Research Methods in Computer Science

WKKim 08/16/2024

Overview

Explore the varying ways in which research is conceptualised, designed, conducted, analysed and how the results might be disseminated. The Research Methods module provides you an in depth understanding on how research fits into the context of the Computer Science discipline and the Computing profession. This module walks you through the various stages of research, from choosing a research topic, developing the proposal, conducting research using the standard approaches for research design and data collection, managing the research process, including evaluating and analysing the outcomes and presenting or disseminating the results. Throughout the module, we look at aspects related to Legal, Social, Ethical and Professional issues, especially important given the global environment in which we work.

Engage in a deep and systematic exploration of the nature of computing research projects, including those that offer new opportunities for innovative research
Examine how the various phases of conducting a research project within the Computer Science discipline includes considerations for Legal, Social, Ethical and Professional Issues (LSEPI), including the use of codes of practice
Discuss how a research project proposal is developed from an initial idea, a review of the currently available literature, identification of aims, objectives, goals and outcomes, risks, potential benefits and outcomes and the potential for contributions to academic and commercial research communities
Explore and enhance the ability to critically review the literature, develop an in-depth understanding related to managing research projects utilizing tools and techniquesDemonstrate an application of the key characteristics related to the nature of strategic computing projects that harness recent development within the Computer Science discipline and profession
Identify the various stages of a research project, conducted within the academic or commercial environments
Develop a research project proposal, for purposes of presentation and preparation for approval
Evaluate research methodologies and methods for appropriate selection, given the research context
Demonstrate an in-depth contextual understand of the fundamental concepts associated with Legal, Social, Ethical and Professional issues within the context of conducting research in the Computer Science discipline
Identify the ways in which research project results can be disseminated, through a variety of venues and contexts

Topic

Exploring and Applying Research Methodologies and Methods suitable for research conducted within the Computer Science discipline

The choice of a research methodology and related research methods reflect not only the context of the research but also include decisions related to the way the research is initially framed through the hypothesis statements or research questions.

This discussion invites you to research and reflect on the different research methodologies and related research methods, in light of a research idea or topic that holds your interest.

Conduct an initial review of research articles related to the research idea or topic you've chosen, reviewing the types of research approaches used by the researchers, their choice of methodology and methods and why these approaches were chosen.
Reflect on your own personal experiences with Computer Science projects that might have required you to investigate the feasibility of a particular technology or application of a technology within a context, from a 'problem solving' perspective.

Post

Several research methodologies have been used in AI and machine learning especially in creating efficient and reliable models used in making decisions in various fields. Two highly informative researches help in distinguishing between various methodological approaches and their use in AI and ML fields.

In their article “Systematic literature review of machine learning methods used in the analysis of real-world data for patient-provider decision making,” Brnabic and Hess (2021) explores the use of ML approaches in real-world data used to inform patient and provider decision. In their assessment of 34 publications, they identified a plethora of approaches, such as decision tree and random forest. Internal validation was widely employed, but external validation was used selectively, suggesting that there may be a lack of broader applicability of the findings. The authors suggest the utilisation of various machine learning techniques and elaborate validation procedures to enhance evidence supporting patient care actions.

In contrast, the method formulated by Zaki Ahmed and Diaz (2022) in their article “A Methodology for Machine-Learning Content Analysis to Define the Key Labels in the Titles of Online Customer Reviews with the Rating Evaluation” emphasises content analysis of online customer reviews in the context of airline services. Their unique approach entails the conversion of titles of qualitative reviews to quantitative forms, which can be subjected to regression analysis. The methodology is useful in determining important labels that affects the general ratings and may be assisted in improving customer satisfaction in the airline industry.

The critical discussion of these methodologies exposes their strengths and weaknesses. Although decision tree and random forest techniques provide several methods to handle large amount of data, there are some restrictions on generalisation of the results mainly because of lack of external testing in some of the works. In the same way, while turning qualitative data into quantitative variables helps to transform sentiment analysis, this process might distort complex meanings of words and overall emotions stated in customers’ reviews.

The methodologies and methods presented in these research articles resonate with my study on creating an AI model to predict students’ academic performance in learning environments (Jiao et al., 2022). When applied to the educational domain, decision tree and random forest can be useful in predicting performance of each student. In addition, quantitative conversion from qualitative feedback of students can help enhance the data gathered on student satisfaction and engagement level to design more effective educational interventions (Rajabalee & Santally 2021).

