Academia.eduAcademia.edu

Figure 29 – uploaded by Esther Weltevrede

See full PDFdownloadDownload figure
Figure 33: Controversial issues in the ‘global warming’ Wikipedia article. Controversialness was calculated on the full edit history until April 16, 2014. The more controversial a wiki link, the redder it is. The images are converted to gray scale. The link to greenhouse gas is among the most controversial. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.

Figure 33 Controversial issues in the ‘global warming’ Wikipedia article. Controversialness was calculated on the full edit history until April 16, 2014. The more controversial a wiki link, the redder it is. The images are converted to gray scale. The link to greenhouse gas is among the most controversial. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.

Related Figures (41)

Figure 2: How Twitter tracked the News of the World scandal. Interactive visualization showing the amount of tweets per minute with the #notw hashtag. Source: Richards et al. 2011.
Figure 2: How Twitter tracked the News of the World scandal. Interactive visualization showing the amount of tweets per minute with the #notw hashtag. Source: Richards et al. 2011.
Figure 3: Google Scraper result displaying the number of web pages mentioning ACTA issues on the website access- vector.org. The issues ‘enforcement’ and ‘transparency’ are found the most on pages of accessvector.org. Screenshot taken from the Google Scraper tool on February 2, 2012. Source: Google Scraper 2007.
Figure 3: Google Scraper result displaying the number of web pages mentioning ACTA issues on the website access- vector.org. The issues ‘enforcement’ and ‘transparency’ are found the most on pages of accessvector.org. Screenshot taken from the Google Scraper tool on February 2, 2012. Source: Google Scraper 2007.
Querying historical controversies in dominant devices and platforms, the question we ask is what kind of history are we accessing on each device? More Information Nee Sal Sa Figure 4: Historical Controversies Now. Screenshot from interactive visualization showing results for the query [9/11] from Twitter, Facebook, Google News, Google Blog, YouTube, Flickr, Google Web, Google Scholar and Google Books, 18 August 2010. The y-axis shows from top to bottom: day, week, month, year, decade, century, undefined. The X-axis shows results for particular platforms. The markers are either circles for hot controversies, squares for cold controversies, and triangles for undefined types of controversies. The color of the marker is red if it is a present controversy, green if it is past controversy, and grey if the controversy is undefined. The figure shows that Twitter displays recent content about 9/11, while for example Google books displays older content. For the full interactive version see Dagdelen et al. 2010.
Querying historical controversies in dominant devices and platforms, the question we ask is what kind of history are we accessing on each device? More Information Nee Sal Sa Figure 4: Historical Controversies Now. Screenshot from interactive visualization showing results for the query [9/11] from Twitter, Facebook, Google News, Google Blog, YouTube, Flickr, Google Web, Google Scholar and Google Books, 18 August 2010. The y-axis shows from top to bottom: day, week, month, year, decade, century, undefined. The X-axis shows results for particular platforms. The markers are either circles for hot controversies, squares for cold controversies, and triangles for undefined types of controversies. The color of the marker is red if it is a present controversy, green if it is past controversy, and grey if the controversy is undefined. The figure shows that Twitter displays recent content about 9/11, while for example Google books displays older content. For the full interactive version see Dagdelen et al. 2010.
Figure 5: Co-words related to ‘austerity’ derived from the Twitter streaming API on January 1-31, 2012 concerning over 100.000 tweets. Visualization created by loading the co-word network from DMI-TCAT into Gephi and using a Force Atlas 2 layout algorithm, excluding top 1% most connected terms including austerity and RT, at least 100 co-occurrences are retained, nodes scaled by degree, 2012.
Figure 5: Co-words related to ‘austerity’ derived from the Twitter streaming API on January 1-31, 2012 concerning over 100.000 tweets. Visualization created by loading the co-word network from DMI-TCAT into Gephi and using a Force Atlas 2 layout algorithm, excluding top 1% most connected terms including austerity and RT, at least 100 co-occurrences are retained, nodes scaled by degree, 2012.
