Contributions to personal teaching environments

Resumo

Context Students use technology in various ways to learn. The differences in their learning goals and needs make it challenging for teachers to be more engaged in designing educational tasks, assisting, and coping with their students learning. Hence, the continuous effort to provide novel solutions to the educational field makes it necessary to engage those technologies to improve the experiences gained from education. This thesis addresses Personal Teaching Environments (PTEs), which help teachers in fulfilling their educational tasks. During the time they spend teaching a course, teachers usually use one or more teaching methods. It varies depending on the design of the course, the content that needs to be illustrated, and the characteristics of the students. To accomplish the objective of using the chosen methodology, the teacher must design and effectively manage some activities, tools, and data sources for each methodology. The state-of-the-art focus on investigates how PTEs have been applied in educational settings, and presents the concepts of PTEs, applications, systems, and techniques that help teachers in their educational tasks from planning to assessing (including feedback). The information extracted from the state-of-the-art gave a useful overview of PTEs, such as new methods, possible advantages, and drawbacks when using PTEs. Although several studies have reported on the use of PTEs, prior research falls short of addressing the structure of PTEs and how it works. Hence, a lot of tools, activities, teaching methods, and data that can help teachers in their tasks are available already but teachers need to know what is the more efficient way to work with them, and what guidance they need to build their PTE to facilitate delivering their teaching. Subsequently, one of the goals of this study is to define the concept of PTE, its structure, and its components (tools, connections and activities, data sources, and teaching methods). As well, to investigate the most popular non-traditional teaching methods, and to relate PTEs to these methods with applications’ examples. Therefore, this thesis is addressing some of the non-traditional teaching methods' advantages, frameworks, and successful implementation stories, and categorizes the tool types of the selected non-traditional teaching methods to serve PTE applications. A review was conducted to achieve these goals, and a total of 93 peer-reviewed articles pertaining to the most frequently studied non-traditional teaching methods were comprehensively studied and analyzed. The analysis resulted in practical guidelines, including the benefits and tool types of the five studied non-traditional teaching methods (flipped classroom, problem-based learning, gamification, case study, and social media-centered). Based on the results, significant examples for teachers who aim to use one or more of those non-traditional teaching methods were established, through the adoption and utilization of the PTE applications. Non-traditional teaching methods, in contrast to traditional methods, are based on interaction, questioning, explaining, demonstrating, and collaboration techniques. These teaching methods include lecturers demonstrating, using games to help students solve problems, and watching pre-recorded videos of demonstrations. Learning outside of the classroom takes place in a variety of non-traditional contexts. Additionally, this offers students a new way to interact with the material they're studying. However, there are numerous non-traditional teaching methods, and the teacher could not know which one is proper for a specific course. Just as important, each method may use different tools, or use the same tool in a different way. As a result, teachers need non-intrusive and automatic ways to get suggested new ways of teaching (e.g., non-traditional teaching methods). Likewise, tools to better follow the educational process and deliver the course structure effectively. And that could help the teachers with one of the most challenging tasks they have, assisting their students' learning of course concepts, and determining if they have met the course's objectives. Pre-class, in-class, and post-class activities are typically part of the teaching methods. This work will help the teacher choose a suitable method for a particular course and tools that can fit with this method as the goal of this work is the non-traditional teaching methods and their tools. Depending on a teacher's personality and expertise with the tools, different teachers may use tools differently. Teachers can post updates and notifications on their Facebook pages. Facebook groups can be used by others to organize discussions and stream live lectures. Teachers can post reminders about assignment due dates and useful links to practical quizzes or resource downloads on Twitter. Additionally, they can use one of the many accessible platforms, such as WordPress, Wix, or Blogger, to start discussions, create a hashtag for chats, and start a class blog for discussions. A blog can be used to convey information about assignments, announcements, and resources for the class and to assign blog posts to students as essays. One of the most popular and effective approaches in recommendation systems is collaborative filtering, which involves recommending items to users who are currently active and have previously shown interest in or purchased them. To make recommendations for items that the active user has not yet rated or viewed, collaborative filtering compares the current user ratings for things like movies or products to those of users who are similar to them (nearest neighbours). To enhance the purchase of goods or services, recommendation systems have been developed and adopted in numerous applications; nevertheless, personal teaching settings have not seen a similar adoption. For the successful adoption of PTEs, the need for a recommender system to efficiently perform non-traditional teaching methods within a personal teaching environment becomes even more crucial. The two most popular types of this approach are user-based and item-based approaches. Another typical recommendation system approach is content-based filtering, which recommends an item to users based on their interests and the item description. Among other things, content-based recommendation systems could be used to suggest websites, news articles, restaurants, TV shows, and commercially available items. The main disadvantage of this approach is that the recommended items by the algorithm are probably going to be quite similar to those that the active user has already chosen. Machine learning will be extremely important in a variety of future applications. By enabling computers to detect and learn from the real world, as well as improve performance on particular tasks depending on this new information, machine learning (ML) simulates human learning. A branch of artificial intelligence called machine learning aims to employ smart