國立高雄大學圖資館 |

Language: English

Back

Information Recovery in Online Socia...

Lin, Yu-Cheng.

Information Recovery in Online Social Networks: Identification and Globalness Detection.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Information Recovery in Online Social Networks: Identification and Globalness Detection.
Author:	Lin, Yu-Cheng.
Published:	Ann Arbor : ProQuest Dissertations & Theses, 2018
Description:	97 p.
Notes:	Source: Dissertations Abstracts International, Volume: 80-09, Section: B.
Notes:	Publisher info.: Dissertation/Thesis.
Notes:	Advisor: Wu, Shyhtsun Felix.
Contained By:	Dissertations Abstracts International80-09B.
Subject:	Web Studies.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10980050
ISBN:	9780438930568

Information Recovery in Online Social Networks: Identification and Globalness Detection.
Lin, Yu-Cheng.

Information Recovery in Online Social Networks: Identification and Globalness Detection. - Ann Arbor : ProQuest Dissertations & Theses, 2018 - 97 p.

Source: Dissertations Abstracts International, Volume: 80-09, Section: B.

Thesis (Ph.D.)--University of California, Davis, 2018.

This item must not be sold to any third party vendors.

The prominence of online social network (OSN) within the context of the modern zeitgeist gives social scientists the unique opportunity to bolster their research with data on a larger scale than ever possible previously. Among the various data attributes on OSN, location attributes are of particular interest to social scientists since these digital footprints provide indications in many aspects. To fully utilize the Facebook public page data, there are mainly two challenges to solve. First, it's not feasible for social scientists to do large-scale comprehensive research if there are many missing data attributes – which happens quite often due to privacy, user's option, or simply lack of data. Second, the delicacy of the labeling is critical: mixing distinct items within the same class would result in diverging conclusions in some research; conversely, dividing items into incorrect classifications would result in poor results as well. This dissertation focuses on the Facebook public page graph, which represents the "Like" connections among public pages on Facebook. First, we devise an efficient algo- rithm to predict the country location of Facebook public pages with 90.25% accuracy. In addition, we compute various properties of the graph including page distribution by country and category, graph connectivity, and country international impact. We have these four findings: (1) the usage of the public page on Facebook is imbalanced. (2) nearly all countries are closely connected on Facebook. (3) the country's PageRank is highly correlated with the number of nations to which the country's pages connect. (4) by clustering the countries, we find that the Islamic countries are clustered into a separate group, which supports Huntington's argument that despite advances in technology, there is still a divide between cultures. Secondly, we develop a framework utilizing the connectivity of the page-like graph to predict the missing geo-location information based on Breadth-First Search (BFS). Our method achieves a satisfyingly high accuracy (89%) on identifying the state location attribute of unknown United States (US) pages. Our empirical results benefit the research of regional social analysis and target audience broadcasting. Finally, we propose an innovative methodology to detect those nodes with globalness property in OSN. While the notion of globalness is quite intuitive, the lack of an objective definition and of a method to measure it directly has greatly hindered scientific inspection. In this work, we operationalize globalness detection of the graph nodes in OSN, in terms of the unbiased relationship they have with the anchor nodes. In addition, we design an operational flow to detect these global nodes along with the local node prediction. We demonstrate our framework on Facebook public page graph. With high precision (89%) and recall (88%) rate of local pages, we have some interesting findings in the two experiments of global page detection. Our results unveil a set of important patterns and provide a better understanding of global pages in OSN.

ISBN: 9780438930568Subjects--Topical Terms:

660405
Web Studies.

Information Recovery in Online Social Networks: Identification and Globalness Detection.
LDR:04204nmm a2200325 4500 001 570691
005 20200514111939.5
008 200901s2018 ||||||||||||||||| ||eng d
020 $a 9780438930568
035 $a (MiAaPQ)AAI10980050
035 $a (MiAaPQ)ucdavis:18398
035 $a AAI10980050
040 $a MiAaPQ $c MiAaPQ
100 1 $a Lin, Yu-Cheng. $3 857327
245 1 0 $a Information Recovery in Online Social Networks: Identification and Globalness Detection.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2018
300 $a 97 p.
500 $a Source: Dissertations Abstracts International, Volume: 80-09, Section: B.
500 $a Publisher info.: Dissertation/Thesis.
500 $a Advisor: Wu, Shyhtsun Felix.
502 $a Thesis (Ph.D.)--University of California, Davis, 2018.
506 $a This item must not be sold to any third party vendors.
520 $a The prominence of online social network (OSN) within the context of the modern zeitgeist gives social scientists the unique opportunity to bolster their research with data on a larger scale than ever possible previously. Among the various data attributes on OSN, location attributes are of particular interest to social scientists since these digital footprints provide indications in many aspects. To fully utilize the Facebook public page data, there are mainly two challenges to solve. First, it's not feasible for social scientists to do large-scale comprehensive research if there are many missing data attributes – which happens quite often due to privacy, user's option, or simply lack of data. Second, the delicacy of the labeling is critical: mixing distinct items within the same class would result in diverging conclusions in some research; conversely, dividing items into incorrect classifications would result in poor results as well. This dissertation focuses on the Facebook public page graph, which represents the "Like" connections among public pages on Facebook. First, we devise an efficient algo- rithm to predict the country location of Facebook public pages with 90.25% accuracy. In addition, we compute various properties of the graph including page distribution by country and category, graph connectivity, and country international impact. We have these four findings: (1) the usage of the public page on Facebook is imbalanced. (2) nearly all countries are closely connected on Facebook. (3) the country's PageRank is highly correlated with the number of nations to which the country's pages connect. (4) by clustering the countries, we find that the Islamic countries are clustered into a separate group, which supports Huntington's argument that despite advances in technology, there is still a divide between cultures. Secondly, we develop a framework utilizing the connectivity of the page-like graph to predict the missing geo-location information based on Breadth-First Search (BFS). Our method achieves a satisfyingly high accuracy (89%) on identifying the state location attribute of unknown United States (US) pages. Our empirical results benefit the research of regional social analysis and target audience broadcasting. Finally, we propose an innovative methodology to detect those nodes with globalness property in OSN. While the notion of globalness is quite intuitive, the lack of an objective definition and of a method to measure it directly has greatly hindered scientific inspection. In this work, we operationalize globalness detection of the graph nodes in OSN, in terms of the unbiased relationship they have with the anchor nodes. In addition, we design an operational flow to detect these global nodes along with the local node prediction. We demonstrate our framework on Facebook public page graph. With high precision (89%) and recall (88%) rate of local pages, we have some interesting findings in the two experiments of global page detection. Our results unveil a set of important patterns and provide a better understanding of global pages in OSN.
590 $a School code: 0029.
650 4 $a Web Studies. $3 660405
650 4 $a Computer science. $3 199325
690 $a 0646
690 $a 0984
710 2 $a University of California, Davis. $b Computer Science. $3 730453
773 0 $t Dissertations Abstracts International $g 80-09B.
790 $a 0029
791 $a Ph.D.
792 $a 2018
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10980050