Figure 8: Barcode chart showing the pace of new content. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines within an interval, the more new results compared to the previ- ous five minute interval. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Figure 8: Barcode chart showing the pace of new content. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines within an interval, the more new results compared to the previ- ous five minute interval. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Figure 9: Pace Online for “Pakistan Floods”. Barcode chart showing the pace of content for the issue Pakistan Floods in Google Web, Google News, Google Blogs, Twitter, YouTube, Flickr, Wikipedia and Facebook in both fresh- ness and relevance mode, 18-19 August 2010. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines, the more new results compared to the previous five minute interval. Facebook’s pace is determined solely by freshness, while for example Google News’ pace is based both on freshness and relevance. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Figure 9: Pace Online for “Pakistan Floods”. Barcode chart showing the pace of content for the issue Pakistan Floods in Google Web, Google News, Google Blogs, Twitter, YouTube, Flickr, Wikipedia and Facebook in both fresh- ness and relevance mode, 18-19 August 2010. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines, the more new results compared to the previous five minute interval. Facebook’s pace is determined solely by freshness, while for example Google News’ pace is based both on freshness and relevance. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Figure 10: Barcode chart showing the pace of new content for the issue ‘Pakistan Floods’ on Twitter, 19 August 2010. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines, the more new results compared to the previous five minute interval. The orange square indicates idle time on Twitter. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Figure 10: Barcode chart showing the pace of new content for the issue ‘Pakistan Floods’ on Twitter, 19 August 2010. Each line represents a new result. Lines are distributed over the five minute intervals, the denser the lines, the more new results compared to the previous five minute interval. The orange square indicates idle time on Twitter. Visualization created by using custom code and Adobe Illustrator, 2010. Source: Borra et al. 2010.
Subsequently, the Internet Archive’s new Wayback Machine was queried for each blog’s URL and the result selected was dated closest to the middle of each investigated year. From the 2,507 URLs requested 946 blogs could be retrieved from the Internet Archive. This method yielded a collection of archived copies of historical Dutch blogs for each year with a timestamp near the middle of the year. Only blogs with a copy in the Internet Archive were retained for further analysis. Table 1 depicts the number of blogs per year serving as a starting point. Table 1: The number of blogs from the expert list that were available in the Internet Archive Wayback Machine, per year from 1999 until 2009. The URLs were retrieved with the Internet Archive Wayback Machine Network Per Year tool, 2011. Source: Weltevrede and Helmond 2012a.
Subsequently, the Internet Archive’s new Wayback Machine was queried for each blog’s URL and the result selected was dated closest to the middle of each investigated year. From the 2,507 URLs requested 946 blogs could be retrieved from the Internet Archive. This method yielded a collection of archived copies of historical Dutch blogs for each year with a timestamp near the middle of the year. Only blogs with a copy in the Internet Archive were retained for further analysis. Table 1 depicts the number of blogs per year serving as a starting point. Table 1: The number of blogs from the expert list that were available in the Internet Archive Wayback Machine, per year from 1999 until 2009. The URLs were retrieved with the Internet Archive Wayback Machine Network Per Year tool, 2011. Source: Weltevrede and Helmond 2012a.
Figure 11: Relative distribution of Top Level Domains (TLDs) in the Dutch blogs from 1999 until 2009. The Dutch blogs in the collection retrieved from the Wayback Machine favor the .nl domain over all other domains throughout the years. Moreover, a significant increase in the .nl domain becomes apparent, whereas the .com domain is stead- ily losing share over time. Visualization created in Google Spreadsheets, 2011. Source: Weltevrede and Helmond 2012a.
