Scheduling Online Oral Assessments Using an Iterative Algorithm: A Profound Software for Educational Continuity

: In response to the socioeconomic effects of the global COVID-19 pandemic, including the disruption of educational continuity, an online oral assessment solution is proposed. While the next pandemic is unpredictable, the need for effective student assessment during lockdown is essential. This paper presents an iterative algorithm for scheduling online oral examinations as a profound software innovation capable of securely assessing students remotely while adhering to COVID-19 safety requirements. The efficacy of the software was tested by university tutors and undergraduate students for 60 days on local and web hosts. The results show that the algorithm offers optimal performance despite extraneous factors such as bandwidth capacity, operating system type, and client-side web browsers. The software manages the systematic random pairing of examiners, sessions, courses, and candidates, as well as the automated generation of question papers with a marking rubric. This invention would largely provide assessment entities with significant pedagogical, financial, and technological benefits while ensuring educational continuity regardless of the pandemic's status.


Introduction
The emergency and widespread COVID-19 pandemic have significantly stimulated software innovation in response to the affected socioeconomic sectors.In the field of education, ensuring continuity in student assessment emerged as a major challenge despite the availability of various online teaching and learning services.Overall, the COVID-19 pandemic has presented significant obstacles to educational delivery (Schleicher, 2020), particularly during periods of lockdown.
Although the majority of schools and universities were impacted (Nicola et al., 2020), many were still able to continue teaching and learning through different forms of online education, including elearning platforms and audiovisual collaboration tools for video conferencing services (Linden & Gonzalez, 2021).While these online learning facilities proved beneficial for both synchronous and asynchronous teaching modes during the COVID-19 pandemic (Daniel, 2020), the main challenge was how to administer examinations, considering that both tutors and students were geographically dispersed and were held under lockdown.During this period, assessment methods that relied on face-to-face sessions and physical classes were completely prohibited, as they failed to meet COVID-19 safety requirements (Adedoyin & Soykan, 2020).Instead, a remote-based online assessment emerged as the sole alternative solution to ensure educational continuity.This form of assessment was the recommended approach (García-Peñalvo et al., 2021) due to its ability to connect participants from anywhere.For example, the University of Twente (UoT, 2020) and Iowa State University (ISU, 2020) utilised video conferences and telephones for student assessment during the COVID-19 pandemic.Other educational institutions employed zoom-invigilated exams to facilitate online remote assessment (Linden & Gonzalez, 2021).However, not all online assessment methods are suitable or compatible with the COVID-19 pandemic (Weiner & Hurtz, 2017).Online written

Development tools and testing plan
OREX development is based on open-source tools such as Linux OS (Ubuntu 20.04 SE, 2020), MySQL database (8.0), PHP programming language (PHP, 2020), Apache web server (Apache, 2020), and the MVC framework for rapid software development.All tools were hosted in a testing box with the following specifications: • Processor Type: Intel® CoreTM i7-8565 CPU@1.80GHz,1.99GHz, • Installed Memory (RAM): 16 GB with 15.7 GB usable, System Type, • System Type: 64-bit operating system, x64-based processor, • Operating System: Windows 10 Home 2019 For efficient OREX testing, the following set of activities were completed: • Development of registration and examination modules.
• User training on the use of OREX software.
• A question bank of 879 courses with a total of 12 questions for type 1 and 36 questions for type 2. • Server configuration with all necessary underlying software programs and drivers.
• OREX guiding tools: a quick short guide for students, OREX templates for Type 1 and Type 2 oral questions, a short guide for the chief examiner (CE), second examiner (SE), and exam observer (EO), a short guide for audiovisual settings, a marking rubric, and a template for the observer's report.
All participants possessed basic web skills and a smart electronic device with audiovisual capabilities, reliable power, and Internet access.

Algorithm testing
As a web-based and real-time application, OREX was tested intensively for its security and efficiency (Ally, 2014).The software was tested for two months, from June 18th, 2020, to August 14th, 2020, using two testbeds configured as local and webhost machines.The web test was configured at http://orex.out.ac.tz,where the server memory was upgraded from 8 to 32 GB of RAM in order to handle increasing concurrent server requests and workloads.Similarly, the Internet speed was increased from 150 to 300 Mbps during the same period (Ally & Oreku, 2022).Table 1 presents important testing parameters for the webhost machine.

