The teaching and learning process of computer programming is a complex task which requires a lot of practice and creativity. Usually, there are numerous solutions to the same problem. Therefore, the student needs that his solutions are evaluated quickly for a faster and effective learning. To face these challenges, teachers and students can rely on resources from the evolution of Information and Communication Technology. Virtual learning environments and online judge systems are attractive alternatives used in this context. This work presents a unified approach to specify and check source code restrictions supported by a static analyzer. Although current tools are able to indicate if the program produced the expected output from a given input, not all are able to determine if the student used (or not) a given programming language construct, such as creating a function and using it in the program. Among those that are capable, there are problems that were solved in the approach proposed in this work, such as: ease of use, unified approach and degree of flexibility. In addition, this work presents an analysis of the database of The Huxley with the purpose of discovering the main restrictions of source code used by the teachers and attended by the students. This analysis was done based on data obtained from the use of the developed static analyzer and in conjunction with a survey applied to the teachers of introduction to programming with the purpose of knowing the main restrictions that would be used by them if they had a tool to specify and check restrictions. / O processo de ensino e aprendizagem da programação de computadores é uma tarefa complexa que requer bastante prática e criatividade. Geralmente há inúmeras soluções para um mesmo problema. Por isso, o aluno precisa que suas soluções sejam avaliadas rapidamente visando um aprendizado mais ágil e eficaz. Para enfrentar esses desafios, os professores e alunos podem contar com recursos provenientes da evolução da Tecnologia da Informação e Comunicação. Os ambientes de aprendizagem virtual e os sistemas de juiz on-line são alternativas atrativas utilizadas nesse contexto. Este trabalho apresenta uma abordagem unificada de especificação e checagem de restrições de código-fonte apoiada por um analisador estático. Apesar das ferramentas atuais serem capazes de indicar se o programa produziu a saída esperada a partir de uma entrada fornecida, nem todas são capazes de determinar se o aluno utilizou (ou não) determinada construção de linguagem de programação, como por exemplo criar uma função e utilizá-la no programa. Entre as que são capazes, existem problemas que foram sanados na abordagem proposta neste trabalho, como: facilidade de uso, abordagem unificada e grau de flexibilidade. Além disto, este trabalho conta com uma análise da base de dados do The Huxley com o objetivo de descobrir quais são as principais restrições de código-fonte utilizadas pelos professores e atendidas pelos alunos. Esta análise foi feita com os dados obtidos da aplicação do analisador estático de código-fonte desenvolvido e em conjunto com um survey aplicado aos professores de introdução à programação com o propósito de conhecer as principais restrições que seriam utilizadas por eles se possuíssem uma ferramenta de especificação e checagem de restrições. / São Cristóvão, SE

The Swedish government has decided to add programming to the Swedish curriculum to strengthen students’ digital skills. The teachers who will teach programming do not always know programming themselves. Because of that, KTH Royal Institute of Technology researchers are planning to start an initiative of creating a Massive Open Online Course (MOOC) in fundamental programming in Swedish for those teachers. Interpreting error messages given by the compiler is one problem with learning programming. An aim of this study is to lower that threshold. The study seeks to identify common misconceptions about programming among novice programmers in order to design a static analyser that investigates code written by students and provides formative feedback to help students in their own learning process. The study combines the constructivist theory of learning with views about formative assessment in order to automate the process that is usually done by a teacher or teaching assistant when assessing code. A phenomenographic study is done in order to identify teachers’ perceptions of common misconceptions about programming by interviewing five active teachers at KTH. The results of that study are used to construct code examples that correspond to these misconceptions. Those results lead to the design and implementation of a software that detects these problems in code. That software is evaluated using a larger set of test data, consisting of 77 errors divided into five larger programs, inserted by independent individuals. From the initial study, five categories of misconceptions are given. Of the 77 errors, the majority are correctly positioned and almost all are given a good hint about the position. About a quarter of the errors are parse errors, which never reach the analysing part of the software that demands the program to be parsable. The study shows that we have succeeded both in designing and implementing a software that detects the identified misconceptions with good results. In the context of a MOOC, the software might require an extension with a more advanced parser and also dynamic analysis to be able to test the correctness of the students’ programs. The