Figure 11: Relative distribution of Top Level Domains (TLDs) in the Dutch blogs from 1999 until 2009. The Dutch blogs in the collection retrieved from the Wayback Machine favor the .nl domain over all other domains throughout the years. Moreover, a significant increase in the .nl domain becomes apparent, whereas the .com domain is stead- ily losing share over time. Visualization created in Google Spreadsheets, 2011. Source: Weltevrede and Helmond 2012a.
Figure 12: Relative distribution of self—hosted blog software & blog platforms in Dutch blogs from 1999 until 2009. The graph shows the rise and popularity of Blogger’s platform, Blogspot, in the beginning of 2000 in the collection of blogs retrieved from the Wayback Machine. The decline of Blogspot coincides with the rise of the Web—Log.nl blogging platform, and other Dutch blog platforms such as BlogNL, Blogo, Blogse, Punt and Blogeiland. The figure clearly shows how from 2004-2005 onwards Dutch bloggers—except for a relatively small number of Blogspot and WordPress.com users—shift to Dutch platforms, which are orange color-coded. Only a few bloggers remain on legacy platforms such as Pitas, which no longer accept new members but are still functional for old members. Visualization created in Google Spreadsheets and Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 12: Relative distribution of self—hosted blog software & blog platforms in Dutch blogs from 1999 until 2009. The graph shows the rise and popularity of Blogger’s platform, Blogspot, in the beginning of 2000 in the collection of blogs retrieved from the Wayback Machine. The decline of Blogspot coincides with the rise of the Web—Log.nl blogging platform, and other Dutch blog platforms such as BlogNL, Blogo, Blogse, Punt and Blogeiland. The figure clearly shows how from 2004-2005 onwards Dutch bloggers—except for a relatively small number of Blogspot and WordPress.com users—shift to Dutch platforms, which are orange color-coded. Only a few bloggers remain on legacy platforms such as Pitas, which no longer accept new members but are still functional for old members. Visualization created in Google Spreadsheets and Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 13: The relative amount of Dutch blog platforms over time compared to other blog platforms from 1999 until 2009. In 2009 almost all bloggers on blog platforms make use of Dutch platforms in the collection of blogs retrieved from the Wayback Machine. Visualization created in Google Spreadsheets, 2011. Source: Weltevrede and Helmond 2012a.
Figure 13: The relative amount of Dutch blog platforms over time compared to other blog platforms from 1999 until 2009. In 2009 almost all bloggers on blog platforms make use of Dutch platforms in the collection of blogs retrieved from the Wayback Machine. Visualization created in Google Spreadsheets, 2011. Source: Weltevrede and Helmond 2012a.
Figure 16: The Dutch blogosphere in transition: the rise and evolution of the Dutch blogosphere 1999-2009. Map- ping the outlinks of the blogs retrieved from the Wayback Machine from 1999 until 2009 allowed to go back in time and study how and where the Dutch blogosphere originated. The network is made with the Internet Archive Wayback Machine Network per Year tool and networks per year are overlaid in Gephi. Thereafter a co-link analysis was applied and the graph was drawn using the Force Atlas 2 layout algorithm. The figure shows the rise, evolution and first signs of decline of the Dutch blogosphere. Grey depicts the hyperlink network of all years together and red the blogosphere of a particular year. The first Dutch bloggers starting mid 1999 were not interlinked into a ‘sphere’, so the beginning of a Dutch blogosphere can be traced back to 2000. