På grund av kalhuggning har förutsättningen förändrats förhålboende fåglar. Bohålsbristen går att åtgärda med placering av nya fågelholkar. Genom att använda GIS går det att strategiskt välja de bästa områdena att sätta ut nya fågelholkar. Med hjälp av crowdsourcing och mobiltelefoner går det kartlägga holkarna som kan användas som underlag för nyplacering eller som hjälp för analys som t.ex. arters spridning. Studien visar hur man kan erhålla ett system för multikriterieanalys i realtid online för mobiltelefoner som utnyttjar crowdsourcing och öppna standarder. PostGIS, Leaflet, Geoserver, PyWPS, GRASS GIS har använts för att uppfylla kraven på prototypen. Tidsåtgången av detta system utvärderades genom att mäta tiden för utförande av algoritmen och tiden för förfrågan till utritningen av resultatet. Resultatet visar att det är möjligt att utföra multikriterieanalyser på kort tid som är acceptabel för användare av smarta mobiltelefoner på fält och att det går att urskilja närliggande fågelholkar. Denna metod skulle kunna anpassas för liknande scenarion.

Online Social Networks (OSNs) have seen an exponential growth over the last decade, with Facebook having more than 1.49 billion monthly active users and Twitter having 135,000 new users signing up every day as of 2015. Users are sharing 70 million photos per day on the Instagram photo-sharing network. Yahoo Answers question-answering community has more than 1 billion posted answers. The meteoric rise in popularity has made OSNs important social platforms for computer-mediated communications and embedded themselves into society’s daily life, with direct consequences to the offline world and activities. OSNs are built on a foundation of trust, where users connect to other users with common interests or overlapping personal trajectories. They leverage real-world social relationships and/or common preferences, and enable users to communicate online by providing them with a variety of interaction mechanisms. This dissertation studies abuse and privacy in online social networks. More specifically, we look at two issues: (1) the content abusers in the community question answering (CQA) social network and, (2) the privacy risks that comes from the default permissive privacy settings of the OSNs. Abusive users have negative consequences for the community and its users, as they decrease the community’s cohesion, performance, and participation. We investigate the reporting of 10 million editorially curated abuse reports from 1.5 million users in Yahoo Answers, one of the oldest, largest, and most popular CQA platforms. We characterize the contribution and position of the content abusers in Yahoo Answers social networks. Based on our empirical observations, we build machine learning models to predict such users. Users not only face the risk of exposing themselves to abusive users or content, but also face leakage risks of their personal information due to weak and permissive default privacy policies. We study the relationship between users’ privacy concerns and their engagement in Yahoo Answers social networks. We find privacy-concerned users have higher qualitative and quantitative contributions, show higher retention, report more abuses, have higher perception on answer quality and have larger social circles. Next, we look at users’ privacy concerns, abusive behavior, and engagement through the lenses of national cultures and discover cross-cultural variations in CQA social networks. However, our study in Yahoo Answers reveals that the majority of users (about 87%) do not change the default privacy policies. Moreover, we find a similar story in a different type of social network (blogging): 92% bloggers’ do not change their default privacy settings. These results on default privacy are consistent with general-purpose social networks (such as Facebook) and warn about the importance of user-protecting default privacy settings. We model and implement default privacy as contextual integrity in OSNs. We present a privacy framework, Aegis, and provide a reference implementation. Aegis models expected privacy as contextual integrity using semantic web tools and focuses on defining default privacy policies. Finally, this dissertation presents a comprehensive overview of the privacy and security attacks in the online social networks projecting them in two directions: attacks that exploit users’ personal information and declared social relationships for unintended purposes; and attacks that are aimed at the OSN service provider itself, by threatening its core business.

