Geo-Lecture : Exploiting location-based recommendation in the context of lecture attendance tracking using a geographical information system (GIS)

Abstract

To ensure productivity in the higher education sector, tracking lecture attendance of instructors and students is vital. Low attendance often leads to negative consequences in terms of optimal output for both individuals and the institution. Traditional methods of locating lecture rooms and tracking attendance are inefficient and error-prone. In this article, we propose Geo-Lecture, a location-based recommender algorithm which leverages a geographical information system (GIS). Using student and location data from a Technical University in Ghana, Geo-Lecture provides accurate, and real-time lecture attendance tracking. Evaluation results demonstrate that Geo-Lecture significantly outperforms contemporary methods with the highest precision of 0.31, recall of 0.34, and an F1-score of 0.32. Additionally, the mean absolute error (MAE) and Normalized MAE both illustrate results of 0.69 and 0.17, respectively, which are the lowest comparatively. These results validate Geo-Lecture's superiority in terms of accuracy, reliability, and usability in comparison to existing approaches.

Keywords

attendance tracking geographical information systems higher education geospatial technology location-based recommender systems

1 Introduction

TRACKING and monitoring instructor and student attendance for lectures are essential for maintaining productivity in the higher education sector (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015). Globally, most higher education institutions (HEIs) currently devote a significant amount of time and resources to improving their teaching and learning strategies. However, issues with low student attendance, a subject that has attracted more attention in recent years, could jeopardize these efforts (Arulampalam, Naylor, and Smith, 2012; Khong et al., 2016; Westerman et al., 2011). Furthermore, in earlier research, it has been shown that attending lectures regularly helps students perform better academically. Although there are some disagreements about how much student attrition in lecture halls skews teaching and learning outcomes, a full classroom is essential to the current trend of student-centered pedagogy in higher education (Arulampalam, Naylor, and Smith, 2012; Khong et al., 2016; Westerman et al., 2011).

Low attendance has detrimental effects on students’ and teachers’ ability to produce their best output in terms of performance. In the traditional approach, when multiple people, including workers, supervisors, and payroll administrators, perform duties that require recording the numbers, the time and attendance data are vulnerable to human errors (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015). Thousands of students and faculty members pass through large and crowded college campuses every day. On campus, it is simple to become disoriented, perplexed, and angry due to the abundance of buildings, floors, and indoor and outdoor elements (Clemenson et al., 2021; McMahon, Cihak, and Wright, 2015; Vasudevan et al., 2020). This begs the question: “How can colleges and universities relieve students who are having trouble getting to the lecture room on campus?” Campus navigation systems have been developed and implemented by some researchers, such as Clemenson et al. (2021), McMahon, Cihak, and Wright (2015), and Vasudevan et al. (2020). However, in terms of lecture room navigation, these systems lack important location data provided by location-based recommender algorithms/systems (Ding et al., 2018; Ravi et al., 2019; Yue-Qiang et al., 2019).

The development of location-based recommender algorithms/systems has opened the door to better methods of recommending sites that users might find interesting. Predicting the location of a user at any given time is essential in today's society. Numerous services become more significant when the location of a person is accurately determined (Ding et al., 2018; Ravi et al., 2019; Yue-Qiang et al., 2019). Based on past usage, location prediction helps identify neighborhood points of interest and forecasts the future location of a specific user. Today, most service providers, such as Uber, Yango, and others, use these algorithms to locate restaurants or solve traffic congestion problems (Ding et al., 2018; Ravi et al., 2019; Yue-Qiang et al., 2019).

Additionally, a location-based recommender system makes more relevant suggestions for goods, services, or places of interest by utilizing location data, such as travel times, in its algorithms. When users search for a restaurant, hotel, or job, for example, they want every result to be a location that they can quickly get to. When making decisions, users will consider accessibility, especially when traveling to a destination (Al-Nafjan, Alrashoudi, and Alrasheed, 2022; Sánchez and Bellogín, 2022).

Compared to traditional approaches and campus navigation systems, location-based recommendation systems can be used to improve methods of monitoring and tracking lecture attendance of instructors and students, especially in the cases of educational establishments with very large campuses (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015). A geographical information system (GIS) consists of computer hardware and software that stores, manages, analyzes, edits, and visualizes geographic data. GIS has been widely used for location-based services by researchers such as (Asabere et al., 2021; Bruno and Giannikos, 2015; Huang et al., 2018; Jin et al., 2020; Risimati and Gumbo, 2018).

In this article, we propose a location-based recommender algorithm called Geo-Lecture which uses a GIS approach. Our proposed Geo-Lecture solution uses a location-based recommender algorithm integrated with GIS to eliminate errors and improve analytical report writing of lecture attendance. We do this using data from a Technical University in Ghana to provide location and student information. Our proposed Geo-Lecture solution requires a student to be within a certain distance of the lecture hall to have their attendance recorded. Our main contributions to this study are as follows.

(a)
We present a comprehensive framework designed to improve lecture attendance monitoring and tracking through the integration of GIS and a location-based recommender algorithm. Our method involves a multistep process, beginning with data modeling and exportation to an ArcGIS Diagrammer tool for further definition and population, ensuring well-defined domains and feature class attributes.
(b)
Our proposed Geo-Lecture framework validates the symbiotic relationship between the Data Schema and the CSV file as pivotal in establishing a robust framework for capturing, organizing, and harmonizing both geospatial and attribute information.
(c)
Additionally, our proposed Geo-Lecture framework incorporates the import of the Geodatabase file into ArcMap, where general settings and symbology for each asset are defined. Subsequently, the web-based map is described and published as a service on the ArcGIS platform, allowing users to access, view, and interact with geographic data through various client applications.
(d)
Furthermore, through an experimental evaluation procedure, using real-time data, we compare our proposed Geo-Lecture framework with existing state-of-the-art methods.

The rest of the article is organized as follows. Section 2 reviews the literature on lecture attendance monitoring and tracking. The operational concept and algorithmic design of our Geo-Lecture framework are discussed in Section 3. In Section 4, we present the performance evaluation of our proposed method. Finally, Section 5 concludes the article.
2 Review of the literature

In this section, we review the literature about the concept of attendance systems in higher education, the relationship between location-based system (LBS) and GIS, and related studies.

2.1 Concept of attendance systems in higher education

Traditionally, the procedure for taking lecture attendance in higher education involves an approach in which students and lecturers have to physically sign attendance sheets using pen and paper. The current challenges with this method are that students have to fill or sign their names and IDs manually; sheets are also prone to errors and fraud since attendance marking data can be manipulated by students, leading to inaccurate attendance data being collected and a lot of time wasting (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015).

