without altering meaning, for example by replacing words with synonyms or inserting typos. For a single sentence, there is a combinatorially large set of possible word-level perturbations, so guaranteeing correctness on all perturbations of an input requires new techniques that can reason about this combinatorial space. I will present two methods for building NLP systems that are provably robust to perturbations. First, certifiably robust training creates robust models by minimizing an upper bound on the loss that the worst possible perturbation can induce. Second, robust encodings enforce invariance to perturbations through a carefully constructed encoding layer that can be reused across different tasks and combined with any model architecture. Our improvements in robustness are dramatic: certifiably robust training improves accuracy on examples with adversarially chosen word substitutions from 10% to 75% on the IMDB sentiment analysis dataset, while robust encodings improve accuracy on examples with adversarially chosen typos from 7% to 71% on average across six text classification datasets from the GLUE benchmark. In the second half of the thesis, I will consider robustness failures that stem from the unrealistic narrowness of modern datasets. Datasets for tasks like question answering or paraphrase detection contain only a narrow slice of all valid inputs, so models trained on such datasets often learn to predict based on shallow heuristics. These heuristics generalize poorly to other similar, valid inputs. I will present methods for both constructing more challenging test data and collecting training data that aids generalization. For the task of question answering, I will use adversarially constructed distracting sentences to reveal weaknesses in systems that standard in-distribution test data fails to uncover. In our adversarial setting, the accuracy of sixteen contemporaneous models on the SQuAD dataset drops from an average of 75% F1 score to 36%; on a current state-of-the-art model, accuracy drops from 92% F1 score to 61%. For pairwise classification tasks, I will show that active learning with neural sentence embedding models collects training data that greatly improves generalization to test data with realistic label imbalance, compared to standard training datasets collected heuristically. On a realistically imbalanced version of the Quora Question Pairs paraphrase detection dataset, our method improves average precision from 2% to 32%. Overall, this thesis shows that state-of-the-art deep learning models have serious robustness defects, but also argues that by modifying different parts of the standard deep learning paradigm, we can make significant progress towards building robust NLP systems

In a world where almost everything seems to come with a tl; dr, how do we make effective use of the large amount of knowledge that surrounds us and is growing every day? This dissertation focuses on addressing this question for the growing amount of knowledge that is encoded in the form of text with the help of natural language processing (NLP) systems. At a high level, it attempts to tackle two distinct problems: how to enable NLP systems to handle our complex information needs by enabling them to perform complex reasoning, and how to communicate efficiently and ask useful questions in a conversation when the request cannot be stated completely in the form of a single question. This dissertation presents several distinct approaches to tackle these problems. As these approaches are designed to solve relatively complex reasoning problems on our behalf, it is important to build trust between the user and the system to make sure the system is not just arriving at the right answers, but also doing so for the right reasons. Therefore, all of the approaches presented in this dissertation are also aimed at making the NLP systems involved more explainable for human understanding, and sometimes more controllable in their behavior through the same mechanism for explanation. Specifically, I first present my work on making use of linguistic information to aid the extraction of knowledge bases from textual data. Here, linguistically-motivated techniques combined with neural networks results in a new state of the art on knowledge extraction from text, which enables robust complex reasoning with this knowledge. Then, I move on to describe how we can complement knowledge-based approaches to question answering by extending it into a schema-less text-based setting. Here, we collect one of the first large-scale datasets for open-domain text-based multi-hop question answering, and then I present a system that iteratively retrieves supporting documents from a large collection of text to answer these text-based complex questions. Finally, as we improve NLP systems' capability of performing complex reasoning to answer questions, I note that it is important that they also accommodate our information needs that are sometimes too complex or under-defined to express in a single complex question. To this end, I present how to train NLP systems to ask inquisitive questions to gather knowledge in the face of information asymmetry. This not only helps them gather important information to help us resolve our information needs, but also allows systems to reason about how we will gather information in an interaction, and present textual knowledge in a more efficient manner to reduce unnecessary confusion. By defining the informativeness of inquisitive questions and optimizing for information gathering, the resulting system generates curiosity-driven questions to help the system learn more about previously unknown knowledge in a conversation. By demonstrating and examining these systems, I also hope to show how designing NLP systems for explainability can help us attain various notions of efficiency necessitated by the need to process and present textual knowledge from large collections of text we wish to make use of

This thesis describes a simple approach to neural arc-factored dependency parsing, building on neural machine learning techniques that have gained considerable popularity in recent years. Dependency parsing is a way of identifying the latent syntactic and semantic relationships between words in a sentence, with solid foundations in linguistic theory that I describe in some detail. In this work, I introduce new classification techniques that extend the affine softmax classifier ubiquitous in machine learning that would otherwise be inappropriate for parsing. What's more, I demonstrate that the new biaffine classification techniques can be derived mathematically from the same principles that yield the affine softmax classifier. Related works either use an alternative to the proposed biaffine classifiers---based on feedforward neural attention---or else use an entirely different parsing algorithm---known as transition-based parsing---based on constituency parsing. In this work, I find evidence that the biaffine classifiers outperform the traditional attention-based classifiers, and that the arc-factored system outperforms transition-based parsers more broadly. I also demonstrate that the hyperparameter choices are optimal or near optimal, with significant deviations either leading to overfitting or underfitting. Consequently, any modifications to the architecture that yield better accuracy are unlikely to be due to simply compensating for poor hyperparameters. The basic system can be batched to parse large documents very quickly, and achieves accuracy comparable to state-of-the-art on the most popular English benchmark. However, the original system makes a few design choices that introduce complications for other languages, namely a reliance on whole word tokens and part-of-speech tags. To solve the first limitation, I have the system construct word representations from characters, so that the model can learn how morphology expressed through orthography reflects syntactic structure. To solve the second, I minimally adapt the architecture of the parser so it can be trained as a sequence labeler. A tagger that directly uses insights gleaned from the parser can be trained on any dependency treebank with gold part-of-speech tags. This approach achieved the highest performance at tagging and parsing on the 2017 CoNLL shared task on dependency parsing, inspiring most of the top-performing systems of the 2018 shared task. I also extend the system for multitask tagging, such that morphological features and language-specific part-of-speech tags are conditioned on the predicted coarse-grained universal tag. Finally, I modify the edge classifier to condition predictions directly on the relative location of words, so the system can more effectively leverage linearization and distance. Both of these make statistically significant improvements to accuracy. In order to accommodate dependency formalisms that don't adhere to strict tree structures, I minimally adapt the parser once more to produce arbitrary dependency graphs instead of dependency trees. I again ablate the system to explore how important the different hyperparameters and components of the system are, finding that while most of them do make a statistically significant difference, in general the differences are very small and the system is very robust. The work in this thesis not only contributes narrowly to the field of dependency parsing, but also more broadly provides tools for tasks with more complex dependencies than sequence labeling or classification.