In recent years, knowledge graphs(KGs) have become essential tools for visualizing concepts and retrieving contextual information. However, constructing KGs for new and specialized domains like Positive Energy Districts (PEDs) presents unique challenges, particularly when dealing with unstructured texts and ambiguous concepts from academic articles. This study focuses on various strategies for constructing and inferring KGs, specifically incorporating entities related to PEDs, such as projects, technologies, organizations, and locations. We utilize visualization techniques and node embedding methods to explore the graph's structure and content and apply filtering techniques and t-SNE plots to extract subgraphs based on specific categories or keywords. One of the key contributions is using the longest path method, which allows us to uncover intricate relationships, interconnectedness between entities, critical paths, and hidden patterns within the graph, providing valuable insights into the most significant connections. Additionally, community detection techniques were employed to identify distinct communities within the graph, providing further understanding of the structural organization and clusters of interconnected nodes with shared themes. The paper also presents a detailed evaluation of a question-answering system based on the KG, where the Universal Sentence Encoder was used to convert text into dense vector representations and calculate cosine similarity to find similar sentences. We assess the system's performance through precision and recall analysis and conduct statistical comparisons of graph embeddings, with Node2Vec outperforming DeepWalk in capturing similarities and connections. For edge prediction, logistic regression, focusing on pairs of neighbours that lack a direct connection, was employed to effectively identify potential connections among nodes within the graph. Additionally, probabilistic edge predictions, threshold analysis, and the significance of individual nodes were discussed. Lastly, the advantages and limitations of using existing KGs(Wikidata and DBpedia) versus constructing new ones specifically for PEDs were investigated. It is evident that further research and data enrichment is necessary to address the scarcity of domain-specific information from existing sources.

Submitted by JOSIANE SANTOS DE OLIVEIRA (josianeso) on 2018-10-10T12:33:16Z No. of bitstreams: 1 Paola Gabriela Konrad_.pdf: 4247741 bytes, checksum: 078c837af4cb5360b2b5636814fdd828 (MD5) / Made available in DSpace on 2018-10-10T12:33:16Z (GMT). No. of bitstreams: 1 Paola Gabriela Konrad_.pdf: 4247741 bytes, checksum: 078c837af4cb5360b2b5636814fdd828 (MD5) Previous issue date: 2018-08-28 / UNISINOS - Universidade do Vale do Rio dos Sinos / A prática de perguntar e de responder consiste em uma atividade trivial na vida das pessoas, seja na fala-em-interação mundana ou institucional. Nesta dissertação, analisam-se sequências de perguntas e respostas, bem como as consequências por elas ocasionadas, em um evento interacional de caráter institucional permeado pela prática de perguntar e de responder: o interrogatório policial. O objetivo consiste em investigar, por meio do arcabouço teórico-metodológico da Análise da Conversa de base etnometodológica (SACKS; SCHEGLOFF; JEFFERSON, 1974), como ocorre a busca vs. o resguardo de informações acerca dos crimes sob investigação em interrogatórios policiais de três Delegacias de Polícia do Estado do Rio Grande do Sul, Brasil. Propõe-se, de maneira específica, analisar e descrever as implicações interacionais: (1) dos formatos das perguntas utilizados pelos policiais investigadores na busca por informações concernentes aos crimes; e (2) dos recursos por meio dos quais os participantes dos interrogatórios resguardam informações acerca dos crimes sob investigação. O corpus deste estudo advém de gravações em áudio e/ou vídeo de dez interrogatórios policiais ocorridos em três Delegacias de Polícia Civil, entre abril de 2017 e janeiro de 2018. No que concerne à busca dos fatos dos crimes, a análise evidencia que é a partir das perguntas de formatos menos abertos e de formato fechado, ou então do estreitamento sequencial de perguntas de formato aberto para perguntas de formato fechado, que informações acerca dos crimes são alcançadas pelos policiais investigadores. Essas informações são seguidas de justificativas e/ou informações adicionais que podem não apenas ser substanciais para as investigações, como também podem ser usadas em favor da própria inocência dos interrogados. Em relação ao resguardo das informações acerca dos crimes sob investigação, a análise revela que ele pode ser: (1) realizado pelos interrogados em seus turnos de fala responsivos; e (2) oportunizado pelos policiais investigadores em suas perguntas. Os interrogados resguardam os fatos dos crimes ao resistirem ao provimento das informações solicitadas pelos policiais e ao fornecerem respostas não conformativas àquelas tornadas relevantes nas perguntas, cujas ações consistem em declarações de desconhecimento, deslembrança e dessaber, dentre outras. Os policiais oportunizam que informações concernentes aos crimes sejam resguardadas pelos interrogados quando realizam