To replace the traditional paper-based attendance method with a digital one, IT solutions have been used. Several methods, such as near field communication (NFC), radio frequency identification (RFID), biometrics, quick response code (QR-Code), etc., have been used for this purpose. At the turn of the millennium, the Internet and the web drastically changed every aspect of our lives. We now employ the use of specialized Internet/web-based information systems for communication, transactions, all-round development, sharing and exchanging ideas and information, learning, and some other regular leisure activities (Kumar, Qadeer, and Gupta, 2009; Stojanovic and Djordjevic-Kajan, 2001; Zhao et al., 2023).

2.2 Relationship between LBS and GIS

LBS represents a real-world application of the endless potential of GIS. LBS is distinct from other conventional GIS and web mapping apps because it can adjust the appearance and content based on the context in which its users are now using it thus, LBS is a fusion of mobile telecommunications and GIS. However, with its roots in GIS, LBS is a fusion of several technologies, including telecommunications, the mobile Internet, and GIS. Global positioning systems (GPS) and location-enabled Wi-Fi are two examples of LBS that are inextricably linked to GIS (Asabere et al., 2021; Bruno and Giannikos, 2015; Huang et al., 2018; Jin et al., 2020; Risimati and Gumbo, 2018).

To satisfy the requirements of LBS, it is crucial to connect mobile computing and GIS technology, which is regarded as one of the most promising uses of GIS. Therefore, to make LBS work, a mobile device or internet-capable device, positioning capabilities, a communication network, and a service provider are necessary requirements. To pinpoint the current location of the user, the positioning component primarily uses GIS technology (Asabere et al., 2021; Bruno and Giannikos, 2015; Huang et al., 2018; Jin et al., 2020; Risimati and Gumbo, 2018; Seetharaman, Mathew, and De’, 2024).

2.3 Related studies

In this section, we review some related studies. Geolocation is the first step in providing location-based services. The most widely used location technologies are GPS, Wi-Fi, cell phone, Bluetooth, infrared, and RFID (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015). Researchers work on these technologies to improve services, such as accuracy and environmental effects. As mentioned earlier, the use of RFID or NFC technology is one method of creating class attendance systems (Le et al., 2020; Mohandes, 2017; Moores, Birdi, and Higson, 2019; Sultana, Enayet, and Mouri, 2015).

For instance, Hameed et al. (2015) used RFID technology in an attendance system by creating a Java program specifically for this purpose. This specific application creates connections between the data gathered from RFID devices to automatically track attendance and add entries to the database. Typically, student ID cards contain special RFID tags embedded in them. Students must therefore provide their student ID cards and scan them using a reader. However, one disadvantage of such a process is that anyone can use identity cards. As a result, verifying the student's identity is insufficient to register their attendance.

Bluetooth Low Energy has attracted a lot of interest from researchers as a technology for close-proximity network node communication and its potential in attendance tracking systems in schools. Lodha et al. (2015) created a Bluetooth-enabled device application for an attendance management system. Their technique requires Bluetooth electronic tags that are imprinted on student ID cards. This system increases productivity because it can connect to a wireless network and utilize Bluetooth technology, eliminating the need for students to hold their cards in front of a reader.

Similar to Lodha et al. (2015), Apoorv and Mathur (2016) proposed a computerized solution that helps teachers track student attendance in the classroom. The system uses Bluetooth low-energy beacons placed in the classroom to identify the student's identity and attendance. Students must bring their student ID card to class. The application on an Android device that connects to the Bluetooth Low Energy beacon and gathers the sensor data must still be used by the teacher to manually extract attendance data.

Saraswat and Garg (2016) created a beacon-based application for faculty administrative tasks. Through Web links and an application, their proposed solution immediately communicates and engages with students. Additionally, the technology continuously collects student attendance data throughout the lesson based on where they are in the room. When the learner is close to the beacon (near and immediate), the system starts with a clock timer (in the program background). This approach primarily uses the student's smartphone to carry out the computation. On the student's mobile device, this calls for a large amount of processing. Similarly, Sultana et al. (2015) proposed a student attendance tracking solution using an Android-powered smartphone. The implemented system made use of the phone's GPS which is used to track workers’ location within the organization. Furthermore, in relation to geographic location, Soko (2024) proposed a web-based geofencing system for automating attendance. Lecturers (rather than students) record attendance based on their GPS location relative to a predefined virtual boundary, eliminating manual roll calls. Similarly, Lala et al. (2022) proposed a mobile/web app that only allows attendance of students to be recorded only when they are within a predefined geolocation radius of the classroom using the browser's PHP backend, MySQL database, and Geolocation API.

Some researchers have also applied GIS in location-based recommender algorithms/systems. For example, Acharya et al. (2023) presented a long-short-term memory (LSTM) based approach for POI suggestion. The key component of the strategy is the spatial binning that groups the venues according to how close they are to the user's previous venue visits. Additionally, Riehl (2020) developed a property recommender system that includes the following elements: Incorporating hotspot analysis and data analytic methods to determine which areas are most ideal for each type of property based on historical and current data and using derived geospatial knowledge as new features and viewpoints for a better overall understanding of a given property. Furthermore, despite progressing from tensor factorization to RNN-based neural networks, previous techniques suffered from the sparsity problem and failed to exploit spatial data efficiently. To do this, Lian et al. (2020) presented a geography-aware sequential recommender for location suggestion that is based on the self-attention network, or GeoSAN for short. Concerning tourism, the integration of this multimedia GIS into a geo-tourism recommender system was the main focus of Zhou et al. (2021). Zhou et al. (2021) presented a smart tourist recommender system based on collaborative weighted filtering and geographic cellular clustering. After the challenges were examined and concluded, research concepts and approaches were presented to address the issues. Their proposed recommender system can tailor its communication to each user specifically. It is suggested that the user visit tourist attractions that are of geological interest and align with their interests, needs, and knowledge.

Related studies show that the drawback of the NFC, RFID Bluetooth tracking, and monitoring systems for students’ attendance requires additional preparation and equipment installation such as face detection cameras. Furthermore, most GIS applications in location-based recommenders have been done in the field of tourism and have not tackled issues of attendance tracking and monitoring of users. To the best of our knowledge, this is the first time GIS has been applied to a location-based recommender algorithm in the higher education sector. In utilizing indoor localization techniques, we decided to use pre-existing equipment (smartphones) in our proposed Geo-Lecture solution to reduce operational expenses and difficulties. The integration of GIS and a location-based recommender algorithm with accessible web-based maps for both mobile and desktop systems constitute our proposed Geo-Lecture framework and ensures a seamless user experience. Geo-Lecture emphasizes data collection, involving the acquisition of geographical information about the university campus, as well as collecting detailed data about class schedules, including times and locations.