Denna studie har undersökt vilka användbarhetsfaktorer och funktioner som gör att en karttjänst får hög användbarhet och ökad tillgänglighet till allmänheten. En karttjänst utvecklades som utgångspunkt i studien. Karttjänsten kan lagra aktuella positionsdata för vägbommar. Karttjänsten användes i samband med användbarhetstester för att evaluera användbarheten. Resultaten från användbarhetsundersökningen avhandlar användbarhetsfaktorer i användargränssnittet. Crowdsourcing antogs som förutsättning och användes som underlag i utformningen av användbarhetsundersökningen.

Contemporary cases of crowdsourcing (CS) and open source development (OS) related to drug development have been selected and studied. Contemporary examples of CS/OS from within and outside of the pharmaceutical industries have been presented to give a background and suggest possible benefits and problems. The main criteria for selection have been that the case must seek to advance drug development and must use crowdsourcing or open source as a mechanism. The cases found in our search show a large diversity in terms of application, usage, and possible implications for the pharmaceutical industry. We found that crowdsourcing within a scientific problem context produced good results, but that open source initiatives were either poorly financed and not successful or focused on neglected diseases made possible through strong backing by non-profit organizations. An analysis of which the pharmaceutical companies where that showed activity on the platforms identified R&D-intensive and biotech companies as the most active. Contract research organizations (CROs) and generics manufacturers (GMs) showed almost complete absence. We argue that GMs are not likely to be interested in this kind of R&D, but CROs are an untapped resource. Finally we propose a hypothetical model that takes into account all the findings from our study and the literature. This model is based on a limited type of open source with a limited number of partners making use of the untapped CRO resource through crowdsourcing. / Ett antal pågående fall av samverkans-modellen crowdsourcing (CS) samt fenomenet av ”öppen-innovation”, open-innovation (OI) har studerats i kontexten läkemedelsutveckling. I denna fall-studien presenteras ett antal fall där CS och OI används både inom och utom läkemedelsindustrin. Detta har gjorts för att skapa en bättre förståelse för fenomenen samt att tydliggöra problem och att belysa fördela med öppen-innovation och olika typer av samverkans-modeller. Huvudkriteriet för utvalen av fall har varit ett avancerat stadium i läkemedelsutvecklingen samt att något av de två verktygen OI eller CS används aktivt i utvecklingsarbetet. Fall-studien klargör en tydlig differens mellan olika applikations-typer och hur man i verkligheten implementerar dessa verktyg i forskningsarbete. Vår fall-studie indikerar goda resultat då man använder samverkans-modellen CS inom forskningsarbete, men att ”öppen-innovation” OI inte gav samma goda resultat då immateriella mekanismer bromsar öppenheten i läkemedelsindustrin. Inom läkemedels-utvecklingen för Negleted Diseases fann vi däremot en större aktivitet vilken grundar sig på icke-vinst-drivande aktörers intressen. Genom en analys av aktörer inom läkemedelsindustrin har vi kunna framtyda att forsknings-och utveckligns-företag samt Biotech som de som mest ser och utnyttjar potentiallen i CS och OI. Contract research organizations (CROs) och Generics Manufacturers (GMs) är nästan uteslutande frånvarande i resultaten från vår analys. Våra resultat visar att GM-företagen inte har mycket intresse av varken CS eller OI, men att för CRO-företagen finns det vinning i att utnyttja dessa resurser. Slutligen förespråkar vi en fortsatt forskning i ämnet baserat på fall-studien för att kunna ta fram en hypotetisk-CS modell, vilken skulle baseras på ett begränsat antal av samarbetspartners vars potentiella vinning skulle visas genom utnyttjandet av samverkans-modellen CS. / Johan Evaldsson +733-706822

Using the concept of open innovation companies have started to look beyond their own borders using the crowd as a source for labour, knowledge and innovation. During the last few years the phenomena of crowdsourcing and crowdfunding have become increasingly popular as sources of competitive advantage through the crowd. While several studies have examined crowd based platforms focusing on the users, little attention has been brought to the companies starting these crowd based services. This study addresses this gap in existing knowledge related to crowd based platforms through a qualitative case study examining a Swedish start-up. We demonstrate challenges related to networking, critical mass and quality in the context of launching a crowd based platform.

