「民以食為天」，飲食與我們的生活息息相關。近年來由於食安風暴肆虐，自行烹煮的需求隨之高漲。然而在家自行烹煮時常會面臨不知道該烹煮什麼料理的問題，因此有便利的食譜搜尋系統對烹煮的人而言將是相當方便的。然而使用搜尋系統時，由於我們只知道想用某些特定食材進行烹煮，而不知道哪些食譜含有特定食材，因此在以少數食材進行查詢時不免會得到過多的食譜結果而難以快速找到喜好的食譜。我們建立了一個食譜搜尋的自動完成機制，並依照該機制實做出了食譜搜尋引擎。使用者使用系統進行搜尋時，我們將會依照使用者輸入的食材尋找適合搭配的食材推薦給使用者，幫助使用者在查詢時使用更完整的Query讓搜尋系統可以找到更少更精準的食譜，幫助使用者更快的找到喜歡的食譜。然而只推薦搭配性食材，可能會推薦出與Query中的食材是替代關係的食材，也就是通常不會一起出現的食材，因此我們也進行了替代性食材的研究。給定由兩個食材組成的食材配對，我們研究如何自動的判斷替代性食材。我們將問題轉化成分類問題來解決，並使用One-Class Classification的技術解決分類問題中的Imbalanced Problem。我們使用f1-score觀看One-Class Classification與傳統分類器的比較。經實驗測試，One Class Classification與傳統分類器相比，One Class Classification較能協助我們解決Imbalanced Problem。

Academic fulltext search engine Egothor has recently became starting point of several thesis aimed on searching. Until now, there was no solution available to provide robust set of web content processing tools. This master thesis is aiming on design and implementation of distributed search system working primary with internet sources. We analyze first generation components for processing of web content and summarize their primary features. We use those features to propose architecture of distributed web search engine. We aim mainly to phases of data fetching, processing and indexing. We also describe final implementation of such system and propose few ideas for future extensions.

This thesis deals with the audit of the board game e-commerce web sites. Nowa-days there is a great number of websites of varying quality, so it is necessary to choose between them. The development of the Internet shifted the emphasis from quantity to quality. With this fact comes the topic of auditing web sites. The first part of the thesis concerns with the theoretical bases of audit of web sites in various sections. The second part includes the representation of the mar-ket of board games and its target group. Based on the first theoretical part the evaluation criteria are created, which are used for the audit of seven selected web sides of board games. This audit is evaluated in the next chapter, which then leads into the conclusion of the work in the form of recommendations for changes in order to increase the competitiveness of each evaluated e-commerce.

Cette thèse se situe dans un axe novateur de recherches en ingénierie et en management des systèmes d’information, elle articule à la fois les aspects de quatre domaines de recherche issus de l’Informatique, des Sciences des Systèmes d’information et des Sciences Humaines et des aspects pratiques liées aux entreprises du Web 2.0. Le «Crowdsourcing», comme son nom l’indique, désigne l’approvisionnement par la foule; Les études et les recherches sur cette thèse se font rares mais celles qui existent confirment l’intérêt managérial des plateformes de Crowdsourcing, grâce à leur rôle incontestable dans la création de valeur. Néanmoins, la foule est composée de groupe d’amateurs hétérogènes, c’est pour cela qu’elle représente aussi une source d’incompétence. Dans ce cadre, notre hypothèse opérationnelle pose que l’apprentissage de la foule stimule la création de valeur dans les plateformes Crowdsourcing. Ainsi, notre travail est, principalement organisé autour de la conception et l’élaboration d’un outil pour l’apprentissage de la foule au sein des plateformes de Crowdsourcing. Ce travail est de nature complexe et relève à la fois d’un travail de recherche et d’une pratique d’ingénierie. C’est pour cela que nous optons pour une démarche constructiviste exploratoire de type qualitative moyennant la méthode de recherche ingénierique qui vise à définir et à concevoir une approche d’apprentissage adaptée aux plateformes de Crowdsourcing et à l’implémenter par la suite au sein d’une plateforme Crowdsourcing de test spécialisée dans les Backlinks. Des expérimentations basées sur des entretiens semi-directifs viendront, à la fin de ce travail, confirmer ou infirmer nos hypothèses. / This thesis is situated in an innovative line of research in engineering and management information systems, it articulates both the aspects of four disciplines of research in the Computer Science, Information Systems, Human Sciences and practical aspects related to Web 2.0 companies. The "Crowdsourcing" as its name suggests, refers to the sourcing by the crowd, studies and research on this topic are infrequent but those that exist confirm the managerial interest of Crowdsourcing platforms, thanks to their undeniable role in value creation. Nevertheless, the crowd is composed of heterogeneous group of amateurs that is why it is also a source of incompetence. Our operating hypothesis posits that learning the crowd stimulates the creation of value in the Crowdsourcing platforms. Thus, our work is mainly organized around the design and development of a tool for learning the crowd in Crowdsourcing platforms. This work is complex and involves both a research work and practical engineering. That is why we choose an exploratory qualitative constructivist approach and an ingénierique research method to define and develop a suitable approach of learning adapted to the Crowdsourcing platforms and implement it thereafter within our test Crowdsourcing platform specializes in Backlinking. Experiments based on semi-structured interviews will, confirm or deny our hypotheses.

