基于數(shù)據(jù)挖掘的商品資訊投送平臺(tái)研究與實(shí)現(xiàn)
發(fā)布時(shí)間：2018-10-08 15:46
【摘要】：隨著電子商務(wù)網(wǎng)站信息量的迅速增長(zhǎng),給消費(fèi)者獲取感興趣的商品信息帶來(lái)了困難,賣家也面臨著很難將商品信息精準(zhǔn)發(fā)送給目標(biāo)客戶的困境,而這些困難成為制約電子商務(wù)網(wǎng)站持續(xù)發(fā)展的問(wèn)題。為解決上述問(wèn)題,行業(yè)提出了商品個(gè)性化推薦技術(shù)。目前個(gè)性化推薦技術(shù)的推薦對(duì)象基本以商品本身為主,很少關(guān)注如商品打折、商品廣告、商品促銷等類似新聞的商品資訊文本信息。而向消費(fèi)者發(fā)布這些資訊信息都是線下商店推廣商品很重要的方式,對(duì)消費(fèi)者購(gòu)買(mǎi)行為有很強(qiáng)的影響。針對(duì)該問(wèn)題,本課題構(gòu)建一個(gè)商品資訊投送平臺(tái),幫助賣家將商品的資訊準(zhǔn)確投送給目標(biāo)客戶。 針對(duì)商品資訊投送平臺(tái)的需求,本論文提出了一個(gè)基于潛在語(yǔ)義索引的商品資訊精準(zhǔn)投送模型。通過(guò)分析資訊投送的基本過(guò)程,很顯然影響精準(zhǔn)匹配的最關(guān)鍵問(wèn)題是推薦策略,而推薦策略中最重要的部分是信息表示模型。相比其它信息表示模型,潛在語(yǔ)義索引模型考慮到文本的語(yǔ)義結(jié)構(gòu),并有效地解決了同義詞和多義詞的問(wèn)題,在信息表示的準(zhǔn)確方面比其它幾種模型都更具優(yōu)勢(shì)�；跐撛谡Z(yǔ)義索引的商品資訊投送模型是對(duì)大量訓(xùn)練文本進(jìn)行預(yù)處理,將這些文本集表示成詞-文本矩陣,利用奇異值分解將詞-文本矩陣從高維矩陣降到低維矩陣,形成潛在語(yǔ)義空間。生成的潛在語(yǔ)義空間保留了原詞-文本矩陣的潛在語(yǔ)義關(guān)系,同時(shí)去掉了因具體用詞方式不同而帶來(lái)的噪聲信息,從而提高商品資訊投送的精準(zhǔn)度。除此之外,消費(fèi)者需求模型反映消費(fèi)者的興趣,對(duì)商品資訊投送精準(zhǔn)度有很強(qiáng)的影響,本論文通過(guò)分析消費(fèi)者需求模型的不足,利用本體論改進(jìn)消費(fèi)者需求模型表達(dá)消費(fèi)者需求的能力,從而進(jìn)一步提高商品資訊投送的精準(zhǔn)度。 本論文圍繞實(shí)現(xiàn)商品資訊精準(zhǔn)投放的目標(biāo),創(chuàng)造性地提出了基于潛在語(yǔ)義索引的商品資訊精準(zhǔn)投送模型。除此之外,本論文還將該商品資訊精準(zhǔn)投送模型應(yīng)用到商品資訊投送平臺(tái)系統(tǒng)的構(gòu)建中,并結(jié)合面向?qū)ο笤O(shè)計(jì)知識(shí)完成了對(duì)商品資訊投送平臺(tái)系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)。最后,論文設(shè)計(jì)實(shí)驗(yàn)方案,驗(yàn)證商品資訊精準(zhǔn)投送模型的關(guān)鍵性能。
[Abstract]:With the rapid growth of information on e-commerce websites, it is difficult for consumers to obtain information of interest, and sellers are also faced with the difficulty of accurately sending commodity information to target customers. And these difficulties become the problem that restricts the sustainable development of e-commerce website. In order to solve the above problems, the industry put forward the personalized recommendation technology. At present, the recommendation object of individualized recommendation technology is basically the commodity itself, and little attention is paid to the text information of commodity information such as commodity discount, commodity advertisement, commodity promotion and so on. Releasing this information to consumers is an important way for offline stores to promote products, which has a strong impact on consumers' buying behavior. To solve this problem, this paper constructs a commodity information delivery platform to help sellers accurately deliver the information to the target customers. In order to meet the demand of commodity information delivery platform, this paper proposes a model of accurate delivery of commodity information based on latent semantic index. By analyzing the basic process of information delivery, it is obvious that the most critical problem affecting accurate matching is the recommendation strategy, and the most important part of the recommendation strategy is the information representation model. Compared with other information representation models, the latent semantic index model takes into account the semantic structure of the text, and effectively solves the problems of synonyms and polysemous words, and is superior to other models in the accuracy of information representation. The commodity information delivery model based on latent semantic index preprocesses a large number of training texts, and represents these text sets as word-text matrices, and reduces the word-text matrix from high-dimensional to low-dimensional by singular value decomposition. Form the latent semantic space. The generated latent semantic space preserves the latent semantic relationship of the original word-text matrix and removes the noise information brought about by the different ways of using the words so as to improve the accuracy of the delivery of commodity information. In addition, the consumer demand model reflects the interests of consumers and has a strong impact on the accuracy of the delivery of commodity information. Using ontology to improve the ability of consumer demand model to express consumer demand, so as to further improve the accuracy of commodity information delivery. In this paper, we creatively propose a model of accurate delivery of commodity information based on latent semantic index. In addition, this paper also applies the model to the construction of commodity information delivery platform system, and completes the design and implementation of commodity information delivery platform system based on object-oriented design knowledge. Finally, the paper designs the experimental scheme to verify the key performance of the model.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2011
