【摘要】：隨著近十幾年互聯(lián)網(wǎng)的高速發(fā)展,社交網(wǎng)絡(luò)已經(jīng)變成人與人之間溝通交流的橋梁,其對(duì)人們的信息獲取、思考和生活產(chǎn)生了不可估量的影響。隨著移動(dòng)終端設(shè)備的不斷升級(jí)以及移動(dòng)平臺(tái)操作系統(tǒng)的逐步完善,基于LBS的移動(dòng)社交網(wǎng)絡(luò)由于其操作方便、信息快捷等優(yōu)勢(shì),在互聯(lián)網(wǎng)大家族中扮演著越來(lái)越重要的角色。而在最近幾年中,移動(dòng)社交網(wǎng)絡(luò)逐漸朝著小眾化、創(chuàng)新化發(fā)展,給用戶帶來(lái)定制化的用戶體驗(yàn)。在如此背景下,為了滿足某個(gè)區(qū)域內(nèi)人們之間的信息交流、溝通互動(dòng),本文提出了一種基于區(qū)域的社交模式:區(qū)域內(nèi)用戶之間多對(duì)多實(shí)時(shí)交流。一個(gè)用戶所發(fā)布的信息可以被區(qū)域內(nèi)其他所有用戶獲取并評(píng)論,實(shí)現(xiàn)一種相對(duì)自由的區(qū)域信息分享平臺(tái)。但這種模式下,用戶接受的信息體量會(huì)十分龐大,在如此多的內(nèi)容中,用戶會(huì)被大量無(wú)感的信息所包圍,所以本文提出了針對(duì)這種社交模式的信息分級(jí)系統(tǒng)。該信息分級(jí)系統(tǒng)核心就是信息推薦,也可以叫做信息推薦系統(tǒng),其根據(jù)用戶偏好來(lái)推薦該用戶感興趣的信息。本文首先介紹LBS、推薦系統(tǒng)、LDA等相關(guān)知識(shí)。然后對(duì)系統(tǒng)從功能、非功能、其他三個(gè)方面做了需求分析,其次從功能模塊、系統(tǒng)交互、數(shù)據(jù)庫(kù)表等方面對(duì)系統(tǒng)做了總體設(shè)計(jì)。隨后針對(duì)系統(tǒng)最重要的信息推薦模塊展開(kāi)討論:對(duì)區(qū)域信息推薦模塊,在提出的模塊設(shè)計(jì)基礎(chǔ)上,通過(guò)中文文本預(yù)處理、LDA主題提取、文檔排序等步驟給出了詳細(xì)實(shí)現(xiàn)。對(duì)個(gè)性化信息推薦模塊,本文研究了基于內(nèi)容的推薦和基于用戶的協(xié)同過(guò)濾兩種算法,基于內(nèi)容的推薦充分利用區(qū)域信息推薦中得到的數(shù)據(jù),計(jì)算信息內(nèi)容的相似度,而基于用戶的協(xié)同過(guò)濾則通過(guò)用戶對(duì)信息的評(píng)分來(lái)計(jì)算用戶之間的相似度。兩種算法在計(jì)算相似度時(shí)都在原有基礎(chǔ)上考慮了時(shí)間上下文,得到一種新的相似度計(jì)算方法。為了測(cè)試該相似度計(jì)算方法,本文在CCF競(jìng)賽數(shù)據(jù)集上設(shè)計(jì)實(shí)驗(yàn),通過(guò)準(zhǔn)確率和召回率的比較,最終驗(yàn)證了新算法的有效性。最后本文將所研究的信息分級(jí)推薦系統(tǒng)應(yīng)用于實(shí)踐,實(shí)現(xiàn)了一個(gè)簡(jiǎn)單的應(yīng)用,給出了關(guān)鍵性代碼以及主要的實(shí)現(xiàn)界面,并測(cè)試了分級(jí)推薦相關(guān)功能。
[Abstract]:With the rapid development of Internet in recent years, social network has become a bridge of communication between people, which has an incalculable impact on people's information acquisition, thinking and life. With the continuous upgrading of mobile terminal devices and the gradual improvement of mobile platform operating system, mobile social network based on LBS is playing an increasingly important role in the Internet family because of its advantages of convenient operation and fast information. In recent years, mobile social networks are gradually moving towards a niche and innovative development, bringing customized user experience to users. In this context, in order to meet the information exchange and communication interaction among people in a certain region, this paper proposes a region-based social model: many-to-many real-time communication among users in the region. The information published by a user can be accessed and commented by all other users in the region, thus realizing a relatively free regional information sharing platform. However, in this mode, the amount of information accepted by users will be very large. In such a large amount of content, users will be surrounded by a large number of senseless information, so this paper proposes an information grading system for this social model. The core of the information classification system is information recommendation, which can also be called information recommendation system, which recommends the information that the user is interested in according to the user's preference. This paper first introduces the LBS, recommendation system and other related knowledge. Then the system from the functional, non-functional, the other three aspects of the requirements analysis, followed by the functional module, system interaction, database table and other aspects of the overall design of the system. Then the most important information recommendation module of the system is discussed. On the basis of the design of the proposed module, a detailed implementation is given through the Chinese text preprocessing, LDA topic extraction, document sorting and so on. For the personalized information recommendation module, this paper studies two algorithms: content-based recommendation and user-based collaborative filtering. The content-based recommendation makes full use of the data obtained from the regional information recommendation, and calculates the similarity of the information content. The user-based collaborative filtering computes the similarity between users by scoring the information. The two algorithms take the time context into account when calculating the similarity, and obtain a new similarity calculation method. In order to test the similarity calculation method, this paper designs experiments on the CCF competition data set, and finally verifies the effectiveness of the new algorithm by comparing the accuracy and recall. Finally, this paper applies the information hierarchical recommendation system to practice, realizes a simple application, gives the key code and the main implementation interface, and tests the related functions of hierarchical recommendation.
【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】：2017
【分類(lèi)號(hào)】：TP391.3;TP393.09

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 劉成;;LBS定位技術(shù)研究與發(fā)展現(xiàn)狀[J];導(dǎo)航定位學(xué)報(bào);2013年01期

2 張啟宇;朱玲;張雅萍;;中文分詞算法研究綜述[J];情報(bào)探索;2008年11期

3 黃昌寧;趙海;;中文分詞十年回顧[J];中文信息學(xué)報(bào);2007年03期

4 鄧愛(ài)林,朱揚(yáng)勇,施伯樂(lè);基于項(xiàng)目評(píng)分預(yù)測(cè)的協(xié)同過(guò)濾推薦算法[J];軟件學(xué)報(bào);2003年09期

5 趙亮,胡乃靜,張守志;個(gè)性化推薦算法設(shè)計(jì)[J];計(jì)算機(jī)研究與發(fā)展;2002年08期

，

本文編號(hào)：2246964