A Novel Hybrid Machine Learning Framework to Recommend E-Commerce Products

Marigowda, Chethan (2022) A Novel Hybrid Machine Learning Framework to Recommend E-Commerce Products. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (771kB) | Preview
Preview	PDF (Configuration manual) Download (1MB) | Preview

Abstract

E- commerce is the activity of electronically purchasing or selling products in an online platform. Ecommerce recommender systems provide suggestions of products based on the consumer sentiment and ratings. There is often a mismatch between consumer rating and their sentiment. Identifying the accuracy of the mismatch is a challenge in machine learning. This research proposes a Novel Hybrid Machine Learning Framework to Recommend E-Commerce Products based on consumer sentiment and product descriptions. This proposed frame work combines a text embeddings model, sentiment analysis model and a rating engine. The text embeddings model is implemented using gensim doc2vec for consumer reviews and product descriptions. Further it uses neural networks for capturing the consumer product interactions for collaborative filtering. The sentiment analysis model is implemented by inputting distributed text embeddings into neural networks that are trained to capture content feature of products and sentiment of consumer evaluations. The rating engine is implemented by aggregating several embeddings as attention weights for consumers and products, then outputting the prediction score for the consumer–product interaction. This research makes use of the real-world Amazon product category semi structured baby and digital music semi structured datasets, each of which contains information on consumer reviews and product metadata. Mean absolute error (MAE) and the root mean-square error (RMSE) are considered to evaluate the recommendation performance, thereby measuring the accuracy of prediction ratings. Experimental results on Amazon distinct product dataset with promising accuracy metric MAE and RMSE values by 0.5909 and 0.8080 respectively demonstrates that proposed framework performs better on rating prediction in enhancing consumers experience to find their preferences over the E-commerce products.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Stynes, Paul UNSPECIFIED
Uncontrolled Keywords:	Hybrid recommendation, Consumer review, Product description, Text embeddings; Machine learning; Neural networks; Collaborative filtering
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science H Social Sciences > HF Commerce > Electronic Commerce Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HF Commerce > Marketing
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	19 May 2023 16:35
Last Modified:	19 May 2023 16:35
URI:	https://norma.ncirl.ie/id/eprint/6614

Actions (login required)

View Item