In conclusion, these articles highlight the need to adopt a more rigorous research methodologies in AI and machine learning urging researchers in this field to employ different techniques and come up with better validation methods that make their findings dependable and usable in real-world decision-making scenarios.

References

Brnabic, A., & Hess, L. M. (2021). Systematic literature review of machine learning methods used in the analysis of real-world data for patient-provider decision making. BMC medical informatics and decision making, 21, 1-19.
Jiao, P., Ouyang, F., Zhang, Q., & Alavi, A. H. (2022). Artificial intelligence-enabled prediction model of student academic performance in online engineering education. Artificial Intelligence Review, 55(8), 6321-6344.
Rajabalee, Y. B., & Santally, M. I. (2021). Learner satisfaction, engagement and performances in an online module: Implications for institutional e-learning policy. Education and Information Technologies, 26(3), 2623-2656.
Zaki Ahmed, A., & Rodríguez Díaz, M. (2022). A Methodology for Machine-Learning Content Analysis to Define the Key Labels in the Titles of Online Customer Reviews with the Rating Evaluation. Sustainability, 14(15), 9183.

Summary

The selection of research strategy and relevant methods in Computer Science is informed by the context and goals of the research. This discussion revolves around methodologies and methods that are relevant to the analysis of ethics in AI systems, based on two articles. The research topic I have chosen is "Ethical Considerations in AI Development." The autonomy and the sophistication of the AI systems implemented put into question the ethical concerns of implementing such technologies.

The study by Abazi-Bexheti et al. (2020) “Word Cloud Analytics of the Computer Science Research Publications’ Titles over the Past Half Century,” employed the DBLP database as the data basis to discover trends in the Computer Science research of the past fifty years. These trends were supported with the help of text visualization tools, more specifically word clouds. This quantitative approach was selected because of the amount of data and to observe trends over a long period of time. Word clouds offered an easy and quick means of presenting structures that otherwise would have taken time to explain and comprehend (Gotterbarn et al. 2018). This approach is suitable for my study on AI ethics as it presents a straightforward way of mapping trends and foci over the discourse period.

In contrast, Buchanan et al. (2011) expand on the complex relationship between computer science research, specifically in security, and the ethical treatment of people. As highlighted in their article, IRBs are not easy in the process of adopting these norms that are mainly developed for biomedical and social sciences more so in computer science ethical context. In the current study, the authors use qualitative research methods that include literature, case studies, and conceptual analysis to explain how basic ethical principles of risk management, the protection of confidentiality, and the acquisition of informed consent differ in computer science fields where subjects are usually data or systems and not individuals. This methodology not only reveals the complex ethical issues at stake in computer science research but also provides modifications to conventional ethical approaches that are specific to AI ethics.

Regarding the research topic relating to ethical considerations in AI development, I shall use both quantitative and qualitative approach in data collection and analysis. First, a quantitative research approach consisting of data mining from academic databases and word frequency analysis of the retrieved documents and publications will determine some of the many ethical issues as well as general trends over time (Owen et al. 2013). Second, a qualitative analysis comprising of a literature review of various ethical guidelines and significant conceptual analysis to create a framework for ethical AI engineering. This method will result in the creation of a body of work that perhaps can inform future research and practical applications in the field of AI ethics.

References

Abazi-Bexheti, L., Kadriu, A., & Apostolova, M. (2020, September). Word Cloud Analytics of the Computer Science Research Publications’ Titles over the Past Half Century. In 2020 43rd International Convention on Information, Communication and Electronic Technology (MIPRO) (pp. 887-892). IEEE.
Buchanan, E., Aycock, J., Dexter, S., Dittrich, D., & Hvizdak, E. (2011). Computer science security research and human subjects: Emerging considerations for research ethics boards. Journal of Empirical Research on Human Research Ethics, 6(2), 71-83.
Gotterbarn, D. W., Brinkman, B., Flick, C., Kirkpatrick, M. S., Miller, K., Vazansky, K., & Wolf, M. J. (2018). ACM code of ethics and professional conduct. https://dora.dmu.ac.uk/bitstream/handle/2086/16422/acm-code-of-ethics-and-professional-conduct.pdf?sequence=1
Owen, R., Stilgoe, J., Macnaghten, P., Gorman, M., Fisher, E., & Guston, D. (2013). A framework for responsible innovation. Responsible innovation: managing the responsible emergence of science and innovation in society, 27-50. https://www.researchgate.net/profile/Erik-Fisher-3/publication/263662345_A_Framework_for_Responsible_Innovation/links/5d90e34d458515202b74859d/A-Framework-for-Responsible-Innovation.pdf