Performance metrics
The performance metrics of OREX are determined by the randomisation and automation levels in the registration and examination phases.The number of sessions for a specific course or examiner, however, depends on the number of registered students.The performance test is based on various factors, such as the repetition level of pairing aspects, the number of loop iterations after the condition is met, CPU execution time (ns), and the functional programming loop.The pairing aspects in the registration phase include candidate/course (cc), candidate/date (cd), and candidate/session (cs).In the examination phase, the pairing aspects are examiner/candidate (ec), examiner/examiner (ee), examiner/observer (eo), examiner/subject (es), examiner/date (ed), and examiner/session (es).

Control of extraneous factors
For clarity, an optimal performance level of the OREX algorithm was achieved by controlling extraneous factors by configuring a localhost testing machine.The testbed was configured with an Intel® CoreTM i7-8565 CPU at 1.80 GHz, 16 GB of RAM, a 64-bit OS system type, and an x64-based processor.The following were the controlled extraneous factors: • EF9: Course capacity, i.e., the number of students in a particular subject.
The first five extraneous factors (EF1-EF5) were completely controlled by the localhost machine.The last two factors (EF8 and EF9) were controlled by limiting the number of examiners and candidates by setting the following criteria: • Each course must have at least five examiners.
• The class size should range between 100 and 120 students.
• CPU Execution Time (CET) was used instead of Server Response Time (SRT) because the Internet speed was considered an extraneous factor.• To ensure the computational complexity of the proposed algorithm and the accuracy of the data collection are of high precision, each experiment was repeated three times.

Requirement specification
The OREX system enables oral assessments through audiovisual online collaboration tools.The solution saves time and financial resources, as it allows for remote connection with the web audience.
The system also facilitates instant feedback, as invigilation and marking are completed during the session.Figure 1 illustrates a typical OREX session with four participants connected via a web link.OREX works by assessing only one candidate in a single session.While CE and SE are subjectspecialists, EO can be any academic staff member randomly selected from the pool.The role of each participant is defined in Figure 2. As depicted in Figure 2, OREX must satisfy both hard constraints and soft constraints to be feasible.
The following set of constraints guided the development of the software algorithms:

Hard constraints
For OREX to attain maximum efficacy, seven hard constraints (HC1 to HC7) are prerequisites.
a) HC1: Examiner vs. Session vs. Course vs. Date • No examiner can be assigned more than one session at the same time.
• No examiner can assess more than one course at the same time.
• For convenience, no examiner is assigned two or more sessions on the same day if the event last long.b) HC2: Candidate vs. Session vs. Date • A candidate cannot sit for two or more sessions at the same time.
• For convenience, no candidate is assigned two or more sessions on the same day if the event last long.c) HC3: Course vs. Question vs. Knowledge Area • Each course has at least 12 questions with model answers.
• Each part has six questions.d) HC4: Course vs. Examiners • Each course has at least two examiners registered in the system.
• Examiners are subject specialists who can ask and mark.e) HC5: Candidate vs. Timetable • A candidate has a unique timetable.f) HC6: Uniqueness and Pairing of Examiners (CE/SE/EO) • The pairing of CE, SE, and EO is randomly defined.
• Examiners and Candidates are uniquely identified.g) HC7: Convenience for Examiners and Candidates • Convenience of examiner is determined by the software.
• Convenience of candidate is self-determined.• Large courses (those with 500 students and above) must have at least five examiners.

Soft constraints
• The software should be able to detect the examiner's workload from other courses.

