The mathematical representation of semantics is a key issue for Natural Language Processing (NLP). A lot of research has been devoted to finding ways of representing the semantics of individual words in vector spaces. Distributional approaches—meaning distributed representations that exploit co-occurrence statistics of large corpora—have proved popular and successful across a number of tasks. However, natural language usually comes in structures beyond the word level, with meaning arising not only from the individual words but also the structure they are contained in at the phrasal or sentential level. Modelling the compositional process by which the meaning of an utterance arises from the meaning of its parts is an equally fundamental task of NLP. This dissertation explores methods for learning distributed semantic representations and models for composing these into representations for larger linguistic units. Our underlying hypothesis is that neural models are a suitable vehicle for learning semantically rich representations and that such representations in turn are suitable vehicles for solving important tasks in natural language processing. The contribution of this thesis is a thorough evaluation of our hypothesis, as part of which we introduce several new approaches to representation learning and compositional semantics, as well as multiple state-of-the-art models which apply distributed semantic representations to various tasks in NLP. Part I focuses on distributed representations and their application. In particular, in Chapter 3 we explore the semantic usefulness of distributed representations by evaluating their use in the task of semantic frame identification. Part II describes the transition from semantic representations for words to compositional semantics. Chapter 4 covers the relevant literature in this field. Following this, Chapter 5 investigates the role of syntax in semantic composition. For this, we discuss a series of neural network-based models and learning mechanisms, and demonstrate how syntactic information can be incorporated into semantic composition. This study allows us to establish the effectiveness of syntactic information as a guiding parameter for semantic composition, and answer questions about the link between syntax and semantics. Following these discoveries regarding the role of syntax, Chapter 6 investigates whether it is possible to further reduce the impact of monolingual surface forms and syntax when attempting to capture semantics. Asking how machines can best approximate human signals of semantics, we propose multilingual information as one method for grounding semantics, and develop an extension to the distributional hypothesis for multilingual representations. Finally, Part III summarizes our findings and discusses future work.

Lexical vagueness is a ubiquitous phenomenon in natural language. Most of previous works in natural language processing (NLP) consider lexical ambiguity as the main problem in natural language understanding rather than lexical vagueness. Lexical vagueness is usually considered as a solution rather than a problem in natural language understanding since precise information is usually failed to be provided in conversations. However, lexical vagueness is obviously an obstacle in human robot interaction (HRI) since the robots are expected to precisely understand their users' utterances in order to provide reliable services to their users. This research aims to develop novel lexical vagueness handling techniques to enable service robots to precisely understand their users' utterance so that they can provide the reliable services to their users. A novel integrated system to handle lexical vagueness is proposed in this research based on an in-depth understanding of lexical ambiguity and lexical vagueness including why they exist, how they are presented, what differences are in between them, and the mainstream techniques to handle lexical ambiguity and lexical vagueness. The integrated system consists of two blocks: the block of lexical ambiguity handling and the block of lexical vagueness handling. The block of lexical ambiguity handling first removes syntactic ambiguity and lexical ambiguity. The block of lexical vagueness handling is then used to model and remove lexical vagueness. Experimental results show that the robots endowed with the developed integrated system are able to understand their users' utterances. The reliable services to their users, therefore, can be provided by the robots.

Defending models against Natural Language Processing adversarial attacks is a challenge because of the discrete nature of the text dataset. However, given the variety of Natural Language Processing applications, it is important to make text processing models more robust and secure. This paper aims to develop techniques that will help text processing models such as BERT to combat adversarial samples that contain misspellings. These developed models are more robust than off the shelf spelling checkers.

<div>Media polarization is a serious issue that can affect someone's views, ranging from a scientific fact to the perceived results of a presidential election. The media outlets in the United States are aligned along political spectrum representing different stances on various issues. Without providing any false information (but usually by omitting some facts), media outlets can report events by deliberately using the words and styles that favor particular political positions. <br></div>This research investigated the U.S. media polarization with authorship attribution approaches, analyzing stylistic differences between the left-leaning and right-leaning media and discovering specific linguistic patterns that made the news articles display biased political attitudes. Several models of authorship attribution were tested while controlling for topic, stance, and style, and were applied to media companies and their identity within a political spectrum. Style features that were compared included semantic and/or sentiment-related information, such as stance taking, with features that seemingly do not capture it, such as part of speech tags. The results demonstrate that a successful classification of articles as left-leaning or right-learning is possible regardless of their stance. Finally, we provide an analysis of the patterns that we found.