Topic

The purpose of this summative assessment is to provide an opportunity to conduct a systematic review of the available literature and select 5 academic articles from the search that are representative of topic that is currently considered ‘cutting edge’ in the Computer Science discipline. Drawing upon your own experiences and the literature you have already read, while completing other modules, along with the research articles reviewed in this module, identify a topic for which there is a need for ongoing research, one in which you are particularly interested, especially if you are proceeding to the Dissertation. Example topics might include Deep Learning, Advanced Algorithmic applications involving the use of Blockchain or the adaption of Biotechnology.

The Literature Review Assignment should have the following structure

Begin with an Introduction, where you present the purpose of the paper, including a Thesis statement. 
Present each of the 5 sources you include in the Literature Review.
For each source:
- Present a full citation using the Harvard Referencing Standard and place it ahead of the review.
- In about 350 words, critically analyse the academic source, presenting the key themes and discoveries and assess its merit in terms of providing information that offers support to and informs the current state of ‘cutting edge’ research on the topic.
Conclude your Literature Review Assessment with a short summary that identifies key ideas presented within the paper

The total length of the Literature Review Assessment is between 2000-2500 words. All articles must be from authoritative sources and recent (within three years).  The Literature Review Assessment’s primary goal is to provide you with the opportunity to critically examine recent research that would support extending existing research in a new manner. 

Deep learning has been the most transformative subfield in artificial intelli-gence (AI) as it has given the field another push in fields like computer vision, natural language processing, and healthcare. Being a type of neural network archi-tecture that closely mimics the human brain’s structure, deep learning has been found to be highly adept in tasks that involve complicated pattern recognition and decision making. This paper aims to outline major trends and advances on deep learning, its difficulties, and potential opportunities for further evolution based on the analysis of five foundational papers. These sources give an outline of the theo-retical notions, landmark achievements, and current research in the field. The the-sis statement guiding this review is: Deep learning, through the advancement of its architectures and strategies, has been progressing as a highly influential area in AI and improving the ability to complete intricate operations, but it is not devoid of limitations, including interpretability, dependence on data, and flexibility in han-dling various tasks.

Citation: Bengio, Y., LeCun, Y., and Hinton, G., 2021. Deep Learning for AI. Communications of the ACM, 64(7), pp.58-65.

Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, the laureates of the 2018 ACM A. M. Turing Award, describe the fundamentals and remarkable achievements in deep learning. The authors note that deep learning has achieved growth because of learning higher-level representations necessary for solving intri-cate tasks, such as object recognition or language processing. One major thread that weaves through the discussion is the importance of depth in fully connected networks which enables hierarchical feature learning that shallow networks can only perform poorly. The hierarchical nature helps deep learning models to perform well in tasks that involve multiple layers of abstractions such as contextual under-standing in NLP, or object recognition in images.

The authors provide history of why deep learning was developed begin-ning with the fact that intelligence comes from highly parallel interconnected rela-tively simple non-linear neurons. This insight has led to the creation of artificial neural networks that emulate human biological neurons. It is also evident from the paper that these networks, as a result of layers of activity vectors, learn connection strengths that follow the stochastic gradient of an objective function that quanti-fies performance. This approach, as conceptually straightforward as it is, has been empirically very efficient when dealing with vast data and significant processing power.

Among the major steps pointed out is the use of GPUs (Graphics Pro-cessing Units) for deep learning. GPUs have offered the computation required to train deep networks on large datasets paving the way for many applications. The authors have also expounded on other software platforms that have been devel-oped including Theano, Torch and TensorFlow through which deep learning has been made more accessible; and through which researchers and practitioners in the field can easily build and train deep models.

The paper points out several issues that deep learning must overcome to maintain the rate of success. These include: Unsaid learning; Out of distribution shift; Utility of deep learning in solving system 2 tasks as described by Daniel Kahneman. Solving these problems will involve changes to model structure, train-ing algorithms, and the incorporation of new learning paradigms like reinforcement learning and unsupervised learning. Thus, this paper provides a basic understand-ing of the history and potential development of deep learning and serves as a fun-damental reference for those following this field.