OREX Software Algorithm
The OREX algorithm constantly maintains the automation and randomisation processes for all key software functionalities, including session allocation and question generation.The algorithm controls the random generation of question papers, mark sheets, and computation of the final score.The algorithms are helpful in automating the examination timetable.The randomisation process considers a set of constraints, logics, and conditions for all pairing aspects as follows: • Registration phase: candidate/course (cc), candidate/date (cd), and candidate/session (cs).• Examination phase: examiner/candidate (ec), examiner/examiner (ee), examiner/observer (eo), examiner/subject (es), examiner/date (ed), examiner/session (es).
The following are the set of conditions constrained in the algorithm: • No examiner is assigned more than one session at the same time.
• No examiner is assigned more than one course at the same time.
• A student cannot sit for two or more examinations in the same session.
• No examiner can assess two or more sessions on the same day if there is a substantial time span of OREX days.• The first OREX question comes from the first part of the course (knowledge areas 1-3), while the second question comes from the second part of the course (knowledge areas 4-6).
• Each course has at least two examiners who can assess and mark the questions.
• Each student has a unique examination schedule.
• The software uniquely identifies each CE, SE, and EO using the staff personal file (PF) number.
• The software should uniquely identify each candidate using a student registration number.
• The convenience of the examiner is determined by software.
• The convenience of the candidate is decided by the student himself.
The solution consists of five algorithms focused on candidate registration, session allocation, examiner selection, collision free for candidates as well as collision free for examiners and observers.(a) The first session  1 is initially checked for all days before the second session  2 is checked.(b) The second session  2 is checked for all days before the third session  3 is checked, and so on.(c) Equation 2 is outer loop.(d) Operation Mode: S1d1, S1d2, S1d3, …S1dn, …S2d1, S2d2, S2d3, …S2dn, … S3d1, S3d2, S3d3, …Smdn, … Condition 4: The algorithm iterates all OREX days for the first session, reiterate again for all days for the second session until when a collision free session is found.Outer Loop:  (   = 1;   <= ;   + +) Inner Loop:  (   = 1;   <= ;   + +) Algorithm 5: Collision Free for Examiner and Exam Observer Condition 1: Once the collision free session     is found for a candidate, the system checks for the two examiners{, } .The system checks for examiner collision free session Condition 2: Examiners (CE/SE) are selected from a pool of course experts.{, } Condition 3: If the course is allocated more than two examiners, the system checks for combination or pairing of examiners {, }  (    ).Condition 4: Student collision free is checked:  (    ) = = 1 Condition 5: Examiners pairing is checked:  (, ) = = 1 Condition 6: Examiners collision free is checked:  (    ) = = 1 Condition 7: Exam Observer is selected from the pool of all tutors:  {} Condition 8: Collision free for Exam Observer is checked:  (    ) = = 1, {, , }

Software report production and retrieval
The OREX software produces reports for use at departmental and university levels for exam registration, exam sessions, exam results, and examiner lists.A list of registered students displays information such as the student's full name, registration number, course code, exam date, session time, and confirmation status.The OREX Visa Card (OVC) is simply an examination hall ticket which consists of admission and academic data.Admission data includes student name, student photo, registration number, physical address, exam centre, program name, index number, sponsor, birthdate, and printed date, while academic data includes course code, course name, exam date, and session time.The list of examiners includes staff file number, staff name, course code, role, status, action, web link, and phone number.The system can also generate a list of courses and the number of questions in the database.The question paper comes with a course code, question code, question text, and model answer.The online mark sheet with an inbuilt marking rubric composes score boxes for communication skills, creativity, knowledge and understanding, problem-solving, understanding of learning outcomes, confidence, fluency, and accuracy.The course and student result sheet includes the student registration number, course code, examiner, question, and score awarded.The observer report captures information about students, sessions, infrastructure, and academic profiles.Finally, four basic system profiles include the following: • Student profile (student name, examination center, registration number, program, birthdate, sponsor, and Form 4 index number).
• Session profile (date of exam, time-in and time-out, course code, and the session link), attendance status of each participant (CE, SE, EO, and candidate), verification of the candidate, if CE/SE has worked independently and with fair treatment, participant's dress code).
• Academic profile (question paper generated, mark sheet, examination irregularity matters, number of follow-up questions asked by examiners) • Infrastructure profile (Internet accessibility, availability of the OREX server, Internet speed, number of session breaks and disconnects, and the adequacy of physical arrangements in the exam virtual room, including noise and interference).

