Manufacturers and suppliers in the tech industry, trading and utilizing minerals, are often unable to conduct substantial supply chain due diligence, due to reasons such as lack of competence, the scattered spread of information and fluid nature of their supply chains. Declaring whether a product has been responsibly sourced, or whether it contains conflict minerals or not, is almost impossible. This study is an exploration of the potential role of blockchain in mineral supply chain management, as a supplementary tool for carrying out due diligence. Well-performed supply chain due diligence should demand continuous status records of various measures of social sustainability, identifying impacts on human well-being. So, how may a blockchain solution for traceability in a mineral supply chain contribute towards ensuring responsible sourcing? Blockchain provides traceability of transactions through its immutable chain structure, and knowing an asset’s origin is vital in order to carry out supply chain due diligence. While the blockchain network has the potential to provide information on the digitally registered flow of an asset, the validity of the information of the physical and social qualities of the asset remains dependent on the actor adding it to the blockchain, leading to an inherent problem regarding the interface between the digital and the physical world, in application of blockchain in supply chains. Through a background study and interviews with researchers and professionals, this study proposes a set of requirements to take into account while addressing responsible sourcing via a blockchain solution. The study finds that a blockchain alone cannot ensure responsible sourcing, and further provides insight into the challenges and opportunities present in the industry and discusses the suitability of potential solutions. / Tillverkare och leverantörer inom techindustrin, som handlar med och drar nytta utav mineraler, är ofta oförmögna att genomföra djupgående företagsgranskningar i sina logistikkedjor, på grund av exempelvis kompetensbrist, vida utspridd information och kedjornas flytande natur. Att säkerställa ifall en produkt har utvunnits på ett hållbart sätt eller huruvida den innehåller konfliktmineraler är i det närmaste omöjligt. Denna studie utforskar blockkedjeteknikens potentiella roll i leverantörskedjor för mineraler, som ett kompletterande verktyg för att genomföra företagsgranskningar. Välgenomförda granskningar bör inkludera fortlöpande statusprotokoll för olika åtgärder gällande social hållbarhet, som identifierar utvinningens påverkan på mänskligt välmående. Så, hur kan en blockkedjelösning för spårbarhet i en leverantörskedja för mineraler bidra till att säkerställa hållbar utvinning? En blockkedja möjliggör spårbarhet av transaktioner genom sin oföränderliga kedjestruktur; samtidigt är kännedom om ursprunget hos en resurs avgörande för att genomföra företagsgranskningar i logistikkedjor. Ett blockkedjenätverk har potential att tillhandahålla information gällande det digitalt registrerade flödet hos en resurs, men informationens validitet gällande dess fysiska och sociala kvaliteter är fortsatt beroende av aktören som registrerar resursen på blockkedjan, vilket leder till ett ofrånkomligt problem gällande gränssnittet mellan den digitala och fysiska världen vid applicering av blockkedjor i leverantörskedjor. Utifrån en litteraturgenomgång och intervjuer med forskare och professionella, så föreslås i denna studie en kravlista att ta hänsyn till ifall blockkedjelösningar ska användas för att understödja hållbar utvinning. Studien visar att en blockkedja på egen hand ej kan säkerställa hållbar utvinning och ger vidare insikt i utmaningar och möjligheter inom industrin, samt diskuterar lämpligheten för potentiella blockkedjelösningar i dessa sammanhang.

Completeness as one of the four major dimensions of data quality is a pervasive issue in modern databases. Although data imputation has been studied extensively in the literature, most of the research is focused on inference-based approach. We propose to harness Web tables as an external data source to effectively and efficiently retrieve missing data while taking into account the inherent uncertainty and lack of veracity that they contain. Existing approaches mostly rely on standard retrieval techniques and out-of-the-box matching methods which result in a very low precision, especially when dealing with numerical data. We, therefore, propose a novel data imputation approach by applying numerical context similarity measures which results in a significant increase in the precision of the imputation procedure, by ensuring that the imputed values are of the same domain and magnitude as the local values, thus resulting in an accurate imputation. We use Dresden Web Table Corpus which is comprised of more than 125 million web tables extracted from the Common Crawl as our knowledge source. The comprehensive experimental results demonstrate that the proposed method well outperforms the default out-of-the-box retrieval approach.