Citation: Alzubaidi, L., Zhang, J., Humaidi, A.J., Al-Dujaili, A., Duan, Y., Al-Shamma, O., Santamaría, J., Fadhel, M.A., Al-Amidie, M., and Farhan, L., 2021. Review of deep learning: concepts, CNN architectures, challenges, applica-tions, future directions. Journal of Big Data, 8(1), pp.1-74.

Alzubaidi et al. present a comprehensive survey of deep learning with em-phasis on Convolutional Neural Networks (CNNs) and their diverse uses. The au-thors begin by demarcating the evolution of deep learning within the ML field, remarking on its fast gain in popularity and high performance on various challeng-ing problems. CNNs specifically have become one of the most crucial layers, pri-marily because of their effectiveness in dealing with structured grid data like im-ages. The paper carefully discusses the progression of CNN architectures from AlexNet in 2012, a pioneering model, to the newer and more sophisticated models such as the High-Resolution Network (HRNet).

The review also focuses on discussing the abilities of CNNs in different application areas. In computer vision, CNNs are now ubiquitous for tasks such as image classification, object detection, and segmentation. Their success is not only limited to vision but spans other areas like natural language processing where they are applied in tasks like text categorization and polarity. The paper also discusses the revolutionary application of CNNs in healthcare specifically in the diagnosis of diseases with the help of radiological images that surpasses conventional tech-niques and, at times, even specialists.

Alzubaidi et al. engage their audience in the potential drawbacks of deep learning as a machine learning technique. One of the main issues is the time-consuming, which is connected with the training of deep networks. This is made worse by the fact that it requires large annotated datasets, which are hard to come by in many domains. The authors also note that problems like these have been somewhat mitigated with the use of hardware accelerators such as GPUs and FPGAs, however, entry into such a field remains expensive for small research groups or practitioners.

Citation: Talaei Khoei, T., Ould Slimane, H., and Kaabouch, N., 2023. Deep learning: Systematic review, models, challenges, and research directions. Neural Computing and Applications, 35(31), pp.23103-23124.

Talaei Khoei, Ould Slimane, and Kaabouch present a systematic review on deep learning, and they particularize on the shift to automation. The paper divides deep learning methods into supervised; unsupervised; reinforcement; and hybrid classes, providing a comprehensive analysis of each class. The authors start by point out that deep learning has been able to automise many tasks increasing the performance and reducing complexity in many applications. This transition is rep-resented as a significant shift in the field’s development, creating vast improve-ments in the area of efficiency and effectiveness.

In the review, the author identifies and elaborates on the features and use of each of the learning models. Of these, supervised learning has been explored in detail and is said to be useful in problems with a vast number of labeled data sets. The authors also discuss the concept of unsupervised learning with a focus on its applicability when labeling the data is unfeasible or too expensive. Reinforcement learning is discussed as a solution for decision-making problems, and the medium of the models that incorporate elements of these techniques are described as flexi-ble and potentially capable of solving more difficult tasks.

Several sections of the review are focusing on the difficulties that deep learning is experiencing at the current time. There is a problem of high costs and time needed to train deep models that may include massive computational power. The authors mention the problem with the current models which do not allow for scalability and transfer learning to new tasks or settings. They emphasize that there is a need for models that can learn more efficiently and with minimal supervision and that many of the models being developed still depend on large sets of anno-tated data. Since this systematic review provides a comprehensive state of the art analysis of the current efforts and focus on the identified gaps, the article by Talaei Khoei et al. is a valuable source of information and should be of interest to both academic and industrial researchers.

Citation: Sarker, I.H., 2021. Deep learning: a comprehensive overview on techniques, taxonomy, applications and research directions. SN Computer Science, 2(6), p.420.

Sarker provides a comprehensive description of deep learning in terms of techniques used, its categorization, areas of implementation and potential areas for future study. The paper first provides an overview of the basic approaches of deep learning and divides them into supervised, unsupervised, and semi-supervised learning techniques. This taxonomy gives systematics to the different strategies as well as their implementation. Sarker also portrays that deep learning is mostly suc-cessful since it can model non-linearity within data, which makes it suitable for different tasks.

