Một số phương pháp gợi ý và ứng dụng trong thương mại điện tử

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Thị Hà Hoàng, Nguyễn Thức Ngô

Ngôn ngữ: Vie

Ký hiệu phân loại: 330 Economics

Thông tin xuất bản: Khoa học nông nghiệp Việt Nam 2021

Mô tả vật lý: 520-534

Bộ sưu tập: Metadata

ID: 397337

Thêm vào giỏ Liên kết toàn văn

Tóm tắt
Chủ đề

Hệ gợi ý (Recommender System) là công cụ được thiết kế nhằm cung cấp những khuyến nghị hữu ích về sản phẩm, dịch vụ,...cho người dùng. Hệ gợi ý dựa trên dữ liệu về mối quan hệ giữa người dùng, sản phẩm và các hành vi của người dùng trong quá khứ đối với sản phẩm để đưa ra những gợi ý thông minh, phù hợp với sở thích của từng khách hàng. Hệ gợi ý giúp khách hàng nhanh chóng định vị được những sản phẩm họ quan tâm để từ đó đưa ra quyết định đúng đắn khi mua sắm online. Trong bài báo này, chúng tôi trình bày tổng quan một số phương pháp gợi ý, đánh giá điểm mạnh, điểm yếu, so sánh hiệu quả thực hiện của mỗi phương pháp. Chúng tôi chỉ ra lợi ích mà các hệ gợi ý mang lại cho thương mại điện tử, đồng thời nêu ra những thách thức và giải pháp khắc phục. Kết quả thực nghiệm của chúng tôi trên 4 tập dữ liệu chuẩn (Movielens, Epinions, BookCrossing, LastFM) cho thấy mỗi phương pháp đều có những ưu điểm và hạn chế riêng, không có phương pháp nào là tốt nhất trên tất cả các tiêu chí. Ngoài ra, chúng tôi cũng đưa ra quy trình chung để xây dựng hệ gợi ý trong các website thương mại và thực hiện tích hợp các kỹ thuật hệ gợi ý trong website thương mại điện tử khắc phục vấn đề người dùng mới, sản phẩm mới (vấn đề này còn gọi là "Cold start problem") của các phương pháp gợi ý cá nhân hóa., Tóm tắt tiếng anh, The designed recommender system is a tool to provide important suggestions forusers or customers. Based on the datasets of user relationships, products, andprevious behavior of consumers, smart recommendations for the preferences of eachconsumer are given, which helps consumers tomake good decisions while shopping online. In this article, we present an overview of some methodologies of recommendation systems, techniques of recommender systems, and evaluate the strengths and weaknesses of each technique, as well as comparedsome benefits of recommender systems in e-commerce. Moreover, we report some challenges that the recommender systems are facing and list some solutions to solve these challenges. Our experimental results on the four datasets (Movielens100k, Epinions,BookCrossing, LastFM) showed that there was no best recommendation algorithm in all evaluation metrics. Finally, we built an e-commerce website that integrated some different techniques of recommender systems such as non-personalized methods, personalized methods to recommend the right product for each customer. The experimental system gives some diverse suggestions to overcome the problem "Cold start problem" of personalized methods.

Tạo bộ sưu tập với mã QR