Art dans l’espace public, art public, art à ciel ouvert, art in situ, Land art ou encore musée sans murs, autant de termes polysémiques qui désignent un champ artistique souvent au coeur des polémiques mais finalement peu documenté. Cette thèse en esthétique s’inscrivant dans le mouvement des humanités numériques est consacrée à la problématique de l’inventaire, de la cartographie et de la visualisation des oeuvres d’art dans l’espace public. Le projet de recherche est constitué de deux volets : une plateforme numérique - le site Wiki et l’application mobile Atlasmuseum - et la thèse. L’examen de plusieurs initiatives, nationales et internationales, de cartographie d’oeuvres d’art public, de projets pionniers d’inventaires numériques et de collections institutionnelles sur le Web s’est opéré à partir d’une question centrale : comment concevoir une instance muséale pour des oeuvres situées précisément en dehors du périmètre du musée ? Et de manière corollaire : comment modéliser un « musée réticulaire » dans le contexte de l’ouverture, de l’interopérabilité des données culturelles et du Web sémantique ? Partant du constat de l’hétérogénéité et de l’incommensurabilité constitutive du corpus d’oeuvres ainsi qu’à travers la lecture rapprochée et distanciée d’une sélection d’oeuvres d’art public contemporain (issues notamment du « 1% artistique » et de la commande publique entre 1951 et 2016), la thèse est construite autour de la conception d’une instance muséale reliant à travers leurs notices plusieurs centaines d’oeuvres, d’artistes et des lieux. La plateforme Atlasmuseum est un dispositif à la fois curatorial, muséologique et muséographique qui regroupe selon un mode contributif curators, conservateurs, amateurs d’art, artistes, informaticiens, historiens de l’art, documentalistes, enseignants et étudiants ; elle repose sur le logiciel Mediawiki enrichi de l’extension SemanticMediawiki. A l’intersection de la cartographie, de la muséologie critique et de l’anthropologie de l’art, la thèse est structurée en quatre parties. La première est consacrée à la carte comme médiation entre les oeuvres et les lieux. La carte y est examinée selon différents points de vue afin d’esquisser les grandes lignes d’un outil d’inventaire potentiel pour les oeuvres d’art public. La deuxième partie de la thèse consiste à prolonger les premières hypothèses en opérant un double changement d’échelle, de la carte à l’atlas et symétriquement de l’inventaire au musée. Suit la partie « centrale » de la thèse, qui constitue une version « papier » de la plateforme Atlasmuseum dans ses deux formats (Web et mobile) et dans ses quatre dimensions (exploration, contribution, recherche et sémantisation). Dans la quatrième partie est proposée une lecture « à distance » des données récoltées et des hypothèses explorées dans les parties précédentes, notamment à travers la constitution d’un corpus d’oeuvres restreint. L’hypothèse explorée dans cette partie est celle de la visualisation ontologique des notices d’oeuvres à travers une modélisation et une spatialisation spécifique des données et des métadonnées (selon le modèle CIDOC CRM ). Il s’agit ainsi au fil de la thèse d’opérer plusieurs glissements de terrain successifs et rétroactifs, de la théorie à la pratique, du curating artistique vers la curation de données, de la carte vers l’inventaire, de l’atlas vers le musée, du diagramme vers le réseau. / Public art, site specific, in situ art, Land art or museum without walls, various polysemic terms are used to designate the field of art in public space, much commentated but finally little documented. This thesis in Aesthetics and Digital humanities aims to account to the modeling process of a “reticular museum” for public art. The research project consists of two parts: a digital platform - the Atlasmuseum wiki site and mobile application - The application and the thesis itself. A review of several, national and international initiatives of public art maps, pioneering projects of digital inventories and online institutional collections has been conducted from a central question: how to design a museum structure for artworks located precisely outside the museum traditional perimeter? And consequently: how to model a contributory inventory tool in the context of cultural linked open data and the Semantic Web? Taking account of the immeasurable and heterogeneous corpus of public artworks, and through close and distant readings of a selection of contemporary public artworks (“1% for art in architecture” and public art commissions in France between 1951 and 2016), the thesis is built around the