COMPARISON OF RECOMMENDATION SYSTEMS IN EDUCATIONAL MANAGEMENT
Mukhammadsolayev Akbar, 2Masharipov San'atbek, 3Iskandarov Sanjar, 4Sharipov
Sirojbek
1,4 2nd stage master's students at Urgench branch of TUIT named after Muhammad al-Khorazmi 2The head of the Department of Software Engineering at Urganch branch of TUIT named after
Muhammad al-Khorazmi 3Associate professor of the Department of Information Technologies of the Urganch branch of
TUIT named after Muhammad al-Khorazmi https://doi.org/10.5281/zenodo.7855429
Abstract. Recommender systems have been an important research topic in recent years, especially in the field of educational management. Recommender systems can be used to provide personalized recommendations to students on what to learn next, based on their past activities and performance. This can help to improve student engagement and academic performance. There are various types of recommender systems, including content-based, collaborative filtering, and hybrid recommender systems.
In this paper, we aim to compare and evaluate the performance of different recommender systems in educational management. We will use real-world educational data to evaluate the accuracy, coverage, and novelty of each recommender system. The goal is to provide insights into which recommender systems are most effective for educational management and to identify areas for future research.
Keywords: recommender systems, collaborative filtering, hybrid recommender systems, demographic recommender systems, community-based, artificial intelligence.
INTRODUCTION
A recommendation system is a type of information filtering system that predicts a user's preferences or interests and recommends items or content that the user is likely to enjoy or find useful. Recommendation systems are widely used in e-commerce, social media, content streaming platforms, and many other applications where there is a large amount of data to be processed and users need help discovering relevant items. [1]
There are several types of recommendation systems, including content-based filtering, collaborative filtering, and hybrid approaches that combine the two. Content-based filtering systems analyze the characteristics of items that a user has interacted with in the past, such as their genre or category, and recommend similar items. Collaborative filtering systems, on the other hand, use the preferences of other users who have similar tastes to generate recommendations. Hybrid systems combine these two approaches to improve the quality and accuracy of recommendations.
The concept of recommendation systems has been around for a long time, but the first recorded instance of a recommendation system dates back to the early 1990s when the GroupLens [2] project was initiated by researchers at the University of Minnesota. The project was aimed at developing a personalized recommender system for Usenet news articles. Since then, many researchers and companies have worked on developing recommendation systems for various applications.
Comparison of recommendation systems in educational management
There are several algorithms that are used in modern recommendation systems, such as Collaborative Filtering (CF) Recommender Systems, Knowledge-Based (KB) Recommender Systems, Demographic Recommender Systems, Community-Based Recommender Systems, and Hybrid Recommender Systems. These algorithms allow for personalized and effective recommendations to be generated for users based on their preferences, behaviors, and other
Figure 1. Collaborative Filtering
Collaborative Filtering (CF) Recommender Systems is an algorithmic approach to making recommendations based on the preferences of users. It works by analyzing the behavior and preferences of a group of users and using that data to recommend items to other users with similar tastes. CF Recommender Systems can be further divided into two categories: user-based and item-based. User-based CF recommends items to a user based on the behavior of similar users, while item-based CF recommends items to a user based on the similarity of the items they have interacted with in the past. CF Recommender Systems are widely used in e-commerce, social media, and entertainment industries. Collaborative Filtering (CF) Recommender Systems are a type of recommendation system that suggest items or products to users based on their similarities and preferences to other users. CF systems work by analyzing user behavior and building a model of their preferences to generate recommendations for items that are most likely to be of interest to them. There are two main types of CF systems: user-based and item-based [3].
User-based CF systems recommend items to a user based on the preferences of other users with similar tastes, while item-based CF systems recommend items based on the similarity of the items themselves. CF systems are widely used in e-commerce, social networks, and online media platforms to provide personalized recommendations to users. Content-based recommendation systems are a type of recommendation system that suggest items to users based on their preferences and past interactions with similar items. In content-based systems, the recommendation is made by analyzing the features of the items that a user has interacted with or shown interest in, and recommending similar items that share similar features.
Video watched by two users
Figure 2. Content-based recommendation system For example, if a user has watched several action movies, a content-based system may recommend other action movies that share similar themes, directors, or actors. Similarly, if a user has read several romance novels, a content-based system may recommend other romance novels that share similar plot elements, authors, or settings.
Content-based [4] systems often rely on machine learning algorithms to analyze and classify the features of items, and to make personalized recommendations based on a user's past interactions with similar items. They are particularly effective for recommending items within a specific domain or genre, such as movies, music, or books.
Hybrid recommendation systems are a combination of two or more recommendation techniques such as collaborative filtering, content-based filtering, and/or other approaches [5].
In a hybrid system, the idea is to leverage the strengths of each technique to provide better and more personalized recommendations to users. For example, a hybrid system could use collaborative filtering to identify items that are popular among similar users and then use content-
Figure 3. Hybrid recommendation systems [6]
In order to evaluate the effectiveness of recommendation systems in the context of educational recommenders, we have created a comparison table for the three main types of recommendation systems: hybrid, content-based, and collaborative filtering. Our goal is to assess the strengths and limitations of each system and determine which one is best suited for use in educational settings.