perguntas cuja composição integra verbos de cognição, tais como “saber” e “lembrar”, possibilitando, assim, que os interrogados declarem, em suas respostas, dessaber e/ou deslembrança sem que revelem resistência ou não conformidade em relação à pergunta. A partir desses resultados, reflete-se sobre a interface entre a ciência da linguagem e as ciências jurídicas, bem como sobre as contribuições que este estudo linguístico-interacional tem a oferecer ao contexto de investigação e ao aparato da Análise da Conversa. / Questioning and answering are trivial activities in people's lives, either in mundane or institutional talk-in-interaction. This dissertation analyzes sequences of questions and answers, as well as its consequences, in interactional and institutional events constituted by questioning and answering practices: police interrogations. The objective is to investigate, supported by the theoretical and methodological framework of Conversation Analysis (SACKS; SCHEGLOFF; JEFFERSON, 1974), how pursuit vs. preservation of information concerning crimes under investigation occur in police interrogations of three Police stations in Rio Grande do Sul State, Brazil. More specifically, it proposes to analyze and describe the following interactional implications: (1) question formats used by police agents in the pursuit of information concerning crimes; and (2) resources through which interrogated individuals preserve information related to crimes under investigation. The corpus of this study is composed of audio and/or video recordings of ten police interrogations from three Civilian Police stations, collected between April, 2017 and January, 2018. Regarding pursuit of crime facts, the analysis shows that information about crimes is obtained by police agents when they use less open or closed questions, as well as when they switch from open to closed-question sequences. This information precedes justifications and/or additional information, which not only can be substantial for the investigation process, but also can be used in favor of the innocence of the individual being interrogated. Concerning preservation of information related to crimes under investigation, the analysis demonstrates that it can be: (1) carried out by interrogated individuals in their responsive turns; and (2) enabled by police agents due to the question formats they choose. Interrogated individuals preserve crime facts when resisting to provide information requested by police agents and when supplying nonconforming responses concerning answers that are made relevant by the question format, performing actions such as lack of knowledge, forgetfulness and unawareness declarations, among others. Police agents enable information related to crimes to be protected by interrogated individuals when they choose question formats which include cognition verbs, such as “to know” and “to remember”, which makes possible, thus, that interrogated people, in their answers, declare ignorance and/or lack of memory without demonstrating resistance or nonconformity towards the question being answered. Based on these results, the study reflects on the interface between linguistic and legal sciences, as well as on contributions that this linguistic and interactional study has to offer to the investigated context and to the Conversation Analysis framework.

Developers of reusable software elements, such as libraries, usually have the responsibility to provide comprehensive and high quality documentation to enable eective reuse of those elements. The eective reuse of libraries depends upon the quality of the API (Application Program Interface) documentation. Well established libraries typically have comprehensive API documentation, for example in Javadocs. However, they also typically lack examples and explanations, which may dicult the eective reuse of the library. StackOverow.com (SO) is a Question and Answer service directed to issues related to software development. In SO, a developer can post questions related to a programming topic and other members of the site can provide answers to help him/her solve the problem he/she has at hand. Despite of the increasing use of SO by the software development community, the information related to a particular library is spread along the website. Thus, SO still lacks an organization of its crowd knowledge. In this dissertation, we present a semi-automatic approach that organizes the information available on SO in order to build a kind of documentation for APIs, called cookbooks (recipe-oriented books). The cookbooks generated by the approach are called crowd cookbooks. In order to evaluate the proposed approach, cookbooks were generated for three APIs widely used by the software development community: SWT,LINQ and QT. Desired features that cookbooks must meet were identied and a study with human subjects was conducted to assess to what extent the generated cookbook meet those features. Through the study it was also possible to identify what is the perceived usefulness by the subjects in relation to the use of cookbooks in APIs learning. The results showed that the cookbooks built using the proposed strategy, in general, meet the identied features. Furthermore, most human subjects