3 Proposed Geo-Lecture solution

In this section, we describe our proposed Geo-Lecture solution. In our proposed method, GIS is infused with a location-based recommender algorithm to provide a web-based map for tracking and monitoring mobile attendance of lecture room sessions. The system is an innovative approach to acquiring first-hand information about the students’ attendance. Web-based mobile and desktop systems will be used to deliver required Geo-Lecture services.

Figure 1 shows the basic recommendation procedure of Geo-Lecture and therefore depicts our motivation and innovation through the recommendation entities we utilize. Figure 2 shows that through the augmentation of relevant context, the Geo-Lecture recommender algorithm generates location-based recommendations of lecture rooms by, respectively, collecting students’ names and index numbers, thereby generating profiles and integrating these profiles with contextual information, that is, data and time, user, computing and geolocation using GIS.

Figure 1.

Basic recommendation procedure flow of geo-lecture.

Figure 2.

Geo-Lecture recommendation model.

Our proposed Geo-Lecture recommendation model is illustrated in Figure 2. Figure 2 shows a comprehensive architecture which emphasizes the interaction between users (students and lecturers), data processing components, and the contextual recommendation engine. Figure 2 is organized into seven layers, each denoting a key functional component in the data flow. The corresponding data flow diagram relating to our proposed Geo-Lecture recommendation model is shown in Figure 3.

Users (students and lecturers)

Students (S₁, S₂,K, S_n) and active lecturers are represented as system users.

The profile of each student comprises their name, index number, course enrollment, and schedule.

Student information collector (SIC)

The Geo-Lecture App facilitates the gathering of raw data from the students via the SIC component.

The data collected includes student identity, login timestamp, selected course, and geolocation data (latitude and longitude) of the lecture venue.

The data is then forwarded to the Student Profile Engine (SPE).

SPE

The SPE is responsible for generating and maintaining a detailed profile for each student.

The SPE integrates contextual attributes such as device information, courses data, location and login and consequently feeds the students’ profiles into the contextual post-filtering (CPF).

CPF

Based on the model by Adomavicius et al. (2021), the CPF implements a context-aware recommendation strategy and filters/adjusts recommendations based on real-time contextual factors such as:

○ Device/User ID matching (Is the logged-in student the authorized user?)

○ Time (Is the student within the scheduled lecture period?)

○ Proximity of Geolocation (Is the student within the geo-fenced lecture room?)

Recommendation engine (RE)

After contextual filtering by CPF, the RE is responsible for computing and generating the best-fit lecture room for the student based on matched lecturer data and scheduled classes.

The student receives computed classroom coordinates which are mapped via GIS.

GIS integration layer (GIL)

The GIL utilizes ArcGIS to provide a map with visual markers for lecture room locations.

The lecture room recommendation is delivered through a web-based map interface accessible via a mobile device.

The attendance of students within the defined geofence is automatically marked.

Feedback loop

The Geo-Lecture App logs attendance and sends updates to the Administrator Dashboard, which monitors student attendance trends in real time.

Students can confirm or verify their attendance details.

Our proposed Geo-Lecture method prioritizes GIS integration, utilizing the technology to map the campus, integrate it with lecture schedules, and associate each venue with geographical coordinates. The resulting web-based map is designed with features for easy navigation and visualization of lecture room locations. Mobile attendance is facilitated by implementing a location-based recommendation algorithm for mobile devices, allowing students to verify lecture room directions on the web-based map. The user-friendly interface is a key aspect, with clear instructions for attendance marking and a centralized system to capture and store attendance data. Security measures are implemented to prevent unauthorized attendance marking. In summary, our proposed Geo-Lecture method provides a holistic approach to lecture attendance tracking, leveraging GIS, location-based services, and a web-based map to create an efficient and user-friendly system. Additionally, concerning our proposed method, we further discuss the following: spatial data collection and analysis, system architecture and implementation, and development tools and functionality of Geo-Lecture in the subsections below.

4 Geo-Lecture design

This section presents the design procedure for our proposed Geo-Lecture method in terms of our algorithms, software requirements gathering, and GIS design technologies utilized.

4.1 Algorithmic Design

Our proposed Geo-Lecture pseudocode for location-based recommendation of lecture rooms and the pseudocode for our proposed Geo-Lecture Application are respectfully illustrated in Algorithms 1 and 2 with explanations below.

Figure 3.

Data flow diagram of Geo-Lecture.

Below is an explanation of how Algorithm 1 functions:

Start and initialization

Algorithm 1 begins with declaring and initializing variables: ○

a, b, and z are integer counters for iterating through lists.

○

lecture_room_location[z] is a floating-point variable storing geographical coordinates for recommended lecture rooms.

○

location[n] and time[n] are string arrays storing the location and time information for each lecture.

○

Students[n] is an array holding n students.

○

Lecturers[m] is an array holding m lecturers.

Outer loop for students

Algorithm 1 iterates through each student by starting a loop that runs n times (one for each student).

In each iteration, the student at index [a] in the Students array is selected for comparison.

Inner loop for lecturers

For every student, another loop is started to iterate through all lecturers in the Lecturers array (m times).

This inner loop ensures that each student is checked against all lecturers to find a match.

Conditional check

Inside the inner loop, an if statement checks if: ○

The current lecturer's location (Lecturer[b].location) matches the student's location (Student[a].location).

○

The current lecturer's time (Lecturer[b].time) matches the student's time (Student[a].time).

If both conditions are true, this indicates a match between the student and lecturer based on location and time.

Compute and store classroom coordinates

When a match is found, the algorithm computes the geographical coordinates of the classroom location and stores them in classroom_location[z], where z is the index for storing matched locations.

Generate recommendation

The algorithm then assigns the matched student to the lecturer (Student[a] to Lecturer[b]) and directs the student to the calculated classroom location.

End of loops

The inner loop (b) completes for each student, and the outer loop (a) continues until all students are matched to lecturers as applicable.

End

Below is an explanation of how Algorithm 2 functions:

Algorithm 2 concludes after all students have been assigned a recommended lecture room location based on matching criteria.

Start ○

The algorithm initiates the process by starting the steps outlined below.

Step 1: open the Geo-Lecture app ○

The student opens the Geo-Lecture App on their device to begin the attendance and recommendation process.