<p>When people write a review about a business, they write and rate it based on their personal experience of the business. Sentiment analysis is a natural language processing technique that determines the sentiment of text, including reviews. However, unlike computers, the personal experience of humans emphasizes their preferences and observations that they deem important while ignoring other components that may not be as important to them personally. Traditional sentiment analysis does not consider such preferences. To utilize these human preferences in sentiment analysis, this paper explores various methods of weighting aspects in an attempt to improve sentiment analysis accuracy. Two types of methods are considered. The first method applies human preference by assigning weights to aspects in calculating overall sentiment analysis. The second method uses the results of the first method to improve the accuracy of traditional supervised sentiment analysis. The results show that the methods have high accuracy when people have strong opinions, but the weights of the aspects do not significantly improve the accuracy.</p>

Measuring semantic similarity is a task needed in many Natural Language Processing (NLP) applications. For example, in Machine Translation evaluation, semantic similarity is used to assess the quality of the machine translation output by measuring the degree of equivalence between a reference translation and the machine translation output. The problem of semantic similarity (Corley and Mihalcea, 2005) is de ned as measuring and recognising semantic relations between two texts. Semantic similarity covers di erent types of semantic relations, mainly bidirectional and directional. This thesis proposes new methods to address the limitations of existing work on both types of semantic relations. Recognising Textual Entailment (RTE) is a directional relation where a text T entails the hypothesis H (entailment pair) if the meaning of H can be inferred from the meaning of T (Dagan and Glickman, 2005; Dagan et al., 2013). Most of the RTE methods rely on machine learning algorithms. de Marne e et al. (2006) propose a multi-stage architecture where a rst stage determines an alignment between the T-H pairs to be followed by an entailment decision stage. A limitation of such approaches is that instead of recognising a non-entailment, an alignment that ts an optimisation criterion will be returned, but the alignment by itself is a poor predictor for iii non-entailment. We propose an RTE method following a multi-stage architecture, where both stages are based on semantic representations. Furthermore, instead of using simple similarity metrics to predict the entailment decision, we use a Markov Logic Network (MLN). The MLN is based on rich relational features extracted from the output of the predicate-argument alignment structures between T-H pairs. This MLN learns to reward pairs with similar predicates and similar arguments, and penalise pairs otherwise. The proposed methods show promising results. A source of errors was found to be the alignment step, which has low coverage. However, we show that when an alignment is found, the relational features improve the nal entailment decision. The task of Semantic Textual Similarity (STS) (Agirre et al., 2012) is de- ned as measuring the degree of bidirectional semantic equivalence between a pair of texts. The STS evaluation campaigns use datasets that consist of pairs of texts from NLP tasks such as Paraphrasing and Machine Translation evaluation. Methods for STS are commonly based on computing similarity metrics between the pair of sentences, where the similarity scores are used as features to train regression algorithms. Existing methods for STS achieve high performances over certain tasks, but poor results over others, particularly on unknown (surprise) tasks. Our solution to alleviate this unbalanced performances is to model STS in the context of Multi-task Learning using Gaussian Processes (MTL-GP) ( Alvarez et al., 2012) and state-of-the-art iv STS features ( Sari c et al., 2012). We show that the MTL-GP outperforms previous work on the same datasets.

When computers generate text, they have to consider how to describe the entities mentioned in the text. This situation becomes more difficult when the audience is unknown, as it is not clear what information is available to the addressees. This thesis investigates generation of descriptions in situations when an algorithm does not have a precise model of addressee's knowledge. This thesis starts with the collection and analysis of a corpus of descriptions of famous people. The analysis of the corpus revealed a number of useful patterns, which informed the remainder of this thesis. One of the difficult questions is how to choose information that helps addressees identify the described person. This thesis introduces a corpus-based method for determining which properties are more likely to be known by the addressees, and a probability-based method to identify properties that are distinguishing. One of the patterns observed in the collected corpus is the inclusion of multiple properties each of which uniquely identifies the referent. This thesis introduces a novel corpus-based method for determining how many properties to include in a description. Finally, a number of algorithms that leverage the findings of the corpus analysis and their computational implementation are proposed and tested in an evaluation involving human participants. The proposed algorithms outperformed the Incremental Algorithm in terms of numbers of correctly identified referents and in terms of providing a better mental image of the referent. The main contributions of this thesis are: (1) a corpus-based analysis of descriptions produced for an unknown audience; (2) a computational heuristic for estimating what information is likely to be known to addressees; and (3) algorithms that can generate referring expressions that benefit addressees without having an explicit model of addressee's knowledge.