considered that cookbooks do not have an appropriate format to the learning of APIs. / Desenvolvedores de elementos reusáveis de software, como as bibliotecas, em geral têm a responsabilidade de disponibilizar documentação abrangente e de alta qualidade para permitir o reuso efetivo desses elementos. O reuso efetivo de bibliotecas depende da qualidade da documentação da API (Interface para Programação de Aplicativos). Bibliotecas bem estabelecidas tipicamente têm documentação abrangente, por exemplo em Javadocs. Porém, essa documentação geralmente carece de exemplos e explicações, o que pode dicultar o reuso efetivo da biblioteca. Stackoverow.com (SO) é um serviço de perguntas e respostas (Q&A) direcionado a questões relacionadas ao desenvolvimento de software. No SO, um desenvolvedor pode postar perguntas relacionadas a um tópico de programação e outros membros do site podem disponibilizar respostas para ajudá-lo a resolver o problema que ele tem em mãos. Apesar da utilização crescente do SO pela comunidade de desenvolvimento de software, a informação relação a um biblioteca está espalhada ao longo do site. Assim, o SO ainda carece de uma organização do crowd knowledge nele contido. Nessa dissertação, será apresentada uma abordagem semi-automatizada que organiza a informação disponível no SO para a construção de um tipo de documentação para APIs, conhecido por cookbooks (livros orientados a receitas). Os cookbooks produzidos pela abordagem proposta são chamados crowd cookbooks. Para avaliar a abordagem proposta foram gerados cookbooks para três APIs amplamente utilizadas pela comunidade de desenvolvimento de software: SWT, LINQ e QT. Foram identicadas características desejáveis de cookbooks e realizado um estudo com sujeitos humanos para entender em que grau os cookbooks construídos atendem a estas características. Por meio estudo também foi possível compreender melhor os pers de uso dos cookbooks mais apropriados em relação ao aprendizado de APIs. Os resultados mostraram que os cookbooks construídos pela estratégia proposta, em geral, atendem às características identicadas. Além disso, a maior parte dos sujeitos humanos considerou que cookbooks não possuem um formato adequado ao aprendizado de APIs. / Mestre em Ciência da Computação

Due to the increasing online commerce in Sweden there are more and more companies who are focusing on their online business and more and more customer are buying clothes online. This means that there has been a transition from the physical store form to the online commerce. Based on this development in the market it aroused an interest in studying customer service, online. Therefore, the purpose of this paper is to study customers experience of customer service from clothing companies within the online industry in Sweden. This carries out by studying if customers experience that they are being encouraged to customer service, what kind of customer service they prefer and the reasons for it. Therefore, we initially make an introductory study of some of the websites belonging to the following clothing companies: H&M, Gina Tricot, Nelly, Boozt, Ellos and Bubbleroom. The main empirical study consists of 15 participants who are divided into three focus groups. To be able to analyse the collected material, three theories are applied: The Customer Contact Theory, Customer Service Life Cycle theory and Social Presence Theory. In terms of the results we alltogether, observe that the participants experience that companies are encouraging customers to use customer service, mainly through different options available on their website, instead of direct contact over the phone. It is also generally shown that the participants prefer screen-based customer service rather than telephone-based customer service. This study is written in the Swedish language. / På grund av den ökande näthandeln i Sverige fokuserar allt fler företag på sin online verksamhet och allt fler kunder handlar kläder online där det har skett en övergång från den fysiska butiksformen till online handeln. Utifrån denna utveckling på marknaden väckte det ett intresse att studera kundservice online. Syftet med denna undersökning är att undersöka konsumenters upplevelse av kundservice som erbjuds av företag som bedriver online handel inom klädbranschen i Sverige. Detta sker genom att studera om konsumenter upplever att de uppmuntras till kundservice, vilken form av kundservice de föredrar och vad som står bakom detta. Därmed görs det först en inledande studie av en del webbsidor som tillhör klädföretag inom online handeln, närmare bestämt: H&M, Gina Tricot, Nelly, Boozt, Ellos och Bubbleroom. Den huvudsakliga empiriska studien består av 15 deltagare fördelat på tre fokusgrupper. För att analysera det insamlade materialet tillämpas tre teorier: The Customer Contact Theory, Customer Service Life Cycle Theory och Social Presence Theory. När det gäller resultatet observerar vi sammantaget att deltagarna upplever att företag uppmuntrar kunder att använda kundservice, främst genom olika alternativ som finns tillgängliga på deras webbsida istället fördirekt kontakt via telefon. Det visar sig även generellt sett att kunder föredrar skärmbaseradkundservice framför telefonbaserad kundservice.