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 16: The Dutch blogosphere in transition: the rise and evolution of the Dutch blogosphere 1999-2009. Map- ping the outlinks of the blogs retrieved from the Wayback Machine from 1999 until 2009 allowed to go back in time and study how and where the Dutch blogosphere originated. The network is made with the Internet Archive Wayback Machine Network per Year tool and networks per year are overlaid in Gephi. Thereafter a co-link analysis was applied and the graph was drawn using the Force Atlas 2 layout algorithm. The figure shows the rise, evolution and first signs of decline of the Dutch blogosphere. Grey depicts the hyperlink network of all years together and red the blogosphere of a particular year. The first Dutch bloggers starting mid 1999 were not interlinked into a ‘sphere’, so the beginning of a Dutch blogosphere can be traced back to 2000. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 17: The pre-blogosphere in 1999: early blogs link outward. The network is made with the Internet Archive Wayback Machine Network per Year tool and visualized with Gephi using the Force Atlas 2 layout algorithm; colors were produced with the modularity algorithm. Note that in this figure no co-link analysis was performed. Some of the known Dutch bloggers, as mentioned in Meeuwsen (2010), together with less well-known bloggers, are present but do not form a blogosphere yet. Most notably Alto169, ~wzweers and ~onnoz reach out to other Dutch blogs and may be seen as an effort to establish a community between blogs. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 17: The pre-blogosphere in 1999: early blogs link outward. The network is made with the Internet Archive Wayback Machine Network per Year tool and visualized with Gephi using the Force Atlas 2 layout algorithm; colors were produced with the modularity algorithm. Note that in this figure no co-link analysis was performed. Some of the known Dutch bloggers, as mentioned in Meeuwsen (2010), together with less well-known bloggers, are present but do not form a blogosphere yet. Most notably Alto169, ~wzweers and ~onnoz reach out to other Dutch blogs and may be seen as an effort to establish a community between blogs. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 18: Partial screenshot of the reconstructed Dutch blogosphere in 2000. All blogs retrieved from the Wayback Machine have been color-coded based on the type of host. Blue nodes are personal homepages, pink nodes are stu- dent pages, and yellow pages are early blog platforms. Bloggers on personal homepage providers (blue) and student pages (pink) dominate. For the full interactive version with the G-Atlas tool see Weltevrede and Helmond 2011.
Figure 18: Partial screenshot of the reconstructed Dutch blogosphere in 2000. All blogs retrieved from the Wayback Machine have been color-coded based on the type of host. Blue nodes are personal homepages, pink nodes are stu- dent pages, and yellow pages are early blog platforms. Bloggers on personal homepage providers (blue) and student pages (pink) dominate. For the full interactive version with the G-Atlas tool see Weltevrede and Helmond 2011.
Figure 19: Partial screenshot of the reconstructed Dutch blogosphere of 2005: a cluster of marketing blogs (in pink). Marketing blogs are marked in pink. The Dutch marketing cluster emerged in 2005 and was still a very dominant cluster in the Dutch blogosphere in 2011. Data retrieved from the Wayback Machine. Visualization cre- ated with G-Atlas and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2011.
Figure 19: Partial screenshot of the reconstructed Dutch blogosphere of 2005: a cluster of marketing blogs (in pink). Marketing blogs are marked in pink. The Dutch marketing cluster emerged in 2005 and was still a very dominant cluster in the Dutch blogosphere in 2011. Data retrieved from the Wayback Machine. Visualization cre- ated with G-Atlas and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2011.
Figure 20: Partial screenshot of the reconstructed Dutch blogosphere in 2004: all co-linked Bloggers also link to the Dutch web statistics service Nedstat Basic. Data retrieved from the Wayback Machine. Visualization created with G-Atlas and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2011.
Figure 20: Partial screenshot of the reconstructed Dutch blogosphere in 2004: all co-linked Bloggers also link to the Dutch web statistics service Nedstat Basic. Data retrieved from the Wayback Machine. Visualization created with G-Atlas and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2011.