This diploma work deals with specific business concept and with possibilities of a company promotion on the Internet. The theoretical segment contains a survey of the Internet business models and a synopsis of current individual forms of promotion used on the Internet. There are described advantages and disadvantages of banner advertising, contextual advertising, email marketing, search engine marketing (SEM), social networks and affiliate marketing. The final part of theoretical section of the thesis is dedicated to the possibilities of propagation efficiency measurements. Practical part of the diploma work introduces author's own business concept and its business model. The description of specific methods of propagation realized in congruence with the noted business concept follows.

Titel:                                  David vs. Goliath Författare:                        Henriksson, Johannes & Ågehäll, Adam Institution:                        Linnéuniversitetet, Ekonomihögskolan Program:                           Ekonomprogrammet inriktning Marknadsföring Kurs:                                 Företagsekonomi III – Marknadsföring, Examensarbete, 15hp Handledare:                      Leif Rytting Examinator:                      Kaisa Lund   Syfte:                                 Uppsatsens syfte är att undersöka huruvida mikroföretag kan skapa marknadsfördelar av att implementera Search Engine Marketing. Ytterligare en avsikt är att uppsatsen skall bidra med kunskap kring hur mikroföretag kan stärka sitt varumärke genom Search Engine Marketing.    Metod:                               Kvalitativ studie innehavande en mix mellan induktiv- och deduktiv forskningsansats. Även kallad abduktiv forskningsansats. Den empiriska datan är insamlad genom semi-strukturerade intervjuer bestående av tio respondenter.   Slutsats:                             Sökmotormarknadsföring är en grundkomponent i den digitala marknadsföringen. Mikroföretag skapas möjlighet att konkurrera mot större varumärken genom skapad synlighet. Marknadsföring via en nischad marknad visar sig fördelaktig. En sammanstrålad varumärkesidentitet och varumärkesimage bör avspeglas i content.   Nyckelord:                          Sökmotormarknadsföring, SEO, PPC, Google, mikroföretag, sökmotorer, sociala medier, varumärken.   Fördelning av arbete:         Båda författarna har varit lika delaktiga under alla delar av studien. / Title:                                   David vs. Goliath Authors:                             Henriksson, Johannes & Ågehäll, Adam Institution:                          Linnaeus University, School of Business and Economics Programme:                       Business Administration and Economics, specialization Marketing Course:                               Business Administration III – Marketing, Bachelor Thesis, 15hp Supervisor:                         Leif Rytting Examiner:                           Kaisa Lund   Purpose:                             The purpose of this study is to examine how micro enterprises can create market advantages by implementing Search Engine Marketing. Further intention is that the study will provide knowledge about how micro enterprises can invigorate they brand through Search Engine Marketing.    Method:                              Qualitative study using a mix of both inductive- and deductive research approach. Also known as abductive approach. The empirical data is collected using semi-structured interviews, with ten respondants.   Conclusion:                         Search Engine Marketing is a basic component in digital marketing. Micro enterprises creates the opportunity to compete against larger brands through visibility. Niche marketing is found valueable. A converged brand identity and brand image should be mirrored in content.   Keywords:                          Search engine marketing, SEO, PPC, Google, micro enterprises, search engines, social medias, brands.   Allocation of work:             Both authors have been equally involved in all parts of the thesis.