software to give computers the ability to carry out tasks with skill. Statistical learning methods are the foundation of intelligent software that is used to generate machine intelligence. The field of machine learning must be connected to the database field since machine learning algorithms need data to learn. Hence, Learning is the process of gaining knowledge. Humans learn from their experiences in a natural way because of their capacity for reasoning. Computers, on the other hand, follow algorithms to learn instead of thinking. Recently, recommender systems have paid a lot of attention to deep learning. There have been many combinations of deep learning, collaborative recommendation, and content-based recommendation. One of the three most crucial recommendation approaches, hybrid recommendation, rarely works with deep learning. Furthermore, the majority of current deep hybrid models just combine two simple recommendation algorithms in post-fusion, leaving room for the discovery of more effective combinations. The impact of deep learning is equally ubiquitous, and recent studies have shown that these methods are useful for recommender systems and information retrieval. Deep learning in recommender systems is undoubtedly a growing field. The ability to learn deep representations from data, which requires learning several levels of representations and concepts, defines deep learning. For practical reasons, we consider any neural differentiable architecture as deep learning if it maximizes a differentiable objective function using a variant of stochastic gradient descent (SGD). There are now additional opportunities to improve recommender performance thanks to deep learning's significant influence on recommendation structures. The capacity of recently developed deep learning-based recommender systems to overcome the drawbacks of traditional models and generate high-quality recommendations has drawn a lot of interest. Deep learning enables the coding of increasingly complex abstractions as data representations at higher levels, as well as the successful capture of non-linear and non-trivial user-item relationships. Furthermore, it uses a range of data sources, including as contextual, textual, and visual information, to capture the intricate relationships that exist within the data itself. As a response to these challenges, this thesis proposes the use of neural and collaborative filtering techniques to build a system that could recommend non-traditional teaching methodologies and tools for the teacher. The aim is to use this recommender system called Adaptive Neural Collaborative Recommender (ANCoR) in a personal teaching environment (PTE), to assist the learning process while using non-traditional teaching methods. This study enhanced the NeuMF framework and proposed ANCoR, which takes into account an extra inherent tool type factor. The proposed framework has good characteristics that can assist and guide teachers in their work, particularly by recommending them to select suitable non-traditional teaching methods and tools. Systems that make recommendations to users rely on data collected from explicit or implicit user feedback. This work proposed a questionnaire that will assist in gathering the necessary data because there isn't any data currently available to offer methods and tools for teachers. The purpose of the questionnaire is to provide a reliable and valid dataset that could be used with ANCoR. After then, the data gathered from the questionnaire will be used to train ANCoR to recommend teaching methods, activities, and tools for new users to use in their teaching. In order to generate multiple teacher profiles and predict their teaching style based on their responses, the datasets include data that was collected from the teachers through the questionnaire. Hence, a significant questionnaire for the teachers who aim to use one or more of those non-traditional teaching methods was established, through the adoption and utilization of a PTE. The questionnaire passed the validation and reliability test, which means the sample size is sufficient. The data extracted from the questionnaire were used to train the recommendation system. The results show numerical and visual improvements when the proposed model is used compared to K-Nearest Neighbors (KNN) baseline. Afterward, the teacher’s satisfaction with ANCoR was tested. The satisfaction questionnaire results with participants indicate the positive role of ANCoR in recommending non-traditional teaching methodologies and tools. Finally, an attempt has been made to identify future research paths. Objectives and methodology This work aims to introduce the concept of PTE and provide a method for creating PTE. a greater effort is made to illustrate several frequently used non-traditional teaching methods in this study. Additionally, the study focused on outlining the activities, tool types, advantages, frameworks, processes, and examples of successful implementation for the methods chosen in the thesis. Examples of PTEs were given that have helped some nontraditional teaching methods develop. Following that, a recommender system was used to ease the process of choosing nontraditional teaching methods and their components while using the PTE system. In order to match each learner with the educational resources that are most appropriate for his or her profile or needs, recommendation systems (RS) have been utilized and adapted in the field of education [1]. In an educational setting, a recommender system is software that tries to intelligently recommend courses to a teacher or student based on two basic filtering techniques: the first is focused on the user's preferences, and the second is focused on the preferences of the group to which the user belongs [2]. In the current study, This recommendation might be an online activity, such as using a particular tool during in-class activities or adopting a non-traditional teaching method (e.g., Flipped learning) to teach a course. However, despite being adapted and utilized in numerous applications to encourage the purchase of goods or services, recommendation systems have not been similarly applied in personal teaching environments that support teachers in carrying out their pedagogical responsibilities. Data was another issue this study had to deal with since recommendation systems need a lot of data to function and make good recommendations. To gather the essential data, We proposed a questionnaire that would facilitate the collection of that data. The adaptive neural collaborative recommender (ANCoR) framework that this study proposes takes into account the various personalities of different teachers. Similar to this, several teaching methods are used to identify each teacher's preferred method of teaching in order to personalize and customize the experience. The dataset contains information gathered from the teachers by a questionnaire, in order to anticipate their teaching style based on their answers and to create different teacher profiles. The