Figure 21: The Dutch Blogosphere in 2009: a comparison of different actor definitions. Social media platform nodes are highlighted in magenta. Both graphs show the full hyperlink network of the 2009 Blogosphere (i.e. with- out co-link analysis). Top: nodes represent host names. Bottom: nodes represent platform specific actor definitions such as user profiles. Through a more fine-grained link analysis (bottom) it becomes possible to analyze the role of social media platforms within the Dutch blogosphere in more detail. Data Retrieved from the Wayback Machine. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 21: The Dutch Blogosphere in 2009: a comparison of different actor definitions. Social media platform nodes are highlighted in magenta. Both graphs show the full hyperlink network of the 2009 Blogosphere (i.e. with- out co-link analysis). Top: nodes represent host names. Bottom: nodes represent platform specific actor definitions such as user profiles. Through a more fine-grained link analysis (bottom) it becomes possible to analyze the role of social media platforms within the Dutch blogosphere in more detail. Data Retrieved from the Wayback Machine. Visualization created with Gephi and Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 22: Different types of links to social media in the 2009 Dutch blogosphere. The grey circles represent social media platforms in the 2009 blogosphere. The colored circles represent different types of platform links such as user pages, hashtag queries, and status updates. Each circle is scaled proportionally to the number of links received for a particular actor definition. Comparing the various social media platforms, the results suggest that some plat- forms can be defined as ‘media sharing’ platforms, such as YouTube and Flickr, which mainly consist of embedded content links in blogs. Data retrieved from the Wayback Machine. Visualization created in Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 22: Different types of links to social media in the 2009 Dutch blogosphere. The grey circles represent social media platforms in the 2009 blogosphere. The colored circles represent different types of platform links such as user pages, hashtag queries, and status updates. Each circle is scaled proportionally to the number of links received for a particular actor definition. Comparing the various social media platforms, the results suggest that some plat- forms can be defined as ‘media sharing’ platforms, such as YouTube and Flickr, which mainly consist of embedded content links in blogs. Data retrieved from the Wayback Machine. Visualization created in Adobe Illustrator, 2011. Source: Weltevrede and Helmond 2012a.
Figure 23: Climate Change Sceptics on the Web (Frederick Seitz). Tag cloud displaying the number of web pages mentioning skeptic “Frederick Seitz” in the top hundred unique hosts returned by Google for the query [“Climate Change”]. The order of Google results is retained in the visualization. It can be seen that Seitz is not well recognized by the ‘top of the web’. Visualization created in Adobe Illustrator, 30 July 2007. Source: Digital Methods Initiative 2009. One of the Digital Methods program earliest empirical critical works using the out- puts of Google search results studied the effects of PageRank by questioning to what extent the results returned provide mainstream or alternative voices in the top engine results (also see the work of Muddiman 2013; Eklof and Mager 2013). Research with PageRank is for instance suitable for ‘source distance’ research, a technique devel- oped during the first Digital Methods Summer School in 2007, which looks at the sources in which certain terms appear and how they rank for a certain query. The classic example project is ‘climate change skeptics’ which looks at the position of known climate change skeptics in the top results for the query [“climate change”] (see
Figure 23: Climate Change Sceptics on the Web (Frederick Seitz). Tag cloud displaying the number of web pages mentioning skeptic “Frederick Seitz” in the top hundred unique hosts returned by Google for the query [“Climate Change”]. The order of Google results is retained in the visualization. It can be seen that Seitz is not well recognized by the ‘top of the web’. Visualization created in Adobe Illustrator, 30 July 2007. Source: Digital Methods Initiative 2009. One of the Digital Methods program earliest empirical critical works using the out- puts of Google search results studied the effects of PageRank by questioning to what extent the results returned provide mainstream or alternative voices in the top engine results (also see the work of Muddiman 2013; Eklof and Mager 2013). Research with PageRank is for instance suitable for ‘source distance’ research, a technique devel- oped during the first Digital Methods Summer School in 2007, which looks at the sources in which certain terms appear and how they rank for a certain query. The classic example project is ‘climate change skeptics’ which looks at the position of known climate change skeptics in the top results for the query [“climate change”] (see
Figure 24: Issue Animals Hierarchy on The Web (Google). This figure shows how prominent certain animals are for a Google Search query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Google Search. It is found that on the web, for a text query, results are distributed across all the animals and thus do not particularly favor one specific issue animal. Visualization created in Adobe Illustrator, 15 July 2007. Source: Digital Methods Initiative 2007.