Step 2: check for scheduled lecture ○

The app checks if the student has a lecture scheduled at the current time. This is a conditional check.

○

If a lecture is scheduled, the app proceeds to the login process.

○

If no lecture is scheduled, the student exits the Geo-Lecture App (ExitGeoLectureApp()).

Step 3: student login ○

The student logs into the app by entering their indexNumber and password.

○

This step verifies the student's identity to prevent unauthorized access.

Step 4: input class details ○

After a successful login, the student inputs the details of the class, such as the course code, classroom location, and any other required information.

Step 5: data verification ○

The app prompts the student to verify that the entered information is accurate.

○

This step allows the student to review their details to avoid any errors before submission.

Step 6: submit data and receive directions ○

Upon verification, the student submits the data to the system.

○

The app then uses this data to generate directions or recommend the appropriate lecture room based on location-based criteria.

Step 7: logout ○

After submitting the data, the student logs out of the Geo-Lecture App, securing their session.

Step 8: exit the app ○

If there was no lecture scheduled, the student exits the app immediately, skipping steps 3–7.

End ○

The algorithm completes after either the student logs out following data submission or exits the app directly if no lecture is scheduled.

4.2 Gathering of Software Requirements

In this section, we will outline the software requirements that are to be gathered for our proposed Geo-Lecture solution, taking into consideration the possible conflicting requirements of the various users.

User Requirements—It is very important to get users of the system fully involved so that the problem of accessing the system and the uncertainty of not having the system readily available does not occur. The following are user requirements of our proposed Geo-Lecture solution: (i) a system that is easy to use, (ii) a system that provides an attractive interface with easy navigation, and (iii) a system that would show them the exact location of lecture rooms and detailed information.

Functional Requirements—The identified functional requirements are: (i) the system would ensure accurate collection of student's lecture attendance records, (ii) the system would allow students to verify locations of lecture rooms either through mobile and web app, (iii) the system would be user-friendly and provide an attractive interface for easy visualizing and querying of data, and (iv) the system would be able to provide directions to lecture rooms at particular and relevant lecture schedules.

Non-Functional Requirements—The non-functional requirements identified are: (i) Security—Information about the system should be secure and safe, (ii) Maintenance—The system permits to be upgraded and modernized when it is necessary, (iii) Usability—The system can be accessed by the admin and the user without issues or difficulties, (iv) Performance—This system responds fast to users without any delay or complications, and (v) User Friendly—The complete system is developed in a standardized way which makes it user-friendly in terms of the interface and ease in terms of understanding its usage.

4.3 GIS design technologies

Geographic coordinates and characteristics were recovered, processed, and deposited on the ArcMap platform. The following are the procedures involved: (a) create a file geodatabase in ArcCatalog, (b) load the CSV data into the file geodatabase, (c) import the data into the ArcMap environment, (d) change the symbol to represent the points imported, (e) creation of the domains and feature classes of the assets to be collected, (f) configuration of the data collection form, (g) customization of the application settings, and (h) publishing and sharing of the form and map for data collection.

The ArcDiagrammer's procedure for creating and defining classes and domains is illustrated visually in Figure 4. For ultimate exporting into ArcCatalog, configurations were made to enable efficient data structure and utilization as this early stage was crucial. Classes and domains were created and specified within the ArcDiagrammer interface, acting as the essential building blocks of the data structure. To categorize and organize the data, it is required selecting domain categories and assigning coded values. Additionally, the procedure made it possible to enable particular data types, which was crucial for ensuring a correct and consistent data representation. Figure 5 illustrates the Geo-Lecture App collection form for data input by a user (student), which is populated to display the collected data in a table view illustrated in Figure 6.

Figure 4.

Data schema configured in ArcDiagrammer.

Figure 5.

Geo-Lecture app form collection interface.

Figure 6.

Geo-Lecture depicting data in table view.

To address concerns of identity theft, the system integrates biometric security measures such as facial recognition. This biometric method ensures that only enrolled students can register attendance, eliminating the risk of unauthorized access from someone using another person's device. For future iterations, these enhancements provide a critical security layer, aligning with the Geo-Lecture system's goal to secure identity authentication effectively while maintaining ease of access across mobile and desktop platforms. This approach underscores the importance of reliable identity verification in GIS-based attendance tracking.

5 Performance evaluation

In this section, we discuss the evaluation performance of our proposed Geo-Lecture method. We present the data we used during our evaluation, compared baseline methods, evaluation metrics, and evaluation results. We designed a set of questionnaires using Google Forms and sent them to students in a Technical University in Ghana. We purposively sampled respondents, consisting of 215 participants (students). Data collected in terms of gender reveals that among the cherished participants, a graceful majority, 63%, identified as Male, while 37% as Female, which is illustrated in Table 1.

Table 1.
Gender of participants.

Question Variable N

Gender Male 135 (63%)

Female 80 (37%)

Total 215 (100%)

Question	Variable	N
Gender	Male	135 (63%)
	Female	80 (37%)
	Total	215 (100%)

5.1 Evaluation data

Tables 1 –3 show the results of our evaluation after analyzing the responses of the participants using the Statistical Package for the Social Sciences (SPSS) analytical tool. In our evaluation analysis, we used the following parameters: TN = total number, M = mean, SE = standard error, SD = standard deviation, and V = variance illustrated in Tables 2 and 3. A high mean (M) value denotes the respondents’ favorite category. A low SE also equates to a high M. As a result, the M for a category is invalid if its SE is high. Additionally, SD signifies the dispersion of the data, and V indicates the variation in the mean of the Likert scale for a certain category.

Table 2.
User interaction with geo-lecture.