It also analyses how deep learning is key to the Fourth Industrial Revolu-tion and how it provides new solutions in many areas. In healthcare, for example, deep learning is applied for disease diagnostics based on medical images, progno-sis, and individualized treatment. In the context of computer vision, deep learning has raised the bar and changed the way images and videos are processed using ap-plications in autonomous driving, facial recognition, and surveillance. The paper also describes deep learning and how it is used in text analytics for purposes such as sentiment analysis, machine translation, automatic summarization, among others. To tackle these challenges, Sarker proposes the following future research ideas. Enhancing the interpretability of black box models is mentioned as one of the key areas to be addressed. Various approaches including attention mechanisms, visual-ization tools, and explainable AI (XAI) techniques are considered to address the lack of interpretability in deep learning models. Another important area in the fur-ther research is the creation of new approaches, which will need less data samples for training and using of deep learning, in conditions of limited availability of the data. The paper also pays great attention to the need to establish reference archi-tectures and benchmarking practices for deep learning. This paper would be useful for any researchers and practitioners concerned with the state and future of deep learning.

Citation: Wang, K., Song, L., Wang, C., Ren, Z., Zhao, G., Dou, J., Di, J., Barbastathis, G., Zhou, R., Zhao, J., and Lam, E.Y., 2024. On the use of deep learning for phase recovery. Light: Science & Applications, 13(1), p.4.

Wang et al. are mainly concerned with deep learning for phase recovery (PR), which is a key processing method in computational imaging. The paper starts with a brief review of conventional PR techniques, which includes reconstructing the refractive index distribution or topography of objects, and correcting for sphere aberration in imaging systems. These techniques are usually based on itera-tive procedures, which may be computationally expensive and time consuming. The authors oppose this approach due to its inefficiency and inaccuracy and pre-sent deep learning as a more promising tool for solving PR challenges.

In the review, the author discusses how the PR process has incorporated deep learning at different levels. At this stage, deep learning algorithms enable the improvement of raw data by eliminating noise and correcting distortion of the ini-tial input. In the in-processing stage, deep learning models can directly create phase images from the enhanced data without the need for the iterative process. In the post processing stage, these models are also useful in improving the recon-structed images in terms of resolution.

One of the main benefits of using deep learning in PR is its data-driven approach and tolerance to variations in imaging conditions. There are various tradi-tional PR methods that may require accurate data on the imaging system and its characteristics, which can be challenging to gather. In contrast, the deep learning models can learn these parameters from the training data and are easy to imple-ment. The authors describe several examples of application of deep learning in op-tical microscopy, tomography and holography in PR.

However, the paper also reveals several limitations of deep learning in PR. One of the major concerns is the requirement of large amounts of labeled training examples which can sometimes be scarce in certain imaging modalities. The au-thors also present the issue of model generalization, where a model trained on one or several datasets performs poorly on other datasets with different properties. To overcome these, they propose the application of data augmentation approaches, transfer learning, and domain adaptation to enhance model stability and effective-ness.

Lastly, Wang et al. propose the potential research avenues for deep learn-ing in PR. They stress the need to create better training algorithms that shall be able to operate with less data and resources in terms of computation. The authors also encourage the development of better interpretability for deep learning models, which are important for building confidence in their applicability for core imaging tasks. They also advise considering the use of a combination of deep learning and traditional PR strategies to benefit from the advantages of both. The paper is use-ful for the researchers who focus on computational imaging and other related fields because it enhances the understanding of current developments and poten-tial future directions.

Topic

The following research idea has been proposed by a researcher: An institution of higher education provides opportunities for students to develop mobile applications that can be deployed using a cloud environment. However, the time to delivery continues to fall behind expected academic performance standards and the institution needs to better understand the factors that influence the development cycle. The institution believes that the use of Learning Analytics, accomplished using an AI enabled environment, could capture real time performance activity, and provide the data needed to allow the faculty to create support structures that encourage better student self-regulation for performance and contributions to the development project.

focus on research methodology & selection of research methods, data selection and LSEPI.
Conduct an initial review of research papers that explored the use of Learning Analytics within Higher Education
Reflect on past work focused on the use and applications of AI and Data Analytics to examine patterns of behaviour, along with your work experiences with the software development cycle (including work with mobile application development and/or cloud computing)
Discuss the potential choices for selecting the research methodology and methods potentially suitable for the research project describe in the Background section (above).
Drawing upon the resources and your critical analysis of the articles on the use of Learning Analytics within Higher Education, provide an initial recommendation for research methodology, related methods for data selection, including a short description of the types of data to be collected, the justification for the recommendation, and how it would support the aims and objectives of the research.