Figure 24: Issue Animals Hierarchy on The Web (Google). This figure shows how prominent certain animals are for a Google Search query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Google Search. It is found that on the web, for a text query, results are distributed across all the animals and thus do not particularly favor one specific issue animal. Visualization created in Adobe Illustrator, 15 July 2007. Source: Digital Methods Initiative 2007.
Figure 25: Issue Animals Hierarchy in the News (Google News). This figure shows how prominent certain animals are for a Google News query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Technorati. It is found that in the news, for a text query, the polar bear is the animal most associated with climate change, followed by the cow. Visualization created in Adobe Illustrator, 15 July 2007. Source: Digital Methods Initiative 2007.
Figure 25: Issue Animals Hierarchy in the News (Google News). This figure shows how prominent certain animals are for a Google News query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Technorati. It is found that in the news, for a text query, the polar bear is the animal most associated with climate change, followed by the cow. Visualization created in Adobe Illustrator, 15 July 2007. Source: Digital Methods Initiative 2007.
Figure 26: Issue Animals Hierarchy in the Blogosphere (Technorati). This figure shows how prominent certain ani- mals are for a Technorati query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Technorati. It is found that in the blogosphere, for a text query, the polar bear is the animal most associated with climate change, followed by the cow. Visualization created in Adobe Illustrator, 17 July 2007. Source: Digital Methods Initiative 2007.
Figure 26: Issue Animals Hierarchy in the Blogosphere (Technorati). This figure shows how prominent certain ani- mals are for a Technorati query for [“climate change”]. Animals are scaled by the number of results (in text and image) returned by Technorati. It is found that in the blogosphere, for a text query, the polar bear is the animal most associated with climate change, followed by the cow. Visualization created in Adobe Illustrator, 17 July 2007. Source: Digital Methods Initiative 2007.
Figure 27: Issue Animals Hierarchy on the Web (Google Images). This figure shows how prominent certain animals are for a Google Images query for [“climate change”]. Images are scaled by the number of images returned by Google images. It is found that, except for the polar bear, no animals are particularly ‘favored’. Visualization created in Adobe Illustrator, 17 July 2007. Source: Digital Methods Initiative 2007.
Figure 27: Issue Animals Hierarchy on the Web (Google Images). This figure shows how prominent certain animals are for a Google Images query for [“climate change”]. Images are scaled by the number of images returned by Google images. It is found that, except for the polar bear, no animals are particularly ‘favored’. Visualization created in Adobe Illustrator, 17 July 2007. Source: Digital Methods Initiative 2007.
Figure 28: A Website is gone: the apparent removal of 911truth.org from Google results for the query [9/11], Sep- tember-October 2007. The graph shows how 911truth.org used to be in the top 10 Google results for the query [9/11], but received a dramatic drop starting 16 October, 2007. Source: Govcom.org 2007.
Figure 28: A Website is gone: the apparent removal of 911truth.org from Google results for the query [9/11], Sep- tember-October 2007. The graph shows how 911truth.org used to be in the top 10 Google results for the query [9/11], but received a dramatic drop starting 16 October, 2007. Source: Govcom.org 2007.
Table 2: Example of signals that are used to identify and define the local intent of queries and the locale of websites. Signals derived from: Diligenti et al. 2012; Buron et al. 2010; Heymans et al. 2011.
Table 2: Example of signals that are used to identify and define the local intent of queries and the locale of websites. Signals derived from: Diligenti et al. 2012; Buron et al. 2010; Heymans et al. 2011.
Figure 30: Hierarchies of rights types per country. Top ten distinctive rights types for the query [“rights”] in the local languages of various local Google versions (for example [“oigused”] in Google.ee and [“direitos”] in Google. pt). Results are in the order that Google provided and translated to English. It was found that certain countries have shared concerns, such as human rights, whereas others have unique concerns, such as activist’s rights in Australia. Visualization created in Adobe Illustrator, 2009. Source: Bekema et al. 2010.