To create the comparison table, we analyzed each type of recommendation system based on several key factors. These factors include the ability to handle new items without user feedback, the provision of personalized recommendations based on user interests and preferences.
Recommen Advantages Disadvantages Source of Knowledge Type of Expected
der System Knowledge Range of Accuracy Percenta ge
Content- l.Handles new l.Limited to Uses item features or Uses explicit 60-80%
Based items without recommending characteristics (e.g., knowledge
any user ratings items similar to genre, author, about the
or feedback. the user's past director, actors,etc.) item features
2. Provides interactions or to recommend similar or
personalized preferences. items to the user. characteristi
recommendations 2. Cannot cs to
based on user's capture a user's recommend
interests and current tastes similar items
preferences. and interests. to the user.
3. Helps 3. Cannot
overcome the recommend
cold-start novel items
problem. outside the user's current interests.
Collaborati 1. Provides 1. Needs a Uses user-item Uses 70-90%
ve-Based accurate significant interaction data (e.g., implicit
recommendations amount of data ratings, reviews, knowledge
based on user to provide clicks, purchases, about user
behavior and accurate etc.) to find preferences
feedback. recommendatio similarities between and behavior
2. Can ns. users and recommend based on
recommend 2. Can be items that similar their
novel items biased towards users have liked or interactions
based on the popular items interacted with. with items to
preferences of or users. recommend
similar users. items that
3. Can capture similar users
changes in a have liked or
user's tastes and interests. interacted with.
Hybrid- 1.Combines the 1. More Combines both item Combines 80-95%
Based strengths of both complex than features and both explicit
content-based either content- user-item interaction knowledge
and based or data to provide more about
collaborative- collaborative- accurate and diverse item
based based systems. recommendations. features and
recommender 2.Requires For example, it may implicit
systems. more use item features to knowledge
2. Provides more computational recommend similar about user
accurate and resources.3. items, and then use preferences
diverse Can be difficult collaborative filtering and behavior
recommendations to integrate and to further refine the to
implement. recommendations provide
3. Helps based on the user's more
overcome the interactions with accurate and
limitations of those items. diverse
each individual recommenda
system. tions.
After evaluating each system on these factors, we assigned a percentage score to each type of recommendation system based on its performance in each area. We found that the hybrid recommendation system performed the best overall, with a score range of 80-95%. However, it should be noted that each system has its own strengths and limitations, and the best choice of system depends on the specific needs and requirements of the educational recommender in question [7] and [8].
DISCUSSION
The effectiveness of these recommendation systems in educational management will depend on several factors, including the quality and quantity of data available, the algorithms used, and the user interface and experience design of the system.
If we rely on short-term requirements, it may result in a significant burden on the student to obtain separate test assignments from each individual student, and it is important for the result to be a valuable recommendation for the student. On the other hand, the recommendation system should provide recommendations based on analyses of the results obtained from the student's own performance so far, or by comparing the student's results with those of other students in the system.
In terms of educational recommender systems, research has shown that hybrid recommender systems perform better than content-based or collaborative filtering systems alone. This is because hybrid recommender systems are able to provide more accurate and personalized recommendations by combining the strengths of both approaches. However, the effectiveness of a recommender system depends on many factors, such as the quality of the data, the algorithm used, and the context in which the system is used. Therefore, further research is needed to compare the effectiveness of different types of recommender systems in educational settings.
CONCLUSION
In conclusion, educational recommender systems have the potential to greatly enhance the learning experience by providing personalized recommendations for learners. Three main types of recommender systems are commonly used in educational settings: content-based, collaborative filtering, and hybrid systems. While each approach has its strengths and weaknesses, research suggests that hybrid recommender systems perform better than content-based or collaborative filtering systems alone, as they are able to combine the strengths of both approaches to provide more accurate and personalized recommendations. However, the effectiveness of a recommender system depends on many factors, including the quality of the data, the algorithm used, and the context in which the system is used. Further research is needed to compare the effectiveness of different types of recommender systems in educational settings, and to explore ways to improve the accuracy and usability of these systems for learners and educators alike. Overall, the field of educational recommender systems holds great promise for improving the learning experience, and continued research and development in this area is essential for realizing this potential.
REFERENCES
1. https://en.wikipedia.org/wiki/Recommender_system
2. https://international.binus.ac.id/computer-science/2020/11/03/definition-and-history-of-recommender-systems
3. Hybrid Educational Recommender System, Utilizing Student, Teacher and Domain Preferences and Amazon Alexa Conversation Engine
4. https://link.springer.com/chapter/10.1007/978-3-540-72079-9_10
5. https://towardsdatascience.com/introduction-to-recommender-systems-6c66cf15ada
6. https://in.pinterest.com/pin/125960120818885384/
7. https://www.researchgate.net/publication/356360025_A_Comparing_Collaborative_Filterin g_and_Hybrid_Recommender_System_for_E-Commerce
8. A Hybrid Movie Recommender System and Rating Prediction Model International Journal of Information Technology and Applied Sciences ISSN (2709-2208) Int. J. Inf. Tec. App. Sci. 3, No.3 (July-2021)