concept of a “reticular museum” connecting through their records hundreds of art pieces, artists and sites. Dedicated to public art inventory, through specifically designed notices, maps, atlases and visualizations, the Atlasmuseum platform aims to reunite on a contributory basis, curators, artists, computer scientists, art historians, librarians, art lovers, teachers and students. Atlasmuseum is based on the MediaWiki software enriched with the Semantic MediaWiki extension. It is a curatorial as well as a museological and museographical project. At the intersection of critical cartography, critical museology and anthropology of art, the thesis is structured in four parts. The first is devoted to the map as mediation between artworks and their sites. The map is examined from different points of view in order to sketch the outlines of a potential inventory tool for public art. The second part of the thesis aims to extend the initial assumptions by working a double shift of scale, from the map to the atlas and symmetrically from the inventory to the museum. Follows the “central” part of the thesis, which is a “printed” version of the Atlasmuseum platform in both formats (Web and mobile) and its four dimensions (exploration, contribution, research and semantization). In the fourth part, a close and distant reading of a specific collection of artworks (art pieces created for the tramway networks in France) is conducted through data visualizations. The hypothesis is that an ontological data visualization (using the CIDOC CRM model) can offer a new form of perception and apprehension of these artworks within their environment and context. Thus the thesis operates successive and retroactive landslides, from theory to practice, from artistic curating to data curation, from the map to the inventory, from the atlas to the museum and from the diagram to the network.

Syftet med uppsatsen är att bidra med kunskapsutveckling kring hur ”crowdsourcing” idag används och upplevs som ett stöd för att säkra och utveckla kvalitet inom svensk krisberedskap. Detta uppnås genom tillämpning av en explorativ, pragmatisk och abduktiv ansats av mixad metod. Insamlad empiri utgörs av en enkät med respondenter från 119 kommuner samt fyra kvalitativa intervjuer. Resultatet visar att crowdsourcing inom ramen för svensk krisberedskap inte är ett känt begrepp bland de responderande kommunerna men att intervjupersonerna känner till begreppet i relation till krisberedskap. Idag tillämpar Malmö Stad och Krisinformation.se crowdsourcing inom ramen för deras krisberedskapsrelaterade verksamhet, pilotförsök med crowdsourcing-liknande metoder pågår hos SOS Alarm. Västerås Stad diskuterar området inför eventuell framtida utveckling. Den verksamhet som idag bedrivs inom krisberedskapsrelaterad crowdsourcing inom Sverige är begränsad till endast ett fåtal aktörer som bedriver verksamhet eller pilotförsök. Crowdsourcing som generellt begrepp är ännu inte välbekant bland många svenska kommuner och än mindre ur ett krisberedskapsperspektiv. Vidare finns det kvalitetsmässiga farhågor över till exempel inmatning av felaktiga data som genererar följdkonsekvenser. Det teoretiska ramverket presenterar flera olika forskningsområden som kan tillämpas för att lösa dessa farhågor. Det uppsatsen tillför forskningen är ett sammanställt utgångsläge utifrån existerande forskning och praktisk erfarenhet med goda möjligheter för framtida forskning (eller tillämpning) inom ett för Sverige synnerligen ungt och outforskat forskningsområde. / The purpose of this thesis is to add to the development of a knowledgebase concerning how crowdsourcing is used and experienced as a support to secure and develop quality within Swedish societal crisis management. This is accomplished by applying an explorative, pragmatic and abductive approach of mixed theory. Collected data consists of one survey with 119 respondents (municipalities) along with four qualitative interviews. The result show that crowdsourcing within Swedish societal crisis management is not known by the survey respondents but known within that context by the participants of the interviews. Malmö City and krisinformation.se applies crowdsourcing methods as a part of their crisis preparedness operation today, trials with crowdsourcing-like methods is conducted by SOS Alarm and last Västerås City are discussing this area for potential future development. Crisis related crowdsourcing operations within Sweden today are reduced to a few actors conducting either regular operations or trials. Crowdsourcing as a general term is not yet well known among many Swedish municipalities and even less known with a crisis preparedness perspective. Additionally, there are quality related concerns such as input of erroneous data leading to further consequence. The theoretical framework presents several research areas that can be applied in solution to such quality concerns. This thesis addition to research is a compiled platform based of existing research and practical experiences with plenty of room for future research or practical application within a for Sweden particularly young and unexplored field of study. / <p>2017-06-28</p>

