本文選題：用戶行為 + 加權(quán)軌跡數(shù)據(jù)集　； 參考：《東南大學(xué)》2016年碩士論文

【摘要】：近年來互聯(lián)網(wǎng)經(jīng)濟快速發(fā)展,B2B電商平臺連接生產(chǎn)企業(yè)與消費企業(yè),在互聯(lián)網(wǎng)經(jīng)濟中發(fā)揮著日益重要的作用。生產(chǎn)企業(yè)在平臺發(fā)布的產(chǎn)品描述關(guān)鍵詞直接影響平臺用戶對該產(chǎn)品的檢索效果,對生產(chǎn)企業(yè)的經(jīng)營至關(guān)重要。如何提升生產(chǎn)企業(yè)所發(fā)布產(chǎn)品描述關(guān)鍵詞的受關(guān)注度直接體現(xiàn)電商平臺的服務(wù)效果。論文工作結(jié)合電商平臺實際需求,對基于用戶行為的產(chǎn)品關(guān)鍵詞優(yōu)化進行研究,并開發(fā)原型系統(tǒng)。實現(xiàn)對目標(biāo)產(chǎn)品關(guān)鍵詞優(yōu)化推薦,以期促進電子商務(wù)平臺服務(wù)的個性化和定制化,提升服務(wù)質(zhì)量。主要工作如下：(1)電商平臺積累了大量用戶訪問信息,結(jié)合挖掘用戶搜索行為模式主題,確定目標(biāo)數(shù)據(jù)源,并進行預(yù)處理,提出一種加權(quán)軌跡數(shù)據(jù)集構(gòu)建方法,兼顧數(shù)據(jù)規(guī)模的同時體現(xiàn)用戶行為模式信息,生成挖掘數(shù)據(jù)集；(2)結(jié)合用戶特征及其搜索產(chǎn)品類別對用戶進行分類,設(shè)計頻繁項集挖掘方法獲取各類用戶搜索主題詞與其所關(guān)注產(chǎn)品關(guān)鍵詞間關(guān)聯(lián)模式。并從語義相關(guān)性和產(chǎn)生式前后件關(guān)聯(lián)角度,設(shè)計聚類算法生成關(guān)鍵詞關(guān)聯(lián)詞庫和熱詞庫；(3)結(jié)合所構(gòu)建的關(guān)聯(lián)詞庫和熱詞庫,設(shè)計基于加權(quán)k近鄰思想的關(guān)鍵詞優(yōu)化方法,實現(xiàn)對目標(biāo)產(chǎn)品關(guān)鍵詞優(yōu)化推薦。并設(shè)計評估機制對基于詞庫的產(chǎn)品關(guān)鍵詞優(yōu)化效果進行評估,驗證所提方法的有效性。在此基礎(chǔ)上,實現(xiàn)基于用戶行為的產(chǎn)品關(guān)鍵詞優(yōu)化原型系統(tǒng)。幫助用戶合理制定產(chǎn)品描述關(guān)鍵詞,提高產(chǎn)品的受關(guān)注度。目前,系統(tǒng)已通過委托研發(fā)驗收,處于部署應(yīng)用中。
[Abstract]:In recent years, the rapid development of Internet economy, B2B ecommerce platform connecting production enterprises and consumer enterprises, plays an increasingly important role in the Internet economy. The key words of product description issued by the production enterprise directly affect the retrieval effect of the product by the platform users, and are very important to the management of the production enterprise. How to improve the attention of product description keywords published by manufacturing enterprises directly reflects the service effect of e-commerce platform. According to the actual requirements of e-commerce platform, this paper studies the optimization of product keywords based on user behavior, and develops a prototype system. In order to promote the individualization and customization of E-commerce platform service and improve the service quality, we can optimize and recommend the target product keyword. The main work is as follows: (1) the ecommerce platform accumulates a large amount of user access information, combines with mining user search behavior pattern topic, determines the target data source, carries on the preprocessing, proposes a weighted track data set construction method. Taking into account the data scale and reflecting the user behavior pattern information, the mining data set is generated to classify the user in combination with the user characteristics and their search product categories. The frequent itemsets mining method is designed to obtain the associated patterns between the various user search subject words and their concerned product keywords. And from the point of view of semantic correlation and production relation, we design the clustering algorithm to generate keyword association lexicon and hot lexicon, and design a keyword optimization method based on weighted k-nearest neighbor thought combined with the related lexicon and hot lexicon. To achieve the target product keyword optimization recommendation. An evaluation mechanism is designed to evaluate the effect of product keyword optimization based on thesaurus to verify the effectiveness of the proposed method. On this basis, the prototype system of product keyword optimization based on user behavior is implemented. Help users to develop product description keywords, improve product attention. At present, the system has been commissioned by R & D acceptance, in the deployment of applications.
【學(xué)位授予單位】：東南大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：TP311.13

【參考文獻(xiàn)】

相關(guān)期刊論文 前7條

1 毛佳昕;劉奕群;張敏;馬少平;;基于用戶行為的微博用戶社會影響力分析[J];計算機學(xué)報;2014年04期

2 姚婷;張敏;劉奕群;馬少平;茹立云;;低頻查詢的用戶行為分析和類別研究[J];計算機研究與發(fā)展;2012年11期

3 程啟月;;評測指標(biāo)權(quán)重確定的結(jié)構(gòu)熵權(quán)法[J];系統(tǒng)工程理論與實踐;2010年07期

4 宋威;李晉宏;徐章艷;楊炳儒;;一種新的頻繁項集精簡表示方法及其挖掘算法的研究[J];計算機研究與發(fā)展;2010年02期

5 雷小鋒;謝昆青;林帆;夏征義;;一種基于K-Means局部最優(yōu)性的高效聚類算法[J];軟件學(xué)報;2008年07期

6 楊艷;李建中;高宏;;數(shù)字圖書館系統(tǒng)中基于Ontology的用戶偏好模型[J];軟件學(xué)報;2005年12期

7 邢東山,沈鈞毅,宋擒豹;從Web日志中挖掘用戶瀏覽偏愛路徑[J];計算機學(xué)報;2003年11期

，

本文編號：2026931