Advancements in Artificial Intelligence, especially in the field of natural language processing have opened new possibilities for educational chatbots. One of these is a chatbot that can simulate a conversation between the teacher and the student for continuous learner support. In an up-scaled learning environment, teachers have less time to interact with each student individually. A resource to practice interactions with students could be a boon to alleviate this issue. In this thesis, we present a machine-learning model combined with a heuristic approach used in the creation of a chatbot. The machine learning model learns language understanding using prebuilt language representations which are fine-tuned with teacher-student conversations. The heuristic compares responses and picks the highest score for response retrieval. A data quality analysis is also performed on the teacher-student conversation dataset. For results, the best-base-cased language model performed best for text classification with a weighted F1-score of 0.70. The dataset used for the machine learning model showed consistency and completeness issues regarding labelling. The Technology Acceptance Model has been used to evaluate the model. The results of this evaluation show a high perceived ease of use, but a low perceived usefulness of the present solution. The thesis contributes with the innovative TUM (topic understanding model), an educational chatbot and an evaluation of the teacher-student chatroom corpus regarding the usage for text classification. / Teknologiska framsteg i artificiell intelligens, speciellt inom språkteknologi, har öppnat för nya möjligheter för chatbottar inom utbildningssektorn. Chatbots har sett en ökande användning i olika lärandeändamål. En av dessa är en chatbot som kan simulera en konversation mellan en lärare och en student för lärandestöd. När inlärning sker på en allt större skala, har lärare allt mindre tid att lägga individuellt på varje student. En resurs för att öva på interaktioner med studenter skulle därför kunna vara ett bra hjälpmedel. I denna masteruppsats presenteras en maskininlärnings modell kombinerad med ett heuristiskt tillvägagångsätt i skapandet av en chatbot. Maskininlärningsmodellen använder sig av färdigbyggda språkrepresentationer som är finjusterade med lärare-studentkonversationer. Heuristiken jämför svar och väljer den högsta poängen för svarshämtning. En datakvalité analys är också gjord på lärare-studentkonversations datasetet. För resultat, den BERT-baserade språkmodellen gav bäst resultat för textklassificering med en weigthed-F1- score på 0.70. Datasetet som användes för maskininlärningsmodellen visade konsistens och fullständighet problem rörande etiketter. Teknologi acceptans modellen har använts för att evaluera modellen. Resultatet av evalueringen visade hög upplevd användarvänlighet, men låg upplevd användbarhet. Detta arbete bidrar med TUM (topic understanding model), en utbildningschatbot och en evaluering av datasetet teacherstudent chatroom corpus för användning till textklassificering.

Malawi, as a country with very limited resources, needs to have educational policies in place to maximize effectiveness of the public education system. Policymakers depend on accurate data, but variations in data between sources leaves policymakers uncertain as they attempt to craft policies to address the growing educational crisis in Malawi. A desk study was performed to evaluate the policy implications of employing data from multiple sources using primary school teacher supply and demand in Malawi as an illustration. This study examined one national organization, Malawi's Ministry of Education, Science, and Technology (MoEST); three international aid and assistance organizations (IAAOs), including The Department for International Development (DIFD) from the UK, Japan International Cooperation Agency (JICA), and the United States Agency for International Development (USAID); and one global organization, The United Nations Educational, Scientific and Cultural Organization (UNSECO). The study documented differences and similarities between the data sources. Among the factors considered were the nature of each institution and the effect it could have on data collection, aggregation, analysis and reporting; the definitions used by each organization, and their implications for data use; and each organization's methods of collection, aggregation, analysis and reporting. The study found significant variations in the teacher supply and demand data presented by the five organizations, with variations of up to 333% between sources. To address this problem, it is recommended that the Government of Malawi (GoM) establish a central agency to standardize education data. Three policy scenarios are detailed, presenting the probable outcome of various actions the GoM could take regarding this recommendation.

In power system asset management, component data is crucial for decision making. This thesis mainly focuses on two aspects of asset data: data quality and data availability. The quality level of data has a great impact on the optimality of asset management decisions. The goal is to quantify the impact of data errors from a maintenance optimization perspective using random population studies. In quantitative terms, the impact of data quality can be evaluated financially and technically. The financial impact is the total maintenance cost per year of a specific scenario in a population of components, whereas the technical impact is the loss of a component's useful technical lifetime due to sub-optimal replacement time. Using Monte-Carlo simulation techniques, those impacts are analyzed in a case study of a simplified random population of independent and non-repairable components. The results show that missing data has a larger impact on cost and replacement year estimation than that of under- or over-estimated data. Additionally, depending on problem parameters, after a certain threshold of missing data probability, the estimation of cost and replacement year becomes unreliable. Thus, effective decision making for a certain population of components requires ensuring a minimum level of data quality. Data availability is another challenge that faces power system asset managers. Data can be lacking due to several factors including censoring, restricted access, or absence of data acquisition. These factors are addressed in this thesis from a decision making point of view through case studies at the operation and maintenance levels. Data censoring is handled as a data quality problem using a Monte-Carlo simulation. While the problems of restricted access and absence of data acquisition are studied using event trees and multiphysics modelling.  While the quantitative data quality problem can be abstract, and thus applicable to different types of physical assets, the data availability problem requires a case-by-case analysis to reach an effective decision making strategy. / <p>QC 20210528</p> / CPC5