Abstract This thesis focuses on a systematic assessment of the feasibility and performance of situated crowdsourcing, as well as a basic understanding of the behaviours of its workers. While these aspects have been extensively studied for online and mobile crowdsourcing, this is not the case for situated crowdsourcing mainly due to its relative novelty. Such an assessment is crucial for the development of the crowdsourcing research agenda, so that task requesters and researchers alike can leverage, whenever appropriate, situated technologies for crowdsourcing efforts with more confidence. The key findings of this thesis illustrate how appropriately designed crowdsourcing tasks can perform well even in a complex deployment setting: situated technologies in public spaces. In the articles presented in this thesis, we empirically demonstrate that situated crowdsourcing performance can compete with other means of collecting crowd contributions. While situated technologies have been reported in the past to suffer from credibility and misappropriation issues, one should not forego the use of these technologies for crowdsourcing purposes assuming that the tasks are not haphazardly designed. The thesis also explores the behaviours of situated crowdsourcing workers through in-situ observations, video analysis and longitudinal individual tracking. Towards the end of the thesis, we revisit the research questions put forth in the thesis, and highlight how they were answered. We then discuss the benefits and drawbacks of situated crowdsourcing, and the differences between using non-personal and personal devices for this purpose. In both cases, the decisions made by the task requesters or researchers will ultimately depend on their goals and the task itself. We conclude the thesis by restating the thesis’ research agenda, reflecting on the challenges and opportunities of situated crowdsourcing, and our future work within this area. / Tiivistelmä Tämä väitöskirja keskittyy paikkasidonnaisen joukkoälyn käyttökelpoisuuden sekä tehokkuuden järjestelmälliseen arviointiin. Väitöskirja pyrkii myös ymmärtämään joukkoälyjärjestelmien työntekijöiden käyttäytymistä ohjaavia tekijöitä alustavalla tasolla. Paikkasidonnaisuus on tekijä, jota useimmiten verkossa tai mobiililaitteissa tehtävässä joukkotyötutkimuksessa ei ole mahdollista ymmärtää perusteellisesti. Paikkasidonnaisuus muodostaa kuitenkin elintärkeän osan joukkoälytutkimuksessa, ja sitä hyväksikäyttämällä sekä joukkotyön teettäjät että joukkoälytutkijat voivat soveltuvissa tilanteissa hyödyntää paikkasidonnaisia teknologioita luotettavamman joukkoälytiedon tuottamiseen. Väitöskirjan keskeisimmät löydökset osoittavat kuinka tarkoituksenmukaisesti toteutetut joukkoälytehtävät tuottavat luotettavaa tietoa, jopa monimutkaisissa käyttöympäristöissä kuten paikkasidonnaisia teknologioita hyödyntävissä julkisissa tiloissa. Väitöskirjan artikkelit osoittavat empiirisesti paikkasidoinnaisen joukkoälyn olevan kilpailukykyinen muiden joukkoälyteknologioiden kanssa, vaikka paikkasidonnaisten teknologioiden on aiemmin osoitettu kärsivän uskottavuuden puutteesta sekä väärinkäytöstä. Tämän väitöskirjan löydökset osoittavat, että oikein suunnitellut joukkoälytehtävät sopivat hyvin käytättäväksi kyseisten teknologioiden kautta. Suorituskyvyn sekä tehokkuuden lisäksi väitöskirjassa esitellään empiirisiin havaintoihin, videoanalyysiin, sekä yksilöiden pitkäkestoiseen tutkimukseen pohjautuvia löydöksiä joukkoälytyöntekijöiden käyttäytymismalleista. Väitöskirjan loppuosa käsittelee henkilökohtaisten ja julkisten laitteiden hyötyjä sekä haittapuolia suhteessa paikkasidonnaisten joukkoälyjärjestelmien käyttöön. Löydökset osoittavat, että kummassakin tapauksessa laitetyypin valintaa ohjaavat joukkoälytyön teettäjien tai tutkijoiden asettamat tavoitteet, sekä kyseessä olevat joukkoälytehtävät. Väitöskirjan päätteeksi palataan asetettuihin tutkimuskysymyksiin sekä vastaaviin löydöksiin, ja pohditaan paikkasidonnaisen joukkoälyn tuomia haasteita, mahdollisuuksia sekä tulevaisuuden visioita.