Recursos Educacionais Abertos (REA) podem ser definidos como materiais de ensino, aprendizagem e pesquisa, em qualquer meio de armazenamento, que estão amplamente disponíveis por meio de uma licença aberta que permite reuso, readequação e redistribuição sem restrições ou com restrições limitadas. Atualmente, diversas instituições de ensino e pesquisa têm investido em REA para ampliar o acesso ao conhecimento. Entretanto, os usuários ainda têm dificuldades de encontrar os REA com os mecanismos de busca atuais. Essa dificuldade deve-se principalmente ao fato dos mecanismos de busca na Web serem genéricos, pois buscam informação em qualquer lugar, desde páginas de vendas até materiais escritos por pessoas anônimas. De fato, esses mecanismos não levam em consideração as características intrínsecas de REA, como os diferentes padrões de metadados, repositórios e plataformas existentes, os tipos de licença, a granularidade e a qualidade dos recursos. Esta dissertação apresenta o desenvolvimento de um mecanismo de busca na Web especificamente para recuperação de REA denominado SeeOER. As principais contribuições desta pesquisa de mestrado consistem no desenvolvimento de um mecanismo de busca na Web por REA com diferenciais entre os quais se destacam a resolução de conflitos em nível de esquema oriundos da heterogeneidade dos REA, a busca em repositórios de REA, a consulta sobre a procedência de dados e o desenvolvimento de um crawler efetivo para obtenção de metadados específicos. Além disso, contribui na inclusão de busca de REA no cenário brasileiro, no mapeamento de padrões de metadados para mecanismos de busca na Web e a publicação de uma arquitetura de um mecanismo de busca na Web. Ademais, o SeeOER disponibiliza um serviço que traz um índice invertido de busca que auxilia encontrar REA nos repositórios dispersos na Web. Também foi disponibilizada uma API para buscas que possibilita consultas por palavras chaves e o uso de palavras booleanas. A forma de validação em mecanismos de busca na Web, como um todo, e de forma quantitativa e específica por componentes foi feita em grau de especialidade. Para validação de qualidade foram considerados 10 participantes com grupos distintos de escolaridade e área de estudo. Os resultados quantitativos demonstraram que o SeeOER é superior em 23.618 REA indexados em comparação a 15.955 do Jorum. Em relação à qualidade o SeeOER demonstrou ser superior ao Jorum considerando a função penalizada e o score utilizada nesta pesquisa. / Open Educational Resources (OER) has been increasingly applied to support students and professionals in their learning process. They consist of learning resources, usually stored in electronic device, associated with an open license that allows reuse, re-adaptation and redistribution with either no or limited restrictions. However, currently the Web search engines do not provide efficient mechanisms to find OER, in particular, because they do not consider the intrinsic characteristics of OER such as different standards of metadata, repositories and heterogeneous platforms, license types, granularity and quality of resources. This project proposes a Web search engine, named SeeOER, designed to recover OER. Main features of SeeOER are: schema-level con ict resolution derived from the heterogeneity of OER, search for Brazilian OER repositories, query considering data provenance and the development of an effective crawler to obtain specific metadata. In addition, our project contributes to the inclusion of the search OER research issues in the Brazilian scenario, to the mapping of metadata standards to Web search engine. In addition, SeeOER provides a service which internally has an inverted index search to find the OER which is different from traditional Web repositories. We also provide an API for queries which make it possible to write queries based on keywords and boolean. The validation of the search engine on the Web was both qualitative and quantitative. In the quantitative validation it was observed in level of specialty of the search engines components. In conclusion, the quality and quantitative results experiments showed that SeeOER is superior in OER indexed 23,618 compared to 15,955 the Jorum. In relation to the quality SeeOER shown to be superior to Jorum 27 points considering the metric used in project.