S/N Question TN M SE SD V

1. Were you able to use the placeholders to navigate the User Interface (UI) when using Geo-Lecture? 215 4.60 0.05 0.72 0.52

2. Is it possible to make changes to the data you have already entered in using Geo-Lecture? 215 4.58 0.05 0.76 0.58

3. Is it possible to monitor your progress using Geo-Lecture? 215 4.44 0.07 1.03 1.05

4. Were you able to visualize your data before sending it to Geo-Lecture for further processing or synchronization? 215 4.44 0.07 1.03 1.05

5. When you log in to Geo-Lecture, do you get a user tutorial on how to utilize it? 215 4.37 0.06 0.89 0.79

6. Were you able to effectively log off Geo-Lecture? 215 4.27 0.06 0.87 0.76

7. Is Geo-Lecture allowing you to acquire the information you need quickly? 215 4.20 0.06 0.83 0.68

8. Were you able to verify yourself when using Geo-Lecture? 215 4.20 0.05 0.71 0.50

S/N	Question	TN	M	SE	SD	V
1.	Were you able to use the placeholders to navigate the User Interface (UI) when using Geo-Lecture?	215	4.60	0.05	0.72	0.52
2.	Is it possible to make changes to the data you have already entered in using Geo-Lecture?	215	4.58	0.05	0.76	0.58
3.	Is it possible to monitor your progress using Geo-Lecture?	215	4.44	0.07	1.03	1.05
4.	Were you able to visualize your data before sending it to Geo-Lecture for further processing or synchronization?	215	4.44	0.07	1.03	1.05
5.	When you log in to Geo-Lecture, do you get a user tutorial on how to utilize it?	215	4.37	0.06	0.89	0.79
6.	Were you able to effectively log off Geo-Lecture?	215	4.27	0.06	0.87	0.76
7.	Is Geo-Lecture allowing you to acquire the information you need quickly?	215	4.20	0.06	0.83	0.68
8.	Were you able to verify yourself when using Geo-Lecture?	215	4.20	0.05	0.71	0.50

Table 3.

Geo-Lecture usability attributes and preferences.

S/N	Question	TN	M	SE	SD	V
1.	Geo-Lecture is appropriate in terms of usability quality.	215	4.62	0.06	0.92	0.85
2.	For navigation and locating of lecture rooms, I prefer to use a Geo-Lecture.	215	4.62	0.06	0.92	0.85
3.	Geo-Lecture is effective in terms of usability aspects.	215	4.53	0.06	0.93	0.87
4.	When using Geo-Lecture are you more satisfied with locating lecture rooms with relevant data in comparison to the traditional approach of physically asking people?	215	4.44	0.07	1.03	1.05
5.	When using Geo-Lecture are you more satisfied with locating lecture rooms with relevant data in comparison to using google maps?	215	4.39	0.05	0.76	0.58
6.	Geo-Lecture is fault tolerant in terms of usability aspects.	215	4.39	0.05	0.76	0.58
7.	In comparison to the old way, I prefer Geo-Lecture for locating lecture rooms with relevant data?	215	4.25	0.10	1.39	1.93
8.	In terms of usability, Geo-Lecture is simple to learn.	215	4.16	0.08	1.14	1.31

In summary, for all the questions (1 to 8) in Table 2, participants generally had a positive perception of Geo-Lecture's functionalities, as indicated by the means being close to 1 (strong agreement). The low standard errors and standard deviations across the questions suggest that the responses of the participants were consistent and not widely varied. This implies that most participants had similar experiences with Geo-Lecture’s features. The variance values further support this observation. Additionally, analytical data in Table 3 indicate that Geo-Lecture is generally well received by users, offering a user-friendly, effective, and satisfactory experience for locating lecture rooms in ATU.

5.2 Baseline methods and evaluation metrics

In terms of our baseline methods, we compared our proposed Geo-Lecture solution with two recently related contemporary methods in literature, namely the work done by Soko (2024) represented as M1, Lala et al. (2022) denoted as M2. In addition to these baseline methods, we performed an experimental comparison of Geo-Lecture with the current ATU Campus Navigation (physical map) of locating lecture rooms denoted as M3 and Google Maps denoted as M4. In relation to evaluation metrics, the precision metric depicts the probability that the selected approach is extremely suitable for a user (Olmo and Gaudioso, 2008; Silveira et al., 2019; Tamm, Damdinov, and Vasilev, 2021). Therefore, Precision (P) is defined as the number of true positives (tp) over the number of tp plus the number of false positives (fp) as depicted in (1) using Table 3. Higher precision corroborates the higher accuracy of a particular approach. Furthermore, with a focus on quality, the recall metric represents the probability that the selected approach by participants in Table 2 is in the majority (Olmo and Gaudioso, 2008; Silveira et al., 2019; Tamm, Damdinov, and Vasilev, 2021). Recall (R) is therefore defined as the number of true positives (tp) over the number of tp plus the number of false negatives (fn) as depicted in (2). Using (3), the measure F (F1) also known as the harmonic mean (HM) integrates the computed precision and recall values into single values to corroborate the robustness and strength of each approach. Using (4), we further compute the arithmetic mean (AM) of the precision and recall values computed (Olmo and Gaudioso, 2008; Silveira et al., 2019; Tamm, Damdinov, and Vasilev, 2021).

Additionally, we utilized the mean absolute error (MAE) to focus on the prediction accuracy of Geo-Lecture in comparison with the baseline methods. The average variance between the significant values in the dataset and the projected values in the same dataset is known as MAE. Therefore, Precision and MAE can be reformulated using (5), as explained by Olmo and Gaudioso (2008). The MAE is normalized by normalized MAE (NMAE) using (6), which is particularly useful for comparing the MAE of models with various scales. This implies that lower MAE and NMAE values depict higher prediction performance of a recommender algorithm/system (Olmo and Gaudioso, 2008; Silveira et al., 2019; Tamm, Damdinov, and Vasilev, 2021). The dataset was divided into the training set and the test set. We allocated 80% of the data for training and 20% for testing (Anifowose, Khoukhi, and Abdulraheem, 2017).

\begin{aligned} P & = \frac{tp}{tp + fp} \end{aligned}

(1)

\begin{aligned} R & = \frac{tp}{tp + fn} \end{aligned}

(2)

\begin{aligned} F-Measure (F 1) & = \frac{2 \times P \times R}{P + R} \end{aligned}

(3)

\begin{aligned} Aritmetic Mean (AM) & = \frac{P \times R}{P + R} \end{aligned}

(4)

\begin{aligned} MAE & = 1 - Precision \end{aligned}

(5)

\begin{aligned} NMAE & = \frac{MAE}{Mean (Actual\ Values)} \end{aligned}

(6)

To evaluate Geo-Lecture, we answer the following questions:

Using the evaluation metrics above, what is the overall performance of Geo-Lecture in comparison to the other methods?

In terms of the utilized evaluation metrics, what is the operational efficiency and effectiveness of Geo-Lecture in comparison to the other methods?

5.3 Evaluation results and analysis

In terms of precision illustrated in Figure 7, Geo-Lecture is more exact, which corresponds to high precision values. Referring to Figure 7, for most of the evaluation coefficients, Geo-Lecture achieved a higher precision compared to that of M1, M2, M3, and M4. This is an indication that Geo-Lecture is more favorable and preferable for students in comparison with the baseline methods. Additionally, as shown in Figure 7, at the highest evaluation coefficient, Geo-Lecture achieved a higher precision value of 0.31 compared to M1 (0.29), M2 (0.27), M3 (0.19) and M4 (0.17). In these case studies, Geo-Lecture was able to execute a higher coverage of lecture room locations within the pool compared with the other illustrated methods.