【分類號(hào)】：TP311.13
本文編號(hào)：2257332
[Abstract]:With the rapid growth of information on e-commerce websites, it is difficult for consumers to obtain information of interest, and sellers are also faced with the difficulty of accurately sending commodity information to target customers. And these difficulties become the problem that restricts the sustainable development of e-commerce website. In order to solve the above problems, the industry put forward the personalized recommendation technology. At present, the recommendation object of individualized recommendation technology is basically the commodity itself, and little attention is paid to the text information of commodity information such as commodity discount, commodity advertisement, commodity promotion and so on. Releasing this information to consumers is an important way for offline stores to promote products, which has a strong impact on consumers' buying behavior. To solve this problem, this paper constructs a commodity information delivery platform to help sellers accurately deliver the information to the target customers. In order to meet the demand of commodity information delivery platform, this paper proposes a model of accurate delivery of commodity information based on latent semantic index. By analyzing the basic process of information delivery, it is obvious that the most critical problem affecting accurate matching is the recommendation strategy, and the most important part of the recommendation strategy is the information representation model. Compared with other information representation models, the latent semantic index model takes into account the semantic structure of the text, and effectively solves the problems of synonyms and polysemous words, and is superior to other models in the accuracy of information representation. The commodity information delivery model based on latent semantic index preprocesses a large number of training texts, and represents these text sets as word-text matrices, and reduces the word-text matrix from high-dimensional to low-dimensional by singular value decomposition. Form the latent semantic space. The generated latent semantic space preserves the latent semantic relationship of the original word-text matrix and removes the noise information brought about by the different ways of using the words so as to improve the accuracy of the delivery of commodity information. In addition, the consumer demand model reflects the interests of consumers and has a strong impact on the accuracy of the delivery of commodity information. Using ontology to improve the ability of consumer demand model to express consumer demand, so as to further improve the accuracy of commodity information delivery. In this paper, we creatively propose a model of accurate delivery of commodity information based on latent semantic index. In addition, this paper also applies the model to the construction of commodity information delivery platform system, and completes the design and implementation of commodity information delivery platform system based on object-oriented design knowledge. Finally, the paper designs the experimental scheme to verify the key performance of the model.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2011
【分類號(hào)】：TP311.13
【引證文獻(xiàn)】
相關(guān)碩士學(xué)位論文 前1條
1 嚴(yán)水發(fā);基于Agent的個(gè)性化服務(wù)平臺(tái)的應(yīng)用研究[D];中南大學(xué);2012年
，本文編號(hào)：2257332
本文鏈接：http://www.lk138.cn/wenyilunwen/guanggaoshejilunwen/2257332.html
最近更新
教材專著