software is limited to handle the language Javascriptish, which is a subset of JavaScript. / För att stärka den digitala kompetensen bland svenska grundskole och gymnasieelever har programmering lagts in i läroplanen. De lärare som är tänkta att hålla i denna programmeringsundervisning har själva inte alltid tillräckliga kunskaper i programmering, varpå ett initiativ för att hålla en storskalig, öppen och internetbaserad kurs (MOOC) på svenska har startats av forskare på Kungliga Tekniska Högskolan (KTH). En av svårigheterna med att lära sig programmering är att förstå meddelandena som kompilatorn ger. Ett mål med denna studie är att sänka denna tröskel för studenten. Denna studie ämnar identifiera vanliga missuppfattningar om programmering hos nybörjare, för att designa en programvara som utför statisk analys av kod skriven av studenter och därefter ge återkoppling kring dessa missuppfattningar. Studien kombinerar den konstruktivistiska teorin om lärande med tankar om formativ bedömning för att automatisera den process som vanligtvis görs av en lärare eller lärarassistent vid bedömning av kod. En fenomenografisk studie görs för att identifiera lärares uppfattningar om vanliga missuppfattningar kring programmering genom att intervjua fem verksamma lärare på KTH. Resultatet från den studien används sedan för att konstruera kodexempel som belyser de identifierade missuppfattningarna. Därefter designas och implementeras en mjukvara som sedan evalueras genom att analysera totalt 77 fel, konstruerade av oberoende individer, uppdelade på fem större program. Den initiala studien resulterar i fem olika kategorier av missuppfattningar. Av de 77 evaluerade felen ger majoriteten en korrekt positionsangivelse och nästan alla ger en god indikation över var felet ligger. Omkring en fjärdedel av felen är parsningsfel, vilka aldrig når huvudmjukvaran som kräver att programmet är parsningsbart. Studien visar att vi lyckas designa och implementera en programvara som med goda resultat upptäcker vanliga missuppfattningar kring programmering hos nybörjare, baserat på det teoretiska ramverket. I kontexten av en kurs på internet kan programvaran behöva utvecklas med en mer avancerad syntaxanalys (eng. parser) samt lägga till dynamisk analys av program för att även kunna testa programmens korrekthet. Programvaran är begränsad till att analysera kod skriven i språket Javascriptish, vilken är en delmängd till JavaScript.

When students use an online eTextbook with content and interactive graded exercises, they often display aspects of two types of behavior: credit-seeking, and knowledge-seeking. Any given student might behave to some degree in either way in a given assignment. In this work, we look at multiple aspects of detecting the degree to which either behavior is taking place, and investigate relationships to student performance. In particular, we focus on an eTextbook used for teaching Formal Languages, an advanced computer science course. This eTextbook is using Programmed Instruction (PI) framesets to deliver the material. We take two approaches to analyze session interactions in order to detect credit-seeking incidents. We first start with a coarse-grained approach by presenting an unsupervised model that clusters the behavior in the work sessions based on the sequence of different interactions that happens during them. Then we perform a fine-grained analysis where we consider the type of each question in the frameset, which can be a multi-choice, single-choice, or T/F question. We show that credit-seeking behavior is negatively affecting the learning outcome of the students. We also find that the type of the PI frame is a key factor in drawing students more into the credit-seeking behavior to finish the PI framesets quickly. We implement three machine learning models that predict students' midterm and overall semester grades based on their amount of credit-seeking behavior on the PI framesets. Finally, we provide a semisupervised learning model to aid in the work session labeling process. / Master of Science / Students frequently exhibit features of two types of behavior when using an online eTextbook with content and interactive graded exercises: credit-seeking and knowledge-seeking. When solving homework or studying a material, students can behave in either manner to some extent. In this research, we study links between student performance and different elements of recognizing the degree to which either behavior is occurring. We concentrate on an eTextbook used to teach an advanced computer science course, Formal Languages and Automata, using a teaching paradigm called Programmed Instruction (PI). In order to detect credit-seeking instances, we use two ways to study students' behavior in the Programmed Instruction sessions. We begin with a coarse-grained approach by building a model that can categorize work sessions into two groups based on the interactions that occur throughout them. Then we do a fine-grained analysis in which we analyze the question types in the framesets and their effect on the students' behavior. We show that credit-seeking behavior has a negative effect on students' learning outcomes. We discovered that the PI frame type is an important factor in enticing students to engage in credit-seeking behavior in an attempt to finish PI framesets fast. Finally, we present three predictive models that can forecast the students' midterm and total semester grades based on their credit-seeking behavior on the Programmed Instruction framesets.