Clinical skills are often measured using objective structured clinical examinations (OSCEs) in healthcare professions education programs. As with assessment centers, it is challenging to provide learners with effective feedback due to burdensome human capital demands. The aim of this dissertation was to evaluate the viability of using a crowdsourced system to gather OSCE ratings and feedback. Aggregating evaluations of student performance from a crowd of patient proxies has the potential to mitigate biases associated with single-rater evaluations, allow the patient a voice as the consumer of physician behavior, improve reliability, reduce costs, improve feedback latency, and help learners develop a mental model of the diversity of patient preferences. Crowd raters, recruited through Amazon Mechanical Turk, evaluated a set of video-recorded performance episodes designed to measure interpersonal and communication (ICS) and physical exam (PE) skills. Compared to standardized patient (SP) and faculty raters, crowd raters were more lenient and less reliable, when holding the number of raters and spending constant. However, small groups of crowd raters were able to reach acceptable levels of reliability. Crowd ratings were collected within a matter of hours whereas SP and faculty ratings were returned in over 10 days. Learner reactions to crowdsourced ratings were also measured. Blind to the rater source, a majority of learners preferred the crowdsourced feedback packages over the SP and faculty packages. After learning about the potential value of crowdsourced ratings, learners were positive about crowd ratings as a complement to SP and faculty ratings, but only for evaluations of ICS (not PE) and only for formative (not summative) applications. In particular, students valued the volume and diversity of the crowdsourced feedback and the opportunity to better understand the patient perspective. Students expressed their concerns about privacy as well as the accuracy and quality of crowd ratings. A discussion of practical implications considers future best-practices for a crowdsourced OSCE rating and feedback system.

Some of the most pressing challenges facing humanity today, such as how to respond to climate change or govern the internet, are too complex for any individual or small group to completely understand by themselves, but a lot of people each know a piece of the problem or solution. This is somewhat like a billion-piece jigsaw puzzle where somebody threw away the box and mailed each piece to a different person. Attempts are now being made to build platforms where people can bring their pieces and assemble them into solutions. This thesis examines such platforms, how people use and perceive them and their content, and how certain design decisions such as the exposure of discussion behind collaboratively produced content can affect those perceptions. Through a set of studies ranging from qualitative interviews to controlled experiments with hundreds or thousands of participants, this thesis adds to our understanding of how humans use and perceive content on such platforms, including when primary content is presented alongside related content or discussion, so that we can better understand how to design these systems to better achieve their users’ intended goals. The research integrates insights from computer science and social psychology with latent variable modeling techniques in order to increase our understanding of what people are trying to accomplish on an example platform designed to support large-scale collaboration around a complex issue, and experimentally explores how people perceive the content they find on such sites. Data and projects used in this thesis come from a diversity of sources including Wikipedia, the President’s SAVE award ideation contest (facilitated through the IdeaScale ideation platform), and the MIT Climate CoLab.