Please use this identifier to cite or link to this item:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/64612
Title: | 基於交易紀錄自動發掘使用者購買喜好
以增進產品搜尋的效能 Automatically Mining User Preference from Transaction Log for Improving Performance of Product Search |
Authors: | Keng-Wei Cheng 鄭耿維 |
Advisor: | 鄭卜壬 |
Keyword: | 使用者偏好,產品搜尋,以內容為基礎,推薦系統,未標示的資料, User preference,Product search,Content-based,Recommendation,Unlabeled data, |
Publication Year : | 2012 |
Degree: | 碩士 |
Abstract: | 在資訊爆炸的時代,一個商業平台若想要能夠在眾多的競爭中異軍突起,就必須要能夠更快速的捕捉到使用者的偏好,雖然說在使用者的基本資料並不會訴說他們本身具有甚麼樣的偏好,不過在很多商業平台上其實都有紀錄著使用者的交易紀錄,藉由商品特徵中所隱含的概念我們便可間接的探知使用者的偏好,但是在目前的一些平台上,通常都沒有特意藉由交易記錄去了解使用者的需求,在產品搜尋以及產品推薦方面通常都只藉由單一的準則(價格或時間)。雖然過往有些研究也試著想去捕捉使用者的喜好,但是它們都存在著一些問題,有些是藉由關鍵字來表示而沒有去捕捉其中的語意,有些雖然有考量語意但是卻需要花費大量的人工。因此在我們的研究中,我們提出了一個方法,藉由這個方法我們可以自動化地從交易紀錄去挖掘出商品特徵所隱含的概念,而這就間接代表了使用者的偏好,最後我們便可以藉由這些知識為使用者改善產品搜尋的結果,進而提升商業平台的競爭能力。 Information nowadays increase rapidly, a commerce site should capture the preference of user as soon as possible if he want to be a good one in the world. Although user rarely describe his preference in his personal information, but most commerce site has stored the transaction log of user, so we can mine the preference of user from the transaction logs. But most of them do not utilize the rich resource, their recommendation or product search are only based on single criteria , like popularity, price, or time…etc. Some previous research also try to mine the preference from the transaction logs, but they all exist some problems, some of them are keyword-based representation, this method do not consider the semantic meaning of word. Some are type-based representation, although this method has considered the semantic meaning of word, but they need to spend a lot of time to analysis the description of product. We propose a method which can automatically mine the concept of product feature, and use it to represent user preference. We can finally use the knowledge we have mined to improve the performance of product search. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/64612 |
Fulltext Rights: | 有償授權 |
Appears in Collections: | 資訊工程學系 |
Files in This Item:
File | Size | Format | |
---|---|---|---|
ntu-101-1.pdf Restricted Access | 827.43 kB | Adobe PDF |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.