The present study examines ChatGPT-3's ability to generate code solutions for introductory programming courses in computer science and the potential implications for academic integrity. An experiment was conducted where ChatGPT was tested on programming problems from Kattis, an automatic software grading tool for computer programs, used in higher education. The results showed that ChatGPT independently could solve 19 out of 127 programming tasks assessed by Kattis. The study’s results also show that ChatGPT could generate accurate code solutions for simple problems on Kattis but encounters difficulties with more complex programming tasks. A qualitative follow up investigation was also carried out. To provide comments on methodology and discuss cheating in higher education concerning programming courses the two teachers were interviewed. The Kattis system is considered to have useful features for preventing cheating, such as hidden test cases, but it also has limitations in detecting AI-generated code. The report concludes by discussing the implications for various stakeholders, including teachers, students, and researchers. / Studien undersöker ChatGPT-3:s förmåga att generera kodlösningar för grundläggande programmeringskurser inom datavetenskap och de potentiella konsekvenserna för akademisk integritet. Ett experiment utfördes där ChatGPT testades med programmeringsproblem från Kattis, ett automatiskt rättningsverktyg för datorprogram som används inom högre utbildning. Resultaten visade att ChatGPT självständigt löste 19 av 127 programmeringsuppgifter som bedömdes av Kattis. Studien konstaterar att ChatGPT kan generera korrekta kodlösningar för problem med låg svårighetsgrad enligt Kattis, men stöter på svårigheter med mer komplexa programmeringsuppgifter. En kvalitativ uppföljningsundersökning genomfördes även där två lärare från KTH intervjuades för att ge sina kommentarer om metodvalet och diskutera fusket inom högre utbildning när det gäller programmeringskurser. Kattis-systemet anses ha användbara funktioner för att förhindra fusk, såsom dolda testfall, men har också begränsningar när det gäller att upptäcka AI-genererad kod. Rapporten avslutas med att diskutera implikationerna för olika intressenter, inklusive lärare, studenter och forskare.

Many older adults desire to maintain their quality of life by living and aging independently in their own homes. However, it is difficult for older adults to notice and track the subtle changes in their own abilities because these abilities can change gradually over a long period of time. Technology in the form of ubiquitous sensors embedded in objects in the home can play a role in keeping track of the functional abilities of individuals unobtrusively, objectively, and continuously over a long period of time. This work introduces a sensing technique called “task-based embedded assessment” that monitors how well specific tasks important for independence are carried out using everyday objects found in the home with which individuals regularly interact. Following formative studies on the information needs of older adults and their caregivers, a sensing system called “dwellSense” that can monitor, assess, and provide feedback about how well individuals complete tasks, such as taking medications, using the phone, and making coffee, was designed, built, and evaluated. Multiple longterm (over 10 months) field deployments of dwellSense were used to investigate how the data collected from the system could support greater self-awareness of abilities and intentions to improve in task performance. Presenting and reflecting on data from ubiquitous sensing systems such as dwellSense is challenging because it is both highly dimensional as well as large in volume, particularly if it is collected over a long period of time. Thus, this work also investigates the time dimension of reflection and has identified that real-time feedback is particularly useful for supporting behavior change, and longer-term trended feedback is useful for greater awareness of abilities. Traditional forms of assessing the functional abilities of individuals tend to be either biased, lacking ecological validity, infrequent, or expensive to conduct. An automated sensor-based approach for assessment is compared to traditional performance testing by a trained clinician and found to match well with clinician-generated ratings that are objective, frequent, and ecologically valid. The contributions from this thesis not only advance the state of the art for maintaining quality of life and care for older adults, but also provide the foundations for designing personal sensing systems that aim to assess an individual’s abilities and support behaviors through the feedback of objective, timely sensed information.