2.
LITERATURE REVIEW
A recommendation system is based on an algorithm that can learn the user's preferences and
automatically suggest a feature or a product to the users(Chen et al., 2008). These systems are
prevalent nowadays because they assist the sellers on how to keep the customers engaged by
providing the content, products, or features which they would be interested in. ("How Do
Recommendation Engines Work? What Are the Benefits?" 2021; Liu et al., 2021)
Several techniques have been developed over time to create different algorithms which enhance
the recommendation systems. In 1993, "Association Rules Mining (ARM)" was introduced by
Agarwal. This technique proposed the famous Apriori algorithm, used to find relations between
different items in databases. According to this, Alex Tze Hiang Sim, Hosein Jafarkarimi1, and
Robab Saadatdoost came up with a 'Proposed model.'In this model, they proposed to pre-process
the available data, use FP growth to perform ARM, store them in a sparse matrix and use the
matrix to find relations to make a recommendation. (Jafarkarimi1 et al., 2012)
Furthermore, the most popular technique for recommendation is the "Collaborative Filtering
System,"which uses user's preferences to recommend content, products, and features. In contrast,
content-based filtering recommends the viewers products and content based on their past
activity. Asikis and G. Lekakos developed a "Collaborative Metaheuristic Algorithm
(CMA)"based on the collaborative filtering system, which could be used as a metaheuristic and a
constructive algorithm. CMAadded items based on their combined scores to construct a solution
from zero.(Asikis & Lekakos, 2014)
These traditional methods were built keeping in mind the preference of the customers and
neglected profit for the seller. ThereforeMu-Chen Chen, Fei-Hao Hsu, Long-Sheng Chen, Mu-
Chen Chen, and Yuan-Chia Hsu proposed two models, which are the "Hybrid Perspective
Recommender System (HPRS)"and the "Convenience plus Profitability Perspective
Recommender System (CPPRS)."The latter technique recommends products based on purchase
probability and product profitability. Hybrid Perspective Recommender System (HPRS) is a
technique that uses both the purchase probability of similar customers and the product
profitability to make recommendations. Some experiments conducted further indicated that
HPRS can generate profit while keeping more accurate recommendations.(Chen et al., 2008)
ISSN: 2278-4853 Vol 10, Issue 9, September, 2021 Impact Factor: SJIF 2021 = 7.699
Do'stlaringiz bilan baham: |