The question of whether a machine can reproduce human intelligence is older than modern computation, but has received a great deal of attention since the first digital computers emerged decades ago. Language understanding, a hallmark of human intelligence, has been the focus of a great deal of work in Artificial Intelligence (AI). In 1950, mathematician Alan Turing proposed a kind of game, or test, to evaluate the intelligence of a machine by assessing its ability to understand written natural language. But nearly sixty years after Turing proposed his test of machine intelligence—pose questions to a machine and a person without seeing either, and try to determine which is the machine—no system has passed the Turing Test, and the question of whether a machine can understand natural language cannot yet be answered. The present investigation is, firstly, an attempt to advance the state of the art in natural language understanding by building a machine whose input is English natural language and whose output is a set of assertions that represent answers to certain questions posed about the content of the input. The machine we explore here, in other words, should pass a simplified version of the Turing Test and by doing so help clarify and expand on our understanding of the machine intelligence. Toward this goal, we explore a constraint framework for partial solutions to the Turing Test, propose a problem whose solution would constitute a significant advance in natural language processing, and design and implement a system adequate for addressing the problem proposed. The fully implemented system finds primary specific events and their locations in monologue discourse using a hierarchical classification approach, and as such provides answers to questions of central importance in the interpretation of discourse. / text

This thesis explores approaches to modelling individual differences in language use. The differences under consideration fall into two broad categories: Variation of the personality projected through language, and modelling of language alignment behaviour between dialogue partners. In a way, these two aspects oppose each other – language related to varying personalities should be recognisably different, while aligning speakers agree on common language during a dialogue. The central hypothesis is that such variation can be captured and produced with restricted computational means. Results from research on personality psychology and psycholinguistics are transformed into a series of lexically-based Affective Language Production Models (ALPMs) which are parameterisable for personality and alignment. The models are then explored by varying the parameters and observing the language they generate. ALPM-1 and ALPM-2 re-generate dialogues from existing utterances which are ranked and filtered according to manually selected linguistic and psycholinguistic features that were found to be related to personality. ALPM-3 is based on true overgeneration of paraphrases from semantic representations using the OPENCCG framework for Combinatory Categorial Grammar (CCG), in combination with corpus-based ranking and filtering by way of n-gram language models. Personality effects are achieved through language models built from the language of speakers of known personality. In ALPM-4, alignment is captured via a cache language model that remembers the previous utterance and thus influences the choice of the next. This model provides a unified treatment of personality and alignment processes in dialogue. In order to evaluate the ALPMs, dialogues between computer characters were generated and presented to human judges who were asked to assess the characters’ personality. In further internal simulations, cache language models were used to reproduce results of psycholinguistic priming studies. The experiments showed that the models are capable of producing natural language dialogue which exhibits human-like personality and alignment effects.

This thesis describes the development and evaluation of new automated Question Answering (QA) methods tailored to clinical comparison questions that give clinicians a rank-ordered list of MEDLINE® abstracts targeted to natural language clinical drug comparison questions (e.g. ”Have any studies directly compared the effects of Pioglitazone and Rosiglitazone on the liver?”). Three corpora were created to develop and evaluate a new QA system for clinical comparison questions called RetroRank. RetroRank takes the clinician’s plain text question as input, processes it and outputs a rank-ordered list of potential answer candidates, i.e. MEDLINE® abstracts, that is reordered using new post-retrieval ranking strategies to ensure the most topically-relevant abstracts are displayed as high in the result set as possible. RetroRank achieves a significant improvement over the PubMed recency baseline and performs equal to or better than previous approaches to post-retrieval ranking relying on query frames and annotated data such as the approach by Demner-Fushman and Lin (2007). The performance of RetroRank shows that it is possible to successfully use natural language input and a fully automated approach to obtain answers to clinical drug comparison questions. This thesis also introduces two new evaluation corpora of clinical comparison questions with “gold standard” references that are freely available and are a valuable resource for future research in medical QA.