Post

When conducting the present research, it is important to choose an appropriate research method that will best capture and analyze important data for an institution of higher education with regard to the development cycle for student mobile application projects. For this project, it is advisable to employ a quantitative research method. This approach will facilitate the gathering of numerical data for subsequent fact-based analysis with a view of assessing development cycle determinants. Quantitative analyses are helpful in identifying trends and relationships between variables, which is important in understanding the association between Learning Analytics (LA) and students’ performance.

When conducting quantitative research therefore, the choosing of data will involve the use of real time performance data from students participating in mobile application development. This data will be collected using intelligent LA tools which can measure different facets of development. Data that should be gathered include time spent on coding and debugging, the number and kind of errors found, use of development tools and resources, and submission and revision schedules (Berndtsson et al. 2007). These would be self-measured using IDEs and version control systems like GitHub, which contain logs of activity performed by the students.

Quantitative research approach is justified by its characteristics of offering numerical and factual data that can be statistically analyzed. This helps to understand the patterns and relations that are not possible to notice under regular observation. According to Abazi-Bexheti et al. (2020), time series analysis of data plays a key role in the improvement of the research educational processes. Using a quantitative research approach can help the research identify particular performance factors that affect project success and offer clear recommendations to faculty to establish support mechanisms.

Also, quantitative data collection is in line with the institution’s goal of using AI-enabled LA to record performance activity in real-time. The collected data will then be used to develop accurate prediction models which will help in early detection of at-risk students in the early development states. This is consistent with the views of Buchanan et al. (2011) who underscore the significance of systematic data collection and analysis in computer science research. Besides, following the guidelines of the ACM Code of Ethics and Professional Conduct (Gotterbarn et al., 2018) also helps to avoid misuse of students’ data and respect their rights to privacy and confidentiality. Overall, the use of quantitative research method is most appropriate for the proposed project given that it yields numerical values that can be analyzed statistically to determine factors in the development cycle.

References

Abazi-Bexheti, L., Kadriu, A. and Apostolova, M., 2020, September. Word Cloud Analytics of the Computer Science Research Publications’ Titles over the Past Half Century. In 2020 43rd International Convention on Information, Communication and Electronic Technology (MIPRO) (pp. 887-892). IEE. https://ieeexplore.ieee.org/abstract/document/9245141/
Berndtsson, M., Hansson, J., Olsson, B. and Lundell, B., 2007. Thesis projects: a guide for students in computer science and information systems. Springer Science & Business Media.
Buchanan, E., Aycock, J., Dexter, S., Dittrich, D. and Hvizdak, E., 2011. Computer science security research and human subjects: Emerging considerations for research ethics boards. Journal of Empirical Research on Human Research Ethics, 6(2), pp.71-83. https://www.researchgate.net/profile/Scott-Dexter-3/publication/51226554_Computer_Science_Security_Research_And_Human_Subjects_Emerging_Considerations_For_Research_Ethics_Boards/links/0046352b2fa14d36a5000000/Computer-Science-Security-Research-And-Human-Subjects-Emerging-Considerations-For-Research-Ethics-Boards.pdf
Gotterbarn, D.W., Brinkman, B., Flick, C., Kirkpatrick, M.S., Miller, K., Vazansky, K. and Wolf, M.J., 2018. ACM code of ethics and professional conduct. https://dora.dmu.ac.uk/bitstream/2086/16422/1/acm-code-of-ethics-and-professional-conduct.pdf

Summary

The study intended here seeks to fill the existing gap of slow cycle of development of mobile applications within an institution of higher learning. This institution offers students a chance to develop mobile applications that can be deployed from a cloud environment. Even with such opportunities, the time to delivery still fails to meet the expected academic performance standards. In response to this, the institution suggests that incorporating Learning Analytics in an environment enhanced by Artificial Intelligence could provide real-time performance data. Such data would help faculty put in place support structures that would help students better manage themselves so that they could improve on their performance and contribution towards development projects.