Discussion
The development of OREX software provided an opportunity for examiners to receive instant feedback from students.Based on the question paper blueprint, the questions prepared were comprehensive, integrated, and randomly generated from the software.Follow-up questions were also used to prompt more elaborate answers from the students.This assessment system has helped students build confidence and evaluate their communication skills, creativity, application, problemsolving abilities, knowledge, and understanding of the learning outcomes.This was made possible due to the high flexibility and greater objectivity of oral assessment compared to written examinations.
A comparative analysis to evaluate the performance level of the algorithm was theoretically conducted using a mathematical technique called big-O notation.The assessment of the computational complexity and efficiency of the software algorithm using run-time analysis shows that the loops behave equally on the server local host and the server web host if extraneous factors (EF1-EF9) are handled properly.For instance, the performance difference between the web host and localhost was insignificant when the internet speed increased from 150 Mbps to 300 Mbps, regardless of the web browser and operating system used in a client computer with minimum computing specifications, such as a 2 GHz CPU and 4 GB RAM.This concludes that the OREX algorithm offers optimal software performance if extraneous factors are taken care of.Since the estimated run time recorded to measure the time spent when user requests are submitted between the client and server depends on the input size, a slight performance difference was observed for big courses with at least 120 concurrent logins.Run time is directly proportional to the input size for both web host and localhost servers when a linear search algorithm is used.In this case, a localhost machine that runs a linear search program exhibits a linear growth rate.On the other hand, a binary search showed high performance in a web host machine when the input size became large.This means that a web host machine running the binary search program exhibits a logarithmic growth rate because the run time increases by a constant amount.Even though a localhost is a faster machine due to no interference from extraneous factors, a web host machine will inevitably surpass a localhost in runtime because it runs an algorithm with a much slower growth rate.

Conclusion and Recommendation
The OREX algorithm is a state-of-the-art software solution that helps maintain educational continuity.Candidates can be assessed from anywhere through an audiovisual online tool.The algorithm controls automation and random selection of examiners, observers, and question paper generation while managing workloads, convenience, and security.Overall, the development of the OREX algorithm surpasses the challenges posed by COVID-19 to the education industry.Higher learning institutions can adopt this software for efficient management of oral assessments for students participating from anywhere in a cost-effective manner.

Declarations
Funding: This research did not receive any external funding Acknowledgements: I would like to acknowledge the Open University of Tanzania for their support during the entire development of the OREX system.I would also like to thank the programmers in the Directorate of Teaching, Learning, and Examination Services for their technical support.

Conflicts of Interest:
The author declares no conflict of interest.

•
EF1: Internet speed of the client machine, • EF2: Diversification of Internet browsers among examiners and candidates, • EF3: OS type of the client machine, • EF4: CPU speed of the client machine, • EF5: RAM size of the client machine, • EF6: Latency factor from client to server.• EF7: Connect time between user and server, including SSL handshake.• EF8: Number of examiners per subject.

Figure 1 :
Figure 1: Virtual Exam Hall (i~Room)The OREX web audience includes the Chief Examiner (CE), Second Examiner (SE), Exam Observer (EO), and candidate.Candidates are uniquely identified by their registration numbers, while CE, SE, and EO are identified by staff numbers.Users are registered based on their assigned roles and privileges and are given access to the virtual link through software.

Figure 2 :
Figure 2: OREX Structural DesignThe random pairing of CE/SE/EO is of high anonymity level and is determined by an OREX algorithm based on user workload and convenience.Each day, the OREX live sessions lasting for a maximum of 60 minutes cannot exceed 10 per examiner.The assessment covers all learning outcomes by dividing the course into two parts, each with six comprehensive oral questions.Each question is assigned a unique code in the following format: CourseCode_Part_QuestionSerialNumber.The CE ask question 1 while the SE ask question 2. Both examiners are markers of oral responses based on the defined rubric.The question papers and marking schemes are randomly generated and are activated at the time of print (ToP) based on the OREX Visa Card (OVC).The OVC is a timetable printout with the candidate's academic and demographic information.The system computes and saves the final score recorded by CE and SE in an online mark sheet for permanent storage.

Table 1 .
Webhost Collision Free for Candidate Condition 1: OREX session allocated to a candidate based on available CE/SE/EO.Condition 2: The system checks for a student collision free session Condition 3: For student's convenience, the nested loops between OREX days and OREX sessions are checked.