purpose of the survey was to create a reliable dataset for the recommender system. For a particular course, the system has the ability to provide nontraditional teaching methods and tools. It aims to create profiles for various teachers in order to generalize the most prevalent practices toward a future activity. The system functions as a personal assistant for teachers, assisting them in selecting the suitable non-traditional teaching methodology, its activities, tools, and how to use the selected tools to achieve the activities goals. Afterward, a satisfaction questionnaire was administered to measure the participating teachers’ overall satisfaction with the study concerning the interaction, the instruction, and the used technology. The teachers’ participation was voluntary and anonymous to enable them to express their opinions without any pressure. Contributions In this thesis, the concept personal teaching environment (PTE) was defined. This contribution's goal was to provide teachers PTE components (tools, connections, and activities, as well as data sources) that would make it easier for them to complete their educational responsibilities. Moreover, this study helped to integrate nontraditional teaching methods within PTEs' framework. Hence, this thesis contributed to providing the types of tools needed to carry out those methods' activities, examples of those tools, and explanations of how to use those tools when using PTE applications. In order to improve the quality of recommendations in teaching, this work has experimented with a hybrid approach that combines content-based and collaborative filtering approaches. This study proposed constructing a neural collaborative filtering recommendation system for use in personal teaching environments that would take into account nontraditional teaching methods as well as the tools and activities that would be used in each method. A real-world dataset gathered from a conducted questionnaire served as the basis for the experiment. The findings demonstrated that ANCoR is capable of automatically extracting features and providing useful recommendations. Moreover, I compared the model performance with the classic recommender system methods. The main contributions of this work are: 1. Defined the structure of the PTE, its framework, and its components while applying it in informal learning. 2. Provided guidelines and examples for the creation of PTEs while using non-traditional teaching methods. 3. Improved the Neural matrix factorization (NeuMF) framework and propose an Adaptive Neural Collaborative Recommender (ANCoR). Conclusion and Future work This study defined the concept of Personal Teaching Environment (PTE) that can facilitate the teacher’s fulfillment of their educational tasks. This study summarizes the technical approaches for establishing PTEs, their concepts, framework, and its components (tools, connections and activities, and data sources). Furthermore, this study investigates some of the popular non-traditional teaching methodologies and its frameworks, advantages, and successful implementation stories. This study further discussed how to build examples for these methodologies using PTEs applications. The study concluded that a methodology was formed for the creation of PTEs with several tools based on the specific needs of each teacher. The study focused on how to assist the teacher in delivering several educational tasks, while using one of the five identified nontraditional teaching methodologies. Each of these methods has its activities, tool types, and tools. The study established significant examples to assist the teachers who aim to use one or more of these non-traditional teaching methodologies, through the use of the PTE applications. The study also gave examples of PTE applications through tools types, examples of tools and examples of use of such tools according to the selected non-traditional teaching methodology. These components need to be included in the PTE for successful implementation. This involves more the teacher in the students' formal and informal learning, to monitor and control the quality of learning, and also to improve the teachers' professional development. The study discussed how to help teachers organize their work more efficiently. The study presents a personal teaching environment (PTE) system architecture and examine the ability of the proposed ANCoR recommendation system to provide non-traditional teaching methodologies recommendations to teachers. The system has 3 components: graphical user interface component, recommendation algorithms component, and backend component. Those components work together to manage the user-recommender system interaction, configure the algorithm, and store the data in the database, respectively. This study established a significant questionnaire for the teachers that is able to specify teachers teaching preferences that can help to choose the appropriate non-traditional teaching methodologies, in the PTE context. The reliability of the questionnaire was tested with the Cronbach’s alpha using the Statistical Package for the Social Sciences (SPSS) software version 25. The questionnaire aimed to build a reliable dataset to be used with the recommender system. The collected data improve the predictive capability of ANCoR, and also alleviates the problem of cold start. Afterward, this study improved the NeuMF framework and propose ANCoR, which additionally considers one inherent factor of tool type. The proposed framework presents good characteristics to assist and guide teachers in their work, especially recommending teachers to choose appropriate non-traditional teaching methodologies and tools by providing relevant recommendations. This study conducted an experiment on a real-world dataset collected from the questionnaire. The results indicate that ANCoR can extract features automatically and make appropriate recommendations. Therefore, compared with the baseline methods, our proposed model can achieve higher accuracy and recommendation effects. The items recommended by our model are more suitable for users. Finally, this study examined the teachers’ satisfaction with the system. The results are generally positive, with low refusal levels. The study contains some limitations. The first one observed is that the keywords used to collect an inclusive database may not be extensive enough. However, the study is confident about the number of scholarly work collected. Secondly, this study identifies some of nontraditional teaching methods to adapt them in a personal teaching environment. However, other non-traditional teaching methodologies were not studied. Future studies could investigate the use of the PTE with other non-traditional teaching methodologies not targeted in this study to identify the advantages and impact of using such techniques for teaching. Furthermore, the study considers to test the implementation of the ANCoR PTE recommender system in more running courses to analyze its impact on the overall educational experience.