Figure 7.

Methodological performance on the dataset in terms of precision.

Referring to Figure 8, at the highest evaluation coefficient, Geo-Lecture attained a higher recall value of 0.34 compared to M1 (0.19), M2 (0.17), M3 (0.16) and M4 (0.14). Consequently, Geo-Lecture was able to carry out a greater coverage of lecture room locations and recommendations within the pool of classroom venues for lectures. After computing the results of precision and recall metrics, we further computed F-measure (F1). The experimental results of F1 are illustrated in Figure 9. Figure 9 corresponds to the precision and recall results obtained respectively in Figures 7 and 8. The experimental results shown in Figure 9 show that Geo-Lecture outperformed M1, M2, M3, and M4 in terms of F1. Furthermore, as illustrated in Figure 9, Geo-Lecture achieved a higher F1 value of 0.31 at the highest evaluation coefficient compared to M1 (0.23), M2 (0.21), M3 (0.18) and M4 (0.15). These scenarios demonstrate the robustness and strength of Geo-Lecture in terms of information retrieval of lecture room locations per the dataset. Additionally, in terms of AM, the experimental results shown in Figure 10 illustrate that Geo-Lecture performed better than all the baseline methods. This validates the strength and suitability of Geo-Lecture in terms of information retrieval of lecture room locations per the dataset. Our method confirms that, compared to the finding of F1 (harmonic mean), the AM results obtained for Geo-Lecture were the highest. This supports the idea that AM should always be higher than F1 to corroborate the effectiveness and precision of a recommender algorithm (Baykan et al., 2011). Summarized experimental results pertaining to precision, recall, F1, and AM are illustrated in Table 4.

Figure 8.

Methodological performance on the dataset in terms of recall.

Figure 9.

Methodological performance on the dataset in terms of F1.

Figure 10.

Methodological performance on the dataset in terms of AM.

Table 4.

Summarized evaluation performance of P, R, F1, and AM.

Method	Evaluation coefficient	P	R	F1	AM
Geo-Lecture	1.0	0.31	0.34	0.31	0.32
M1	1.0	0.29	0.19	0.23	0.24
M2	1.0	0.27	0.17	0.21	0.22
M3	1.0	0.19	0.16	0.18	0.18
M4	1.0	0.17	0.14	0.15	0.15

The experimental results for Geo-Lecture demonstrated reduced MAE values on most evaluation coefficients, as shown in Figure 11, indicating superior performance when compared to the other methods. Figure 11, respectively, shows that Geo-Lecture had the lowest MAE evaluation coefficients in most cases when compared to M1, M2, M3, and M4, which signifies additional evidence of Geo-Lecture's efficacy over the baseline methods. Similar results for NMAE are illustrated in Figure 12. The MAE and NMAE results for the initial evaluation coefficients in our experiment are summarized in Table 5. The experimental results depicted in Figures 7–12 validate that Geo-Lecture provides very appropriate contextual data such as location, time, and lecture room for effective attendance tracking and monitoring of students. Our experimental findings validate that Geo-Lecture outperforms the other methods in terms of all the evaluation metrics we utilized. The superior performance of Geo-Lecture suggests that a novel blend of geospatial data and an ArcMap can extract valuable insights from users and user groups on social networks to obtain high recommendation accuracy. As a result, our experimental data indicate that a decrease in errors (MAE and NMAE) is correlated with an improvement in accuracy. Furthermore, analytical results of our experiment perfectly align with the idea that students can more appropriately verify their lecture rooms using Geo-Lecture by providing relevant and appropriate geographical data that support the accuracy of the recommendations. Table 6 further illustrates the novelty of Geo-Lecture.

Figure 11.

Methodological performance on the dataset in terms of MAE.

Figure 12.

Methodological performance on the dataset in terms of NMAE.

Table 5.

Summarized evaluation performance of MAE and NMAE.

Method	Evaluation coefficient	MAE	NMAE
	Geo-Lecture	1.0	0.69	0.17
M1	1.0	0.70	0.18
M2	1.0	0.73	0.18
M3	1.0	0.80	0.20
M4	1.0	0.83	0.21

Table 6.

Summary of performance evaluation and novelty.

		Contextual	GIS mapping &	Student–side	Recommender metrics
Method	Approach	filtering	recommendation	automation	evaluated
Google Maps	Geolocation-based web app	None	No	Yes	No
ATU Navigation	Geolocation-based web app	None	No	Yes	No
Lala et al. (2022)	Geolocation-based web app	None	No	Yes	No
Soko (2024)	Lecturer-recorded geofenced system	None	No	Lecturer-side only	No
Geo-Lecture	Student-side + GIS + recommendation	✓	✓	Automated attendance	✓

Moreover, students may access, see, and engage with Geo-Lecture thanks to the developed web-based map that is made available as a service on the ArcGIS platform. With a variety of client apps, Geo-Lecture gives students the ability to access, view, and interact with geographic data. Additionally, Figure 13 displays the dashboard of the Geo-Lecture Administrator, which illustrates data of all students in terms of their attendance trends. In summary, Geo-Lecture's recommendation accuracy varies the least when compared to all the other methods. This demonstrates that Geo-Lecture handles data sparsity and solves issues regarding large scholarly data (Khan et al., 2017; Xia et al., 2017) as well as provision of benefits such as optimum precision monitoring and tracking, user-friendly navigation, reliability, positive usability and cost-effectiveness. Furthermore, Geo-Lecture demonstrates an appealing feature in that it achieves excellent accuracy even in a limited training set. This saves a ton of time in an experiment when testing Geo-Lecture over a medium-sized subset of the original user-user matrix.

Figure 13.

Geo-Lecture administrator dashboard.

