用戶移動端與社交端行為建模與模式分析
發(fā)布時間：2018-10-31 11:37
【摘要】：隨著互聯(lián)網(wǎng)技術(shù)的迅猛發(fā)展,各種服務(wù)商在互聯(lián)網(wǎng)中開啟的流量入口也越來越多,形式各不相同,硬件上,用戶可以通過PC、智能平板、手機(jī)等使用相關(guān)服務(wù),在軟件上也包含了如社交網(wǎng)絡(luò)、電子商務(wù)、游戲等各類應(yīng)用渠道。也就是說,在現(xiàn)今的互聯(lián)網(wǎng)大環(huán)境下,有很多種渠道可以留下用戶的行為軌跡,這些行為軌跡能從某種程度上能夠折射出用戶的個性化特征,通過挖掘這些個性化特征可以幫助我們了解用戶日常的行為習(xí)慣,并對用戶進(jìn)行較為準(zhǔn)確的服務(wù)推薦。用戶在線使用服務(wù)的行為一般在兩種場景下發(fā)生,一種是圍繞用戶個人產(chǎn)生的行為,只是滿足自己的需求,并不與他人發(fā)生直接關(guān)系,比如在線音樂,手機(jī)閱讀等,另一種是多個用戶互相協(xié)作以滿足某種需求,比如社交網(wǎng)絡(luò)。那么我們針對這兩種場景展開對用戶在線行為的研究,即用戶在移動端的行為與用戶在社交端的行為,通過對行為的建模與行為模式的分析來研究用戶使用服務(wù)中的行為習(xí)慣。對于用戶在移動端的行為研究,我們挑選安卓用戶作為我們的研究群體,并進(jìn)行了如下工作:通過手機(jī)程序采集用戶在移動端發(fā)生的行為以上下文信息,并對數(shù)據(jù)進(jìn)行預(yù)處理;提出四種基于上下文的用戶的行為模式;設(shè)計挖掘基于上下文的行為模式算法并進(jìn)行算法的評價;提出若干移動端用戶行為預(yù)測策略;通過可視化系統(tǒng)來展示挖掘結(jié)果。對于用戶在社交端的行為研究,因為每種社交平臺給用戶提供的行為類型都不同,所以我們挑選行為類型較多,用戶活躍度較高的這種比較代表性的社交協(xié)作平臺github來進(jìn)行研究,并進(jìn)行了如下工作:通過信息抓取等技術(shù)采集用戶在github的行為并對數(shù)據(jù)進(jìn)行預(yù)處理;對用戶在線行為特征進(jìn)行分析;提出兩種行為模式:面向具體開發(fā)者的行為模式(PP)與面向抽象開發(fā)者的行為模式(RP);設(shè)計社交端用戶行為模式挖掘算法并實證分析;提出若干社交端用戶行為預(yù)測策略;通過可視化系統(tǒng)來展示挖掘結(jié)果。研究結(jié)果發(fā)現(xiàn),用戶在線行為是遵循一定的行為模式,用戶在移動端使用以個人為中心的服務(wù)時,與上下文結(jié)合可以更好地解釋用戶的行為模式,而用戶在社交端與其他用戶協(xié)作時,用戶間的模式之間差異與共性并存。
[Abstract]:With the rapid development of Internet technology, various service providers open more and more traffic ports in the Internet in different forms. In hardware, users can use related services through PC, smart tablets, mobile phones, etc. Software also includes various application channels such as social networks, e-commerce, games and so on. That is to say, in today's Internet environment, there are many channels that can leave the user's behavior track, which can to some extent reflect the personalized characteristics of the user. Mining these personalized features can help us to understand the user's daily behavior habits and make a more accurate service recommendation to the user. The behavior of users using services online generally takes place in two scenarios. One is the behavior around the user, which only meets his own needs and does not have a direct relationship with others, such as online music, cell phone reading, and so on. The other is the collaboration of multiple users to meet certain needs, such as social networks. Then we study the online behavior of users in these two scenarios, that is, the behavior of users on the mobile side and the behavior of users on the social side, and the behavior habits of users in the use of services are studied through the modeling of behavior and the analysis of behavior patterns. For the research of users' behavior on mobile side, we select Android users as our research group, and do the following work: collect the behavior of users on mobile side with context information through mobile phone program, and preprocess the data; Four context-based user behavior patterns are proposed; context-based behavior pattern algorithms are designed and evaluated; several mobile user behavior prediction strategies are proposed; and mining results are displayed through a visual system. In terms of the behavior of users on the social side, because each social platform provides users with different types of behavior, we choose more types of behavior. Github, a representative social cooperation platform with high user activity, is studied. The following works are done: collecting the behavior of users in github and preprocessing the data through information capture and other technologies; This paper analyzes the characteristics of users' online behavior, proposes two kinds of behavior patterns: (PP) for specific developers and (RP); for abstract developers, designs and empirically analyzes the algorithm of social user behavior pattern mining. Several social user behavior prediction strategies are proposed, and the mining results are displayed through a visualization system. The results show that the online behavior of the user follows a certain behavior pattern. When the user uses a personal-centric service on the mobile side, it can better explain the behavior pattern of the user by combining with the context. When users cooperate with other users on the social side, the differences and commonalities exist among users.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：TP311.13
，
本文編號：2302039
[Abstract]:With the rapid development of Internet technology, various service providers open more and more traffic ports in the Internet in different forms. In hardware, users can use related services through PC, smart tablets, mobile phones, etc. Software also includes various application channels such as social networks, e-commerce, games and so on. That is to say, in today's Internet environment, there are many channels that can leave the user's behavior track, which can to some extent reflect the personalized characteristics of the user. Mining these personalized features can help us to understand the user's daily behavior habits and make a more accurate service recommendation to the user. The behavior of users using services online generally takes place in two scenarios. One is the behavior around the user, which only meets his own needs and does not have a direct relationship with others, such as online music, cell phone reading, and so on. The other is the collaboration of multiple users to meet certain needs, such as social networks. Then we study the online behavior of users in these two scenarios, that is, the behavior of users on the mobile side and the behavior of users on the social side, and the behavior habits of users in the use of services are studied through the modeling of behavior and the analysis of behavior patterns. For the research of users' behavior on mobile side, we select Android users as our research group, and do the following work: collect the behavior of users on mobile side with context information through mobile phone program, and preprocess the data; Four context-based user behavior patterns are proposed; context-based behavior pattern algorithms are designed and evaluated; several mobile user behavior prediction strategies are proposed; and mining results are displayed through a visual system. In terms of the behavior of users on the social side, because each social platform provides users with different types of behavior, we choose more types of behavior. Github, a representative social cooperation platform with high user activity, is studied. The following works are done: collecting the behavior of users in github and preprocessing the data through information capture and other technologies; This paper analyzes the characteristics of users' online behavior, proposes two kinds of behavior patterns: (PP) for specific developers and (RP); for abstract developers, designs and empirically analyzes the algorithm of social user behavior pattern mining. Several social user behavior prediction strategies are proposed, and the mining results are displayed through a visualization system. The results show that the online behavior of the user follows a certain behavior pattern. When the user uses a personal-centric service on the mobile side, it can better explain the behavior pattern of the user by combining with the context. When users cooperate with other users on the social side, the differences and commonalities exist among users.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：TP311.13
，
本文編號：2302039
本文鏈接：http://sikaile.net/jingjilunwen/dianzishangwulunwen/2302039.html
最近更新
教材專著