Learning Analytics is defined as the accumulating, monitoring, evaluating and communicating of data regarding learners and their environments. It seeks to explain and enhance learning and the contexts in which learning takes place. In a study by Abazi-Bexheti et al. (2020), it was found out that the integration of Learning Analytics with AI can capture live performance activity and thus offer crucial data regarding student learning and development. AI is capable of processing data faster and has the capability to detect information that human cannot find easily. Such real time feed backing can assist students to change their learning behaviors and enhance their outcome. For instance, if a student is discomforted in a certain area such as app development part then the system could suggest materials or help to clear these issues (Berndtsson et al., 2007). This approach is expected to improve efficiency of individual students and at the same time improve performance of developmental projects.

Additionally, this research will outline possible affordances where Learning Analytics can be incorporated into the learning system in order to constantly feedback to the learners. Such integration could result to proper infrastructural support systems that would be flexible for better performance as determined by the students’ performance. This instance shows that with integration of AI, the institution is capable of catering for each learner to give them the help they require so that they can excel (Gotterbarn et al., 2018). This might consist of responses to the learner, learning resources that could be in a position to adapt to developmental phase, and actions that can be taken in any given stage that new emergences during the developmental phase. Moreover, this study will also explore the ethical usage of both AI and Learning Analytics particularly within a learning environment. The study conducted by Buchanan et al. (2011) observed that the privacy of the data collected for the research must first be ensured, and students’ consent should be sought at various phases of data collection. The approach also looks at offering a better school and learning environment that students can use to enhance performance and come up with set standards of performance.

References

Abazi-Bexheti, L., Kadriu, A. and Apostolova, M., 2020, September. Word Cloud Analytics of the Computer Science Research Publications’ Titles over the Past Half Century. In 2020 43rd International Convention on Information, Communication and Electronic Technology (MIPRO) (pp. 887-892). IEE. https://ieeexplore.ieee.org/abstract/document/9245141/
Berndtsson, M., Hansson, J., Olsson, B. and Lundell, B., 2007. Thesis projects: a guide for students in computer science and information systems. Springer Science & Business Media.
Buchanan, E., Aycock, J., Dexter, S., Dittrich, D. and Hvizdak, E., 2011. Computer science security research and human subjects: Emerging considerations for research ethics boards. Journal of Empirical Research on Human Research Ethics, 6(2), pp.71-83. https://www.researchgate.net/profile/Scott-Dexter-3/publication/51226554_Computer_Science_Security_Research_And_Human_Subjects_Emerging_Considerations_For_Research_Ethics_Boards/links/0046352b2fa14d36a5000000/Computer-Science-Security-Research-And-Human-Subjects-Emerging-Considerations-For-Research-Ethics-Boards.pdf
Gotterbarn, D.W., Brinkman, B., Flick, C., Kirkpatrick, M.S., Miller, K., Vazansky, K. and Wolf, M.J., 2018. ACM code of ethics and professional conduct. https://dora.dmu.ac.uk/bitstream/2086/16422/1/acm-code-of-ethics-and-professional-conduct.pdf

Introduction

The purpose of this summative assessment is for you to develop and present a research idea that could be developed into a full proposal and conducted within either a commercial or academic setting, as might be presented to a senior executive or academic leadership group. It should present your rationale for the research project, including the anticipated benefits derived from conducting the research, once disseminated. Consider which of the two types (commercial or academic) of research projects might be appropriate, given your own experiences or expected career goals, including the attainment of a graduate degree.

Video presentation should include the following key components

Personal introduction (who you are, a short background on your experience)
An introduction to the purpose of the video – provides interest to the audience and includes what to expect from the video
Description of the context of the research idea – provide the audience information about the context as a vehicle to ‘set the stage’ for why the research is positioned as a worthy endeavor. Include an explanation on how you would manage the research project
Hypothesis / research questions and, if appropriate, a problem statement that connects how addressing the hypothesis or answering the research questions would achieve new insights into the research context
The research methodology and proposed methods that would be used and why
The data collection strategy – and how the data will inform how you approach either discovering more about the research and conditions associated with the research context and/or testing/validating the hypothesis, answering the research question(s) or solving the problem through the production of an artefact. 
A preliminary review of the analysis of the results – depending on the research idea, this may include analysing quantitative data, conducting a thematic analysis of ‘rich data’ or both.  Include a short explanation of what you think you may uncover in terms of finding
Offer some comments about how your research idea could contribute to the current body of knowledge, either within an academic setting or a commercial/industry-oriented environment
Remember to end your presentation with a short set of concluding sentences

※ PPT and Script File