Figure 14 shows a layout of geofenced classrooms of Geo-Lecture, illustrating the importance of georeferencing in attendance monitoring and tracking systems. As shown in Figure 14, each classroom is represented as a colored square, with a surrounding, slightly larger shaded area representing the geofence. This geofenced zone uses geographic coordinates to establish a virtual boundary, which enables the Geo-Lecture platform to recognize when students and instructors are physically present in the designated area. Each student, represented by an orange circle icon, is strategically positioned within their respective classrooms to visually indicate attendance within the geofenced zones. Geofencing ensures that only students within the exact spatial boundaries are recorded as present, mitigating issues related to manual attendance errors or identity theft. By using geofencing boundaries, Geo-Lecture can automatically detect each student's presence and accurately track attendance in real time. This layout underscores the precision offered by geofencing, particularly when classrooms or educational facilities span large, and complex campus environments. In the case of Geo-Lecture, the GIS integration not only provides a detailed spatial reference but also enhances security, efficiency, and accuracy, ensuring that attendance data is robust and reliable.

Figure 14.

Geofenced classroom layout of Geo-Lecture.

5.4 Uniqueness of Geo-Lecture

The novelty of our proposed Geo-Lecture solution lies in its integration of contextual recommendation, automated attendance tracking, and real-time geolocation. This has currently not been addressed holistically in existing systems. Specifically, the uniqueness of Geo-Lecture is as follows:

Fusion of recommender algorithms and GIS for attendance: While GIS is widely used in tourism and logistics, Geo-Lecture is the first known application of a GIS-driven location-based recommender system for tracking student attendance in a higher education context.

Context-aware recommendation for lecture room assignment: Unlike NFC/RFID systems or typical static classroom navigation tools, Geo-Lecture considers context such as location, time, and student identity through CPF to dynamically recommend the correct lecture room and verify presence.

Real-time attendance verification through geofencing: To reduce manual errors and fraud common in signature or card-swipe systems, Geo-Lecture introduces geofenced classroom zones, which automatically record attendance if a student is within a virtual boundary.

Device-agnostic mobile/web access with integrated GIS visualization: In contrast to systems that rely heavily on institutional infrastructure like beacons or biometric kiosks, Geo-Lecture offers a web-based interactive GIS map, accessible on both mobile and desktop platforms, for seamless student navigation.

Empirical evaluation using educational metrics and recommender system key performance indicators (KPIs): The performance evaluation incorporates user (quantitative data) perception (Likert scale) with evaluation metrics (precision, recall, F1, MAE, NMAE). This offers a more robust performance assessment than existing systems which often rely on anecdotal or qualitative data alone.

6 Concluding remarks

In this article, we proposed Geo-Lecture, a location-based recommender algorithm that uses a GIS. Our proposed Geo-Lecture solution solves the present physical approach of student lecture attendance tracking, which is prone to inaccurate analytical and reporting data. We evaluated our proposed Geo-Lecture solution using data from a Technical University in Ghana relating to location and student information. The evaluation findings of our proposed solution demonstrate that, in terms of precision (P), recall (R), F-Measure (F1), MAE, NMAE, and AM, Geo-Lecture performs better than other contemporary approaches/methods.

Our proposed Geo-Lecture solution places a high priority on GIS integration, making use of the technology to map the campus, connect to lecture schedules, and link each student to a specific location. The resultant web-based map is developed with features that make it easy to navigate and view lecture room locations. Students can confirm lecture room directions on the web-based map using a location-based recommender algorithm for mobile devices, which facilitates mobile attendance tracking.

Furthermore, concerning user privacy, our proposed Geo-Lecture solution can deduce a great deal of personal data/information about students such as their characteristics, demographics, and personalities as a result of data collected from their smartphones through the Geo-Lecture server. Through geolocation tracking, students may not be aware that their personal data/information may be disclosed to outside parties like marketers. It is therefore crucial that through the institutional and national data protection policies, the Technical University comprehends the kind of data that can be gleaned from student location data, as well as the implications of user privacy. Furthermore, it is crucial to give accessibility and usability adequate thought, effort, and resources toward our proposed Geo-Lecture solution. Improving the experience of the end-user of Geo-Lecture by fully addressing these issues has a positive impact on the effectiveness of higher technical education in Ghana. In the future, we plan to expand our current data to involve more students to improve the resilience and accuracy of our proposed Geo-Lecture solution.

6.1 Operational Considerations, Limitations and Future Research Directions

We acknowledge some limitations of our proposed Geo-Lecture solution as follows: (a)

Data privacy: The location of students and behavioral data introduce privacy concerns. Consequently, if robust safeguards are not utilized, there can be misuse of data or unauthorized access.

(b)

Scalability: Broader implementation and evaluation across diverse academic environments are needed to confirm generalizability since this study is limited to a single technical university in Ghana.

(c)

Device compatibility: Some students with limited device capabilities device requirements: 2.0 GHz processing power, 15GB storage, and stable WLAN may face accessibility challenges.

(d)

GPS accuracy: The reliance on GPS raises concerns about accuracy, especially in indoor settings where signal degradation is common.

(e)

Stress testing: The current architecture requires further stress testing for high-volume scenarios involving multiple campuses or nationwide systems.

Future research relating to our proposed Geo-Lecture solution will involve: (a)

Conducting multi-institutional pilot deployments to test and validate interoperability and scalability.

(b)

Integrating alternative localization methods such as inertial tracking, Wi-Fi fingerprinting, or Bluetooth beacons for indoor accuracy.

(c)

Developing lightweight and device-agnostic versions of Geo-Lecture to improve compatibility across a range of hardware.

(d)

Implementing stronger privacy-preserving techniques such as anonymization and encrypted data transmission.

(e)

Exploring AI-based predictive analytics to detect attendance patterns of students and lecturers, forecast absenteeism, and personalize recommendations.

Footnotes

Acknowledgments

The authors acknowledge the usage of computing (internet) resources kindly provided by Accra Technical University.

ORCID iDs

Nana Yaw Asabere

Gare Lawson

Funding

This research did not receive specific funding but was performed as part of the employment of the authors, namely Accra Technical University, Ghana, University of Professional Studies, Accra-Ghana, and Université Alioune DIOP de Bambey, Senegal.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Acharya

Yadav

Mohbey

(2023) How can we create a recommender system for tourism? A location centric spatial binning-based methodology using social networks. International Journal of Information Management Data Insights 3(1): 100161.

Adomavicius

Bauman

Tuzhilin

, et al. (2021) Context-aware recommender systems: From foundations to recent developments. In: Recommender Systems Handbook. New York, NY: Springer US, 211–250.

Al-Nafjan

Alrashoudi

Alrasheed

(2022) Recommendation system algorithms on location-based social networks: Comparative study. Information 13(4): 188.

Anifowose

Khoukhi

Abdulraheem

(2017) Investigating the effect of training–testing data stratification on the performance of soft computing techniques: An experimental study. Journal of Experimental and Theoretical Artificial Intelligence 29(3): 517–535.

