Cette thèse décrit les applications du traitement automatique des langues (TAL) à la gestion des risques industriels. Elle se concentre sur le domaine de l'aviation civile, où le retour d'expérience (REX) génère de grandes quantités de données, sous la forme de rapports d'accidents et d'incidents. Nous commençons par faire un panorama des différentes types de données générées dans ce secteur d'activité. Nous analysons les documents, comment ils sont produits, collectés, stockés et organisés ainsi que leurs utilisations. Nous montrons que le paradigme actuel de stockage et d’organisation est mal adapté à l’utilisation réelle de ces documents et identifions des domaines problématiques ou les technologies du langage constituent une partie de la solution. Répondant précisément aux besoins d'experts en sécurité, deux solutions initiales sont implémentées : la catégorisation automatique de documents afin d'aider le codage des rapports dans des taxonomies préexistantes et un outil pour l'exploration de collections de rapports, basé sur la similarité textuelle. En nous basant sur des observations de l'usage de ces outils et sur les retours de leurs utilisateurs, nous proposons différentes méthodes d'analyse des textes issus du REX et discutons des manières dont le TAL peut être appliqué dans le cadre de la gestion de la sécurité dans un secteur à haut risque. En déployant et évaluant certaines solutions, nous montrons que même des aspects subtils liés à la variation et à la multidimensionnalité du langage peuvent être traités en pratique afin de gérer la surabondance de données REX textuelles de manière ascendante / This thesis describes the applications of natural language processing (NLP) to industrial risk management. We focus on the domain of civil aviation, where incident reporting and accident investigations produce vast amounts of information, mostly in the form of textual accounts of abnormal events, and where efficient access to the information contained in the reports is required. We start by drawing a panorama of the different types of data produced in this particular domain. We analyse the documents themselves, how they are stored and organised as well as how they are used within the community. We show that the current storage and organisation paradigms are not well adapted to the data analysis requirements, and we identify the problematic areas, for which NLP technologies are part of the solution. Specifically addressing the needs of aviation safety professionals, two initial solutions are implemented: automatic classification for assisting in the coding of reports within existing taxonomies and a system based on textual similarity for exploring collections of reports. Based on the observation of real-world tool usage and on user feedback, we propose different methods and approaches for processing incident and accident reports and comprehensively discuss how NLP can be applied within the safety information processing framework of a high-risk sector. By deploying and evaluating certain approaches, we show how elusive aspects related to the variability and multidimensionality of language can be addressed in a practical manner and we propose bottom-up methods for managing the overabundance of textual feedback data / Тoзи реферат описва приложението на автоматичната обработка на естествен език (ОЕЕ) в контекста на управлението на риска в гражданското въздухоплаване. В тази област докладването на инциденти и разследването на произшествия генерират голямо количество информация, главно под формата на текстови описания на необичайни събития. На първо време описваме раличните типове (текстови) данни, които секторът произвежда. Анализираме самите документи, методите за съхраняването им, как са организирани, както и техните употреби от екперти по сигурността. Показваме, че съвремените парадигми за съхраняване и организация не са добре приспособени към реалната употреба на този тип данни и установяваме проблемните зони, в които ОЕЕ е част от решението. Две приложения, отговарящи прецизно на нуждите на експерти по авиационна сигурност, са имплементирани: автоматична класификация на доклади за инциденти и система за проучване на на колекции, основаваща се върху текстовото сходство. Въз основа на наблюдения на реалната употреба на приложенията, предлагаме няколко метода за обработка на доклади за инциденти и произшествия и обсъждаме в дълбочина как ОЕЕ може да бъде проложено на различни нива в информационнo-обработващите структури на един високорисков сектор. Оценявайки методите показваме, че трудностите свързани с многоизмерността и изменимостта на човешкия език могат да бъдат ефективно адресирани и предлагаме надеждни възходящи методи за справяне със свръхизобилието на доклади за инциденти в текстови формат

The domain of rhetoric has changed dramatically since its inception as the art of persuasion. It has adapted to encompass many forms of digital media, including, for example, data visualization and coding as a form of literature, but the approach has frequently been that of an outsider looking in. The use of comprehensive computational tools as a part of rhetorical analysis has largely been lacking. In this report, we attempt to address this lack by means of three case studies in natural language processing tasks, all of which can be used as part of a computational approach to rhetoric. At this same moment in time, it is becoming all the more important to transition to renewable energy in order to keep global warming under 1.5 degrees Celsius and ensure that countries meet the conditions of the Paris Agreement. Thus, we make use of speech data on climate policy from the Swedish parliament to ground these three analyses in semantic textual similarity, topic modeling, and political party attribution. We find that speeches are, to a certain extent, consistent within parties, given that a slight majority of most semantically similar speeches come from the same party. We also find that some of the most common topics discussed in these speeches are nuclear energy and the Swedish Green party, purported environmental risks due to renewable energy sources, and the job market. Finally, we find that though pairs of speeches are semantically similar, party rhetoric on the whole is generally not unique enough for speeches to be distinguishable by party. These results then open the door for a broader exploration of computational rhetoric for Swedish political science in the future.

More and more aspects of today’s healthcare are becoming integrated with medical technology and dependent on medical IT systems, which consequently puts stricter re-quirements on the companies delivering these solutions. As a result, companies delivering medical technology solutions need to spend a lot of resources maintaining high-quality, responsive customer support. In this report, possible ways of increasing customer support efficiency using machine learning and NLP is examined at Sectra, a medical technology company. This is done through a qualitative case study, where empirical data collection methods are used to elicit requirements and find ways of adding decision support. Next, a prototype is built featuring a ticket recommendation system powered by GPT-3 and based on 65 000 available support tickets, which is integrated with the customer supports workflow. Lastly, this is evaluated by having six end users test the prototype for five weeks, followed by a qualitative evaluation consisting of interviews, and a quantitative measurement of the user-perceivedusability of the proposed prototype. The results show some support that machine learning can be used to create decision support in a customer support context, as six out of six test users believed that their long-term efficiency could improve using the prototype in terms of reducing the average ticket resolution time. However, one out of the six test users expressed some skepticism towards the relevance of the recommendations generated by the system, indicating that improvements to the model must be made. The study also indicates that the use of state-of-the-art NLP models for semantic textual similarity can possibly outperform keyword searches.