發(fā)布時間：2018-05-02 03:02

  本文選題：稀疏特征 + 密集特征��； 參考：《北方工業(yè)大學(xué)》2017年碩士論文

【摘要】：隨著計算機(jī)技術(shù)和通信技術(shù)的不斷進(jìn)步,移動設(shè)備多種多樣,功能更加完善,用戶的移動數(shù)據(jù)可以隨時隨地發(fā)送到服務(wù)器上,進(jìn)行有效保存。日積月累,移動數(shù)據(jù)無論是數(shù)量還是種類都變得相當(dāng)巨大。這對于研究人員挖掘移動數(shù)據(jù)中所蘊(yùn)藏的價值,提取移動用戶數(shù)據(jù)特征,向移動用戶提供個性化服務(wù),帶來了巨大的挑戰(zhàn)。本課題依托于網(wǎng)絡(luò)后臺服務(wù)器以及Android端滑屏App。用戶使用安裝了該滑屏App的智能設(shè)備時,移動數(shù)據(jù)便傳送并保存到后臺服務(wù)器上。本文通過分析移動用戶數(shù)據(jù),根據(jù)用戶的地理位置信息的稀疏密集程度,提取出了移動數(shù)據(jù)稀疏特征和密集特征,建立了移動數(shù)據(jù)稀疏特征模型和密集特征模型。在服務(wù)器端實現(xiàn)了移動用戶數(shù)據(jù)特征模型,并將個性化服務(wù)器信息推送至客戶端。在客戶端實現(xiàn)了移動用戶數(shù)據(jù)特征展示,以便用戶查看個人的數(shù)據(jù)特征。首先,結(jié)合百度地圖API,由地理位置信息轉(zhuǎn)換得到位置語義信息,并對位置語義信息進(jìn)行類別劃分。移動用戶剛進(jìn)入系統(tǒng)時,其移動用戶的地理位置數(shù)據(jù)是稀疏的。為了分析移動用戶的訪問興趣點(diǎn)的行為偏好,提取出移動數(shù)據(jù)稀疏特征—年齡特征、性別特征和類別相似特征。根據(jù)提取的移動數(shù)據(jù)稀疏特征,設(shè)計了移動數(shù)據(jù)稀疏特征模型。其次,隨著時間或地點(diǎn)的不斷變化,移動數(shù)據(jù)無論是數(shù)量還是種類變得越來越密集。根據(jù)當(dāng)前移動數(shù)據(jù)的特點(diǎn),提取出移動數(shù)據(jù)密集特征—地理位置特征、時間特征和類別特征。利用移動數(shù)據(jù)密集特征,設(shè)計了移動數(shù)據(jù)密集特征模型。最后,為了驗證不同模型的正確性,分析了依據(jù)不同模型的得到的推薦結(jié)果,并結(jié)合Android百度地圖API,開發(fā)了 Android端移動用戶個人數(shù)據(jù)特征展示模塊。此模塊包括:移動用戶軌跡追蹤、移動用戶興趣點(diǎn)展示、移動用戶活動區(qū)域分布。移動用戶個人軌跡追蹤包括移動用戶實時軌跡追蹤和歷史軌跡查看。移動用戶興趣點(diǎn)展示是根據(jù)設(shè)計的移動數(shù)據(jù)稀疏特征模型和密集特征模型得到的結(jié)果推薦到客戶端并展示。移動用戶個人活動區(qū)域分布功能可以選擇不同條件來查看自己的活動區(qū)域。
[Abstract]:With the development of computer technology and communication technology, mobile devices have a variety of functions, and users' mobile data can be sent to the server at any time and anywhere for effective storage. Over time, mobile data, both in terms of quantity and variety, has become quite large. This is a great challenge for researchers to mine the value of mobile data, extract the features of mobile user data, and provide personalized services to mobile users. This topic depends on the network background server and the Android terminal slide screen App. When a user uses a smart device that installs the slider App, the mobile data is transferred and saved to the background server. Based on the analysis of mobile user data, the sparse feature and dense feature of mobile data are extracted according to the sparse density of user location information, and the sparse feature model and dense feature model of mobile data are established. The mobile user data feature model is implemented on the server side, and the personalized server information is pushed to the client. The mobile user data feature display is implemented on the client side so that the user can view the personal data feature. Firstly, with the help of Baidu map API, the location semantic information is obtained by the transformation of geographical location information, and the location semantic information is classified into categories. When the mobile user enters the system, the geographic location data of the mobile user is sparse. In order to analyze the behavioral preference of mobile users' access points of interest, the sparse features of mobile data, such as age, gender and category similarity, were extracted. According to the extracted sparse features of mobile data, a sparse feature model of mobile data is designed. Secondly, as time or place changes, mobile data become more and more dense in terms of quantity and type. According to the characteristics of current mobile data, the features of mobile data density, such as geographical location feature, time feature and category feature, are extracted. A mobile data dense feature model is designed using mobile data dense features. Finally, in order to verify the correctness of different models, this paper analyzes the recommended results obtained from different models, and develops the Android mobile user personal data feature display module combined with Android Baidu map API. This module includes: mobile user trajectory tracking, mobile user interest point display, mobile user activity area distribution. Personal trajectory tracking of mobile users includes real-time tracking of mobile users and viewing of historical tracks. Mobile user points of interest display is based on the design of mobile data sparse feature model and dense feature model results are recommended to the client and display. Mobile user's individual activity area distribution function can select different conditions to view their own active area.
【學(xué)位授予單位】：北方工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：P289;TP391.3

【參考文獻(xiàn)】

相關(guān)期刊論文 前8條

1 劉文兵;胡婷婷;江帆;王俊;;基于Web Services的異構(gòu)數(shù)字地圖資源檢索算法[J];地理空間信息;2016年12期

2 殷方勇;王紅;王吉華;;一種緩解互惠推薦系統(tǒng)中數(shù)據(jù)稀疏性的算法[J];濟(jì)南大學(xué)學(xué)報(自然科學(xué)版);2017年01期

3 毛嘉莉;金澈清;章志剛;周傲英;;軌跡大數(shù)據(jù)異常檢測:研究進(jìn)展及系統(tǒng)框架[J];軟件學(xué)報;2017年01期

4 范玖國;王全全;;三維數(shù)字地形圖符號開發(fā)與研究[J];礦山測量;2016年05期

5 李建軍;;基于GIS的數(shù)字地圖的應(yīng)用研究[J];科技風(fēng);2016年15期

6 朱保華;;基于用戶特征和用戶興趣的推薦計算[J];軟件導(dǎo)刊;2014年11期

7 陳佳;胡波;左小清;樂陽;;利用手機(jī)定位數(shù)據(jù)的用戶特征挖掘[J];武漢大學(xué)學(xué)報(信息科學(xué)版);2014年06期

8 袁柳;張龍波;;個性化搜索中的用戶特征模型研究[J];計算機(jī)工程與應(yīng)用;2011年15期

相關(guān)碩士學(xué)位論文 前2條

1 郭銳;農(nóng)村移動用戶特征及移動運(yùn)營商市場策略研究[D];北京郵電大學(xué);2008年

2 何X;基于聚類的用戶特征分析[D];揚(yáng)州大學(xué);2008年

，

本文編號：1832104