During the last decade machine learning has spread rapidly in computer science and beyond, and a central issue for machine learning is data quality. This study was carried out in the intersection of business and Human-Computer Interaction, examining how an interface may be developed for crowdsourced verification of datasets.The interface is developed for efficiency and enjoyability through research on areas such as usability, information presentation models and gamification. The interface was developed iteratively, drawing from needs of potential users as well as the machine learning industry. More specifically, the process involved a literature study, expert interviews, a user survey on the Kenyan market and user tests. The study was divided into a conceptual phase and a design phase, each constituting a clearly bounded part of the study with a prototype being developed in each stage. The results of this study give an interesting insight on what usability factors are important when designing a practical tool-type mobile application, while balancing efficiency and enjoyability. The resulting novel interface indicated on a more effective performance than a conventional grid layout and is more enjoyable to use according to the users. In addition, the ‘rapid serial visual presentation’ can be deemed a well-functioning model for tool-type mobile applications which require a high amount of binary decisions on short time. The study highlights the importance of iterative, user-driven processes, allowing a new innovation or idea to merge with the needs and skills of users. The results may be of interest to anyone developing tool-type mobile applications and certainly if binary decision making on images is central. / Studien implementerar en utvecklingsprocess inspirerad av design thinking metodologi, och undersöker således gränslandet mellan ekonomi och människa-dator interaktion. Initialt undersöks affärsmöjligheter för mobila crowdsourcing applikationer i en Östafrikansk kontext, och baserat på resultaten av denna förstudie, utvecklas ett interface för mobil crowdsourcing. Interfacet ämnar hantera verifikation av bildbaserade dataset genom att samla in beslut från användarna. Målet var att designa för två huvudkriterier - njutbarhet och effektivitet, vilket uppnåddes genom efterforskning inom användbarhet, speed-reading metodologier och gamificationprinciper. Interfacet utvecklades iterativt, baserat på krav från potentiella användare, såväl som input från maskininlärningsindustrin. Mer specifikt involverade processen en litteraturstudie, expertintervjuer, en användarstudie på den Kenyanska marknaden och iterativa användartester. Den konceptuella fasen handlade om att identifiera problemet och leverera en relevant idé av hur lösningen skulle utformas. Således togs ett novellt ”Touch-Hold-Release”- interface fram. Resultaten av denna studie ger en intressant insikt i vilka usabilityfaktorer som är viktiga vid design av en praktisk arbetsinriktad applikation, samtidigt som effektivitet och njutbarhet balanseras. Det novella interfacet som tagits fram indikerar på en mer effektiv prestanda än den konventionella grid-layouten och är mer njutbar att använda enligt användarna. Dessutom kan ‘rapid serial visual presentation’ anses vara ett väl fungerande modell för arbetsinriktade mobilapplikationer som kräver stora mängder binära beslut på kort tid. Studien understryker vikten av att arbeta iterativt, med användarfokuserade processer som tillåter nya innovationer och idéer att möta användarnas faktiska behov och kunskaper. Resultaten kan vara av intresse för den som utvecklar en arbetsinriktad mobilapplikation och särskilt då binärt beslutsfattande är fundamentalt.

no / Efficient management of toxicity information as an enterprise asset is increasingly important for the chemical, pharmaceutical, cosmetics and food industries. Many organisations focus on better information organisation and reuse, in an attempt to reduce the costs of testing and manufacturing in the product development phase. Toxicity information is extracted not only from toxicity data but also from predictive models. Accurate and appropriately shared models can bring a number of benefits if we are able to make effective use of existing expertise. Although usage of existing models may provide high-impact insights into the relationships between chemical attributes and specific toxicological effects, they can also be a source of risk for incorrect decisions. Thus, there is a need to provide a framework for efficient model management. To address this gap, this paper introduces a concept of model governance, that is based upon data governance principles. We extend the data governance processes by adding procedures that allow the evaluation of model use and governance for enterprise purposes. The core aspect of model governance is model representation. We propose six rules that form the basis of a model representation schema, called Minimum Information About a QSAR Model Representation (MIAQMR). As a proof-of-concept of our model governance framework we develop a web application called Model and Data Farm (MADFARM), in which models are described by the MIAQMR-ML markup language. (C) 2013 Elsevier Ltd. All rights reserved.