Apoorv

Mathur

(2016) Smart attendance management using bluetooth low energy and android. In: Proceedings of the IEEE Region 10 Conference (TENCON), pp.1048–1052.

Arulampalam

Naylor

Smith

(2012) Am I missing something? The effects of absence from class on student performance. Economics of Education Review 31(4): 363–375.

Asabere

Lawson

Badu-Marfo

, et al. (2021) Classification of public health centres in Accra through a web-based portal integrated with geographical information system (GIS). Journal of Healthcare Engineering 2021(1): 4178161.

Baykan

Henzinger

Marian

, et al. (2011) A comprehensive study of features and algorithms for URL-based topic classification. ACM Transactions on the Web (TWEB) 5(3): 1–29.

Bruno

Giannikos

(2015) Location and GIS. In: Laporte

Nickel

Saldanha da Gama

(eds) Location Science. Cham: Springer, 509–536.

10.

Clemenson

Maselli

Fiannaca

, et al. (2021) Rethinking GPS navigation: Creating cognitive maps through auditory clues. Scientific Reports 11(1): 7764.

11.

Ding

Jiang

, et al. (2018) Objectives and state-of-the-art of location-based social network recommender systems. ACM Computing Surveys 51(1): 1–28.

12.

Hameed

Saquib

SMT

ul Hassan

, et al. (2015) Radio frequency identification (RFID) based attendance and assessment system with wireless database records. Procedia-Social and Behavioral Sciences 195: 2889–2895.

13.

Huang

Gartner

Krisp

, et al. (2018) Location based services: Ongoing evolution and research agenda. Journal of Location Based Services 12(2): 63–93.

14.

Jin

Nara

Yang

, et al. (2020) Similarity measurement on human mobility data with spatially weighted structural similarity index (SpSSIM). Transactions in GIS 24(1): 104–122.

15.

Khan

Liu

Shakil

, et al. (2017) A survey on scholarly data: From big data perspective. Information Processing and Management 53(4): 923–944.

16.

Khong

Dunn

Lim

, et al. (2016) Why do students attend lectures? Exploring justifications for attendance among undergraduate students from a British University in Asia. The Journal of Developing Areas 50(5): 497–506.

17.

Kumar

Qadeer

Gupta

(2009) Location based services using android (LBSOID). In: Proceedings of the IEEE International Conference on Internet Multimedia Services Architecture and Applications (IMSAA), pp.1–5.

18.

Lala

Aworinde

Ogunbanwo

, et al. (2022) A mobile web application for attendance management using geolocation. Adeleke University Journal of Science (AUJS) 1(2): 459–467.

19.

Antoshchuk

Nguyen

TKT

, et al. (2020) Automated student attendance monitoring system in classroom based on convolutional neural networks. Applied Aspects of Information Technology 3(3): 179–190.

20.

Lian

, et al. (2020) Geography-aware sequential location recommendation. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 2009–2019.

21.

Lodha

Gupta

Jain

, et al. (2015) Bluetooth smart based attendance management system. Procedia Computer Science 45: 524–527.

22.

McMahon

Cihak

Wright

(2015) Augmented reality as a navigation tool to employment opportunities for postsecondary education students with intellectual disabilities and autism. Journal of Research on Technology in Education 47(3): 157–172.

23.

Mohandes

(2017) Class attendance management system using NFC mobile devices. Intelligent Automation and Soft Computing 23(2): 251–259.

24.

Moores

Birdi

Higson

(2019) Determinants of university students’ attendance. Educational Research 61(4): 371–387.

25.

Olmo

Gaudioso

(2008) Evaluation of recommender systems: A new approach. Expert Systems with Applications 35(3): 790–804.

26.

Ravi

Subramaniyaswamy

Vijayakumar

, et al. (2019) Hybrid location-based recommender system for mobility and travel planning. Mobile Networks and Applications 24: 1226–1239.

27.

Riehl

(2020). Property recommendation system with geospatial data analytics and natural language processing for urban land use. Master Thesis, Cleveland State University.

28.

Risimati

Gumbo

(2018) Exploring the applicability of location-based services to delineate the state public transport routes integratedness within the city of Johannesburg. Infrastructures 3(3): 28–41.

29.

Sánchez

Bellogín

(2022) Point-of-interest recommender systems based on location-based social networks: A survey from an experimental perspective. ACM Computing Surveys 54(11s): 1–37.

30.

Saraswat

Garg

(2016) Beacon controlled campus surveillance. In: Proceedings of the International Conference on Advances in Computing, Communications and Informatics (ICACCI), pp.2582–2586.

31.

Seetharaman

Mathew

De’

(2024) Does location matter in IS research? A developing country perspective from India. Information Systems Journal: 1–32. DOI:10.1111/isj.12521.

32.

Silveira

Zhang

Lin

, et al. (2019) How good your recommender system is? A survey on evaluations in recommendation. International Journal of Machine Learning and Cybernetics 10: 813–831.

33.

Soko

(2024) Attendance tracking system using geofencing: A web-based automated solution for higher education. i-Manager’s Journal on Information Technology 13(1): 1–9.

34.

Stojanovic

Djordjevic-Kajan

(2001) Developing location-based services from a GIS perspective. In: Proceedings of the 5th IEEE International Conference on Telecommunications in Modern Satellite, Cable and Broadcasting Service. TELSIKS 2001. Proceedings of Papers (Cat. No. 01EX517), pp. 459–462.

35.

Sultana

Enayet

Mouri

(2015) A smart, location based time and attendance tracking system using android application. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT) 5(1): 1–5.

36.

Tamm

Damdinov

Vasilev

(2021) Quality metrics in recommender systems: Do we calculate metrics consistently? In: Proceedings of the 15th ACM Conference on Recommender Systems, pp. 708–713.

37.

Vasudevan

Venkatachalam

Shree

, et al. (2020) An intelligent and interactive AR-based location identifier for indoor navigation. International Journal of Advanced Intelligence Paradigms 15(1): 32–50.

38.

Westerman

Perez-Batres

Coffey

, et al. (2011) The relationship between undergraduate attendance and performance revisited: Alignment of student and instructor goals. Decision Sciences Journal of Innovative Education 9(1): 49–67.

39.

Xia

Wang

Bekele

, et al. (2017) Big scholarly data: A survey. IEEE Transactions on Big Data 3(1): 18–35.

40.

Yue-Qiang

Xiao-Na

, et al. (2019) A multi-element hybrid location recommendation algorithm for location based social networks. IEEE Access 7: 100416–100427.

41.

Zhao