English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 63753/63753 (100%)
造訪人次 : 18872810      線上人數 : 174
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋

    請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/74791

    題名: 使用者行為分析與商品推薦應用於集點App;User Behavior Analysis and Commodity Recommendation for Point-Earning App
    作者: 陳昱瑾;Chen, Yu-Ching
    貢獻者: 資訊工程學系
    關鍵詞: 推薦系統;機器學習;時間序列性推薦;Structure Learning
    日期: 2017-08-24
    上傳時間: 2017-10-27 14:39:28 (UTC+8)
    出版者: 國立中央大學
    摘要: 近年來網路普及,電子商務網站發展成熟,隨之而生的推薦系統發展蔚為風潮,如何幫助消費者從成千上萬的商品中,找到他所需要的商品,並在最短時間內將「對的商品」在「對的時間」推薦給「對的人」,是很重要的。另一方面,對於商家而言,能將推薦資訊精準的投放給有需要的顧客,為其首要目標。然而,推薦系統背後仰賴的是大量的數據和有效的分析,對於初創發展不久的企業來說是很大的問題。
    在本論文中,我們篩選出消費次數較高的使用者來分析,以解決資料不足的問題。此外,透過設計使用者特徵、商品特徵、使用者與商品的交互特徵以及時間相關的特徵,將資料透過Machine Learning的方法訓練,並建立預測模型,再對以時間切分後的測試資料進行預測。實驗數據顯示在Top5的F-measure由原本的0.0507提升至0.4413,藉此提升推薦系統的效能並進行較為有效的推薦。;The E-commerce website is well developed due to the internet become more popular for the past few year. It is a trend in recommendation system. It is the most important thing, how can we recommend the right thing for the right person at the right moment in the short time. On the other side, the first topic for the store is how can we give an exact information for the customers who needed. However, recommendation system depends on a huge data and analysis. It will be a big problem for a new company.
    This research data is from an App needs to collect point. We find out users’ profile and behavior from APP records and Facebook account. The App is mainly cooperate with a Taiwan makeup chained retailer, that it sells beauty, skin care products and some necessities, they have around 2,177 kinds of item. This kind of consumption recommended is different from the movie recommendation. Consumer product is related to the time and will be bought again by the user.
    In my paper, we select some user, who purchase times is high to resolve the data is not enough. Otherwise, through user feature, item feature, feature between with user and item and related to time. Using Machine Learning to train and built up the forecast model to predicted the testing data by the time. Resulting in F-measure from 0.0507 to 0.4413, Structure leaning with time greatly enhances the efficiency of our recommendation system.
    顯示於類別:[資訊工程研究所] 博碩士論文


    檔案 描述 大小格式瀏覽次數


    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 回饋  - 隱私權政策聲明