Toxic compounds, such as pesticides, are routinely tested against a range of aquatic, avian and mammalian species as part of the registration process. The need for reducing dependence on animal testing has led to an increasing interest in alternative methods such as in silico modelling. The QSAR (Quantitative Structure Activity Relationship)-based models are already in use for predicting physicochemical properties, environmental fate, eco-toxicological effects, and specific biological endpoints for a wide range of chemicals. Data plays an important role in modelling QSARs and also in result analysis for toxicity testing processes. This research addresses number of issues in predictive toxicology. One issue is the problem of data quality. Although large amount of toxicity data is available from online sources, this data may contain some unreliable samples and may be defined as of low quality. Its presentation also might not be consistent throughout different sources and that makes the access, interpretation and comparison of the information difficult. To address this issue we started with detailed investigation and experimental work on DEMETRA data. The DEMETRA datasets have been produced by the EC-funded project DEMETRA. Based on the investigation, experiments and the results obtained, the author identified a number of data quality criteria in order to provide a solution for data evaluation in toxicology domain. An algorithm has also been proposed to assess data quality before modelling. Another issue considered in the thesis was the missing values in datasets for toxicology domain. Least Square Method for a paired dataset and Serial Correlation for single version dataset provided the solution for the problem in two different situations. A procedural algorithm using these two methods has been proposed in order to overcome the problem of missing values. Another issue we paid attention to in this thesis was modelling of multi-class data sets in which the severe imbalance class samples distribution exists. The imbalanced data affect the performance of classifiers during the classification process. We have shown that as long as we understand how class members are constructed in dimensional space in each cluster we can reform the distribution and provide more knowledge domain for the classifier.

It is critical for researchers to be confident of the quality of survey data. Problems with data quality often relate to measurement method design, through choices made by researchers in their creation of standardised measurement instruments. This is known to affect the way respondents interpret and respond to these instruments, and can result in substantial measurement error. Current methods for removing measurement error are post-hoc and have been shown to be problematic. This research proposes that innovations can be made through the creation of measurement methods that take respondents¿ individual cognitions into consideration, to reduce measurement error in survey data. Specifically, the aim of the study was to develop and test a measurement instrument capable of having respondents individualise their own rating-scales. A mixed methodology was employed. The qualitative phase provided insights that led to the development of the Individualised Rating-Scale Procedure (IRSP). This electronic measurement method was then tested in a large multi-group experimental study, where its measurement properties were compared to those of Likert-Type Rating-Scales (LTRSs). The survey included pre-validated psychometric constructs which provided a baseline for comparing the methods, as well as to explore whether certain individual characteristics are linked to respondent choices. Structural equation modelling was used to analyse the survey data. Whilst no strong associations were found between individual characteristics and respondent choices, the results demonstrated that the IRSP is reliable and valid. This study has produced a dynamic measurement instrument that accommodates individual-level differences, not addressed by typical fixed rating-scales.

This thesis details a process for the development of reliable metrics that could be used to assess the quality of combustion pressure measurement data - important data used in the development of internal combustion engines. The approach that was employed in this study was a model based technique, in conjunction with a simulation environment - producing data based models from a number of strategically defined measurement points. A simulation environment was used to generate error data sets, from which models of calculated result responses were built. This data was then analysed to determine the results with the best response to error stimulation. The methodology developed allows a rapid prototyping phase where newly developed result calculations may be simulated, tested and evaluated quickly and efficiently. Adopting these newly developed processes and procedures, allowed an effective evaluation of several groups of result classifications, with respect to the major sources of error encountered in typical combustion measurement procedures. In summary, the output gained from this work was that certain result groups could be stated as having an unreliable response to error simulation and could therefore be discounted quickly. These results were clearly identifiable from the data and hence, for the given errors, alternative methods to identify the error sources are proposed within this thesis. However, other results had a predictable response to certain error stimuli, hence; it was feasible to state the possibility of using these results in data quality assessment, or at least establishing any boundaries surrounding their application for this usage. Interactions in responses were also clearly visible using the model based sensitivity analysis as proposed. The output of this work provides a solid foundation of information from which further work and investigation would be feasible, in order to achieve an ultimate goal of a full set of metrics from which combustion data quality could be accurately and objectively assessed.