Article Preview
Top1. Introduction
In this fast-paced digital world, everyone is drifting towards electronic resources for getting their stuff done, be it books, movies, or an entire learning system. The term E-Learning is a formal teaching methodology which uses electronic resources as their parameter. It just requires you an internet connection which is not difficult to find in this digitalized world. E-Learning is nowadays very popular among people as it allows them to learn the topics from the best-in-class faculties in the world without any discrimination. Unlike traditional learning which requires you to sit in a class for a fixed amount of time, you can study through E-Learning from any place you want and at any pace. It provides a lot of conveniences, that's why it is widely adopted everywhere.
With an increasing population dependent on electronic resources for reading, it would be highly convenient if the website or app itself recommend you the new topics/courses based on specific attributes so that you don't have to search for it every time. Many a time since the E-Learning platform is not well organized, it isn't straightforward for a student to find the next appropriate course. There are lots of advice you can find on the internet, and most of the time it confuses the student. It was a significant disadvantage of E-Learning which is why people still relied on traditional classroom learning methods as their teacher used to tell what you should study after this course. If studied in a mismatch course, It can lead to a lack of motivation for a student to study the entire subject sometimes.
This problem is solved by a course recommender system which takes various parameters into account and then suggests the course. A course recommender system is a subset software based on information filtering concept. It customizes the needs of a student and shows the most relevant courses for an individual and thus creating a personal learning environment. It uses efficient information retrieval techniques for this which can be seen in many other fields too, like Netflix (movie recommendation), YouTube (video recommendation), eCommerce sites.
Based on the various parameters taken into account for the recommendation, it can be divided into three categories: content-based, collaborative, and knowledge-based. Collaborative recommender system takes preferences that people that have a similar liking in the past have a similar liking in future too. The concept of content-based systems is based on the presumption that people who like an item with a particular attribute will also like the same attributes in the future while knowledge-based takes the data of a person to recommend it the suggestions. It is more accurate but takes a lot of data from the consumer, and hence in this privacy concerned world, it is not preferred much. There is one more category of recommendation system, which is formed by combining two or more above types in order to maximize the accuracy of suggestion and reducing the disadvantages. This is called the Hybrid recommender system. This research will focus on Hybrid recommender system.
The following are the novel Propositions in the Proposed Work:
- 1.
The usage of the Web Usage Data of the user requesting the courses for recommendation and also the collective requirements and interests of similar users are taken into consideration.
- 2.
Using a Spectral Clustering technique for grouping the profiles of similar users such that a collective intelligence of individual user profiles can be harvested.
- 3.
The usage of LSTM by appending the features from Semantic Networks formulated based on the User Query and Current User Clicks and Enriching it based on the real-world knowledge form Wikidata is one of the novel contributions.
- 4.
Also, collectively imbibing both clustering and a classification into a single framework and transforming the approach based on knowledge harbored from the external knowledge stores makes it quite novel.
The remainder of the paper is formatted as follows. Section 2 illustrates the Related Work. In Section 3, the problem definition is explained. Section 4 depicts the Proposed Methodology. Section 5 discusses implementation and performance evaluation. Section 6 brings the paper to a conclusion.