東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Fake News Prediction on Facebook: De...

Meyn, Lennie Frederik.

FindBook

Google Book

Amazon

博客來

Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool./
作者:	Meyn, Lennie Frederik.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2018,
面頁冊數:	80 p.
附註:	Source: Masters Abstracts International, Volume: 80-08.
Contained By:	Masters Abstracts International80-08.
標題:	Statistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10982246
ISBN:	9780438863958

Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool.
Meyn, Lennie Frederik.

Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool. - Ann Arbor : ProQuest Dissertations & Theses, 2018 - 80 p.

Source: Masters Abstracts International, Volume: 80-08.

Thesis (M.S.)--University of Nebraska at Omaha, 2018.

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

Social Media Services like Facebook or Twitter are present in the daily routine of millions of active users around the earth. The user behavior on these networks has changed from mostly interacting with friends towards active dissemination of information when News channels and politicians became active. The Social Media Networks have become a place for political debates and rapid news exchange in the last few years. By now, two thirds of Americans get their news from Social Media sites, which makes them an important target to spread false information broadly. With today's technology it is easy to create and distribute digital Fake News, which seem to be trustworthy at first sight. These News are mostly generated to manipulate the users' (political) perception. This approach was most recently seen on a grand scale during the 2016 U.S. election. This thesis explores the disparity in Fake News prediction between different Social Media platforms and tests prediction models to find the best working prediction model for Facebook. For that purpose, existing prediction approaches from other Social Media platforms will be transferred, altered, combined and extended. In addition, a clustering approach (semisupervised) based on the Latent-Dirichlet-Allocation model will be tested. A successful semisupervised prediction model could erase the need for huge pre-labeled datasets, which require excessive time and energy to generate. A total of eight different prediction models have been implemented as part of the thesis: three models from Twitter and Sina Weibo, two Ensemble Method models, one self-adjusted SVM-model, an existing Facebook model and the novel semisupervised LDA-clustering approach. The three models from Twitter and Sina Weibo were successfully transferred, yet using ensemble method models did not increase the performance of the existing models. However, quantitative features such as comments, shares, and reactions have a high impact of the prediction outcome. In this situation, the LDA-approach did perform significantly worse than its supervised counterparts.

ISBN: 9780438863958Subjects--Topical Terms:

517247
Statistics.

Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool.
LDR:03181nmm a2200337 4500 001 2209647
005 20191104073757.5
008 201008s2018 ||||||||||||||||| ||eng d
020 $a 9780438863958
035 $a (MiAaPQ)AAI10982246
035 $a (MiAaPQ)unomaha:11250
035 $a AAI10982246
040 $a MiAaPQ $c MiAaPQ
100 1 $a Meyn, Lennie Frederik. $3 3436744
245 1 0 $a Fake News Prediction on Facebook: Design and Implementation of a Fake News Prediction Tool.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2018
300 $a 80 p.
500 $a Source: Masters Abstracts International, Volume: 80-08.
500 $a Publisher info.: Dissertation/Thesis.
500 $a Advisor: Haas, Christian.
502 $a Thesis (M.S.)--University of Nebraska at Omaha, 2018.
506 $a This item must not be sold to any third party vendors.
520 $a Social Media Services like Facebook or Twitter are present in the daily routine of millions of active users around the earth. The user behavior on these networks has changed from mostly interacting with friends towards active dissemination of information when News channels and politicians became active. The Social Media Networks have become a place for political debates and rapid news exchange in the last few years. By now, two thirds of Americans get their news from Social Media sites, which makes them an important target to spread false information broadly. With today's technology it is easy to create and distribute digital Fake News, which seem to be trustworthy at first sight. These News are mostly generated to manipulate the users' (political) perception. This approach was most recently seen on a grand scale during the 2016 U.S. election. This thesis explores the disparity in Fake News prediction between different Social Media platforms and tests prediction models to find the best working prediction model for Facebook. For that purpose, existing prediction approaches from other Social Media platforms will be transferred, altered, combined and extended. In addition, a clustering approach (semisupervised) based on the Latent-Dirichlet-Allocation model will be tested. A successful semisupervised prediction model could erase the need for huge pre-labeled datasets, which require excessive time and energy to generate. A total of eight different prediction models have been implemented as part of the thesis: three models from Twitter and Sina Weibo, two Ensemble Method models, one self-adjusted SVM-model, an existing Facebook model and the novel semisupervised LDA-clustering approach. The three models from Twitter and Sina Weibo were successfully transferred, yet using ensemble method models did not increase the performance of the existing models. However, quantitative features such as comments, shares, and reactions have a high impact of the prediction outcome. In this situation, the LDA-approach did perform significantly worse than its supervised counterparts.
590 $a School code: 1060.
650 4 $a Statistics. $3 517247
650 4 $a Web Studies. $3 1026830
650 4 $a Information science. $3 554358
690 $a 0463
690 $a 0646
690 $a 0723
710 2 $a University of Nebraska at Omaha. $b Information Technology. $3 1043963
773 0 $t Masters Abstracts International $g 80-08.
790 $a 1060
791 $a M.S.
792 $a 2018
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10982246