Submitted by Filipe dos Santos (fsantos@pucsp.br) on 2018-05-10T12:54:13Z No. of bitstreams: 1 Carolina Ferreira Souza.pdf: 1903774 bytes, checksum: fb12be505b4c7a36b6455c716c3e0891 (MD5) / Made available in DSpace on 2018-05-10T12:54:13Z (GMT). No. of bitstreams: 1 Carolina Ferreira Souza.pdf: 1903774 bytes, checksum: fb12be505b4c7a36b6455c716c3e0891 (MD5) Previous issue date: 2018-03-14 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Fundação São Paulo - FUNDASP / The present work deals with the lack of transparency of the Google search engine, analyzing what the rights of consumers users of this service are affected, as well as the damage caused to the balance of the digital environment. The capture of data and information circulating in the most different pages, programs and applications belonging to the large corporations of the network is increasingly refined and less transparent. In addition, access to existing data and information is increasingly being filtered and directed to partners, advertisers, in more or less explicit ways, depending on the service provided or the page accessed. In this way, the treatment that must be reserved for the digital environment must begin with the immediate and full universalization of Internet access and its data in favor of all without distinction or privileges, exclusions or exceptions, as well as the transparency of the algorithms that operate in the network through machine learning and other forms of artificial intelligence, under penalty of not authorizing those harmed by state inaction and / or by the action of capitalist speculation, to take the administrative and judicial measures applicable and applicable to the species for the realization of their rights. The methodology used was based on a bibliographic review to analyze the data obtained, with the collection of textual documents such as: legislation, jurisprudence, relevant doctrine, and publications of a technical nature, as well as an analysis of some of the terms of use and data policies of Google, in order to verify the lack of transparency and disrespect for human rights and Brazilian consumer rights and help in understanding the problem of commercialization and bias of the digital media. The results of the research indicated that the existing legislation is sufficient for the protection of consumers users of the Google search service in several aspects; however, considers the possibility of new data protection legislation as one more strategy within the law for the protection of human rights to communicate / O presente trabalho trata da falta de transparência do buscador Google, analisando-se quais os direitos dos consumidores usuários desse serviço são afetados, bem como o prejuízo causado ao equilíbrio do meio ambiente digital. A captação de dados e informações que circulam nas mais diferentes páginas, programas e aplicativos pertencentes às grandes corporações da rede está cada vez mais refinada e menos transparente. Além disso, o acesso aos dados e informações já existentes está sendo cada vez mais filtrado e direcionado para parceiros, anunciantes, de formas mais ou menos explícitas, dependendo do serviço prestado ou da página acessada. Desta forma, o tratamento que se deve reservar ao meio ambiente digital deve começar pela imediata e plena universalização do acesso à internet e aos seus dados em favor de todos, sem distinção ou privilégios, exclusões ou exceções, bem como pela transparência dos algoritmos que operam na rede através do aprendizado de máquina e outras formas de inteligência artificial, sob pena de, em não o sendo, autorizar os prejudicados pela inação estatal e/ou pela ação da especulação capitalista, a tomar as medidas administrativas e judiciais cabíveis e aplicáveis à espécie para a efetivação de seus direitos. A metodologia empregada valeu-se de revisão bibliográfica para análise dos dados obtidos, com a coleta de documentos textuais como: legislações, jurisprudências, doutrina pertinente, e publicações de caráter técnico, bem como análise de parte dos termos de uso e de políticas de dados do Google, a fim de se constatar a falta de transparência e o desrespeito aos direitos humanos e direitos do consumidor brasileiros e auxiliar na compreensão da problemática da mercantilização e parcialidade dos meios de comunicação digital. Os resultados da pesquisa indicaram que a legislação já existente é suficiente para a proteção dos consumidores usuários do serviço do buscador Google em diversos aspectos; entretanto, considera a possibilidade de uma nova legislação de proteção de dados como mais uma estratégia dentro do direito para a proteção dos direitos humanos à comunicar-se

<p>Examensarbetet har gått ut på att genom kvalitativa undersökningar ta reda på hur grafisk form ochinformationsstruktur samverkar för att skapa användarcentrerade webbplatser. Målet var att ta fram en grund för huranvändarcentrerade webbplatser bör byggas. Denna grund är tänkt att fungera som riktlinjer för formgivare som villskapa användarcentrerad webbdesign.</p><p>Studien visar att det viktigaste för att skapa användarcentrerad webbdesign är att involvera användarna kontinuerligti utvecklingsprocessen. Detta sker genom användartester och intervjuer med representanter för målgruppen. Vi harkommit fram till att graden av användbarhet spelar stor roll för hur målgruppen uppfattar webbplatsen som sådan,men också vilken uppfattning de får om företaget bakom.</p> / <p>This thesis considers a qualitative study in how info structure and graphical design collaborate to form a usercentredweb site. The aim was to produce guidelines for how to build user-centred web sites. The results from thestudy are built on qualitative interviews, user tests with the target group and expert evaluation.</p><p>The study has shown that the most important thing for creating user-centred web sites is to continuously involve thetarget group in the development process. This should be done through interviews and user tests. The level ofusability determines how the target group experience both the company and their web site.</p>

<p>This thesis examines companies offering their services for free to Internet users, byemploying digital free business models. As a framework Chris Anderson’s classificationsof “free” business models are used. A sample of eleven companies that provide “free”services was selected and divided into four groups. These were search engine, socialnetworking/community, content based and others. Their profitability was then measuredin relation to their valuation with the help of P/E ratios within and among the groups. Aregression analysis was also conducted to compare profitability of either one of two“free” business models used by the researched companies.Findings were that search engine and social networking/community companies appear tohave profits for the period researched. No strong trend for overvaluation could be foundin either of these groups, except for individual companies with high P/E ratios. Neithercompany within the content based group showed any profits. Their marginal costs weretoo high but this may change with technological progress. Regression analysis could notshow any significant results employing either the “Freemium” or the advertising “free”business model to be more profitable than the other. Significant results could be shownbeing a content based company and being unprofitable. Comparison between specificcompanies gave mixed results but network effects appear to create dominant playerswithin each group. Employing more than only the advertising “free” business modelseems to be efficient in raising revenue per user for social networking/communitycompanies.</p>