Figure 30: Hierarchies of rights types per country. Top ten distinctive rights types for the query [“rights”] in the local languages of various local Google versions (for example [“oigused”] in Google.ee and [“direitos”] in Google. pt). Results are in the order that Google provided and translated to English. It was found that certain countries have shared concerns, such as human rights, whereas others have unique concerns, such as activist’s rights in Australia. Visualization created in Adobe Illustrator, 2009. Source: Bekema et al. 2010.
Figure 31: Wikipedia Policies and Guidelines. Partial screenshot. Source: Wikipedia contributors 2012b.
Figure 31: Wikipedia Policies and Guidelines. Partial screenshot. Source: Wikipedia contributors 2012b.
Figure 32: The English Wikipedia article on ‘global warming’ on January 12, 2013 with all plain text blanked out to emphasize Wikipedia’s active content objects. Visualization created in Firefox, 2013. Source: Density Design et al. 2015.
Figure 32: The English Wikipedia article on ‘global warming’ on January 12, 2013 with all plain text blanked out to emphasize Wikipedia’s active content objects. Visualization created in Firefox, 2013. Source: Density Design et al. 2015.
Figure 34: Partial dashboard view of the controversial wiki links in the ‘global warming’ Wikipedia article on April 16, 2014. Each row represents a controversial wiki link. The rows are ordered by how controversial the wiki link is. The redder the square, the more controversial it is overall. Also shown are a timeline of edits and a controversy bar indicating at which time the element was most controversial. Additionally, the type of element and the number of users editing sentences containing that wiki link are shown. The wiki link ‘greenhouse gas’ is the most controver- sial. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 34: Partial dashboard view of the controversial wiki links in the ‘global warming’ Wikipedia article on April 16, 2014. Each row represents a controversial wiki link. The rows are ordered by how controversial the wiki link is. The redder the square, the more controversial it is overall. Also shown are a timeline of edits and a controversy bar indicating at which time the element was most controversial. Additionally, the type of element and the number of users editing sentences containing that wiki link are shown. The wiki link ‘greenhouse gas’ is the most controver- sial. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 36: Detail of the dashboard view of the ‘global warming’ Wikipedia article on April 16, 2014, including a partial edit history. The red color under the ‘edit’ section indicates a deletion and green an insertion of text. The first few rows of the edit table show how models of global warming are put into doubt both through edit activity and via the comments. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015. Figure 35: Detail of the dashboard view ‘intergovernmental panel on climate change’ wiki link in the ‘global warm- ing’ Wikipedia article on April 16, 2014. The wiki link was most edited within the article on global warming in 2005 en 2007 while it was involved in most edit activity in the middle of 2006. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 36: Detail of the dashboard view of the ‘global warming’ Wikipedia article on April 16, 2014, including a partial edit history. The red color under the ‘edit’ section indicates a deletion and green an insertion of text. The first few rows of the edit table show how models of global warming are put into doubt both through edit activity and via the comments. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015. Figure 35: Detail of the dashboard view ‘intergovernmental panel on climate change’ wiki link in the ‘global warm- ing’ Wikipedia article on April 16, 2014. The wiki link was most edited within the article on global warming in 2005 en 2007 while it was involved in most edit activity in the middle of 2006. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 39: Detail of the dashboard view of the ‘global warming’ Wikipedia article from February 1, 2006 to Febru- ary 1, 2008. The second wiki link was used and controversial in 2006, whereas the first wiki link was present in the article throughout most of 2007. The names of the wiki links are struck through as they do not appear in the article on February 1, 2008. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 39: Detail of the dashboard view of the ‘global warming’ Wikipedia article from February 1, 2006 to Febru- ary 1, 2008. The second wiki link was used and controversial in 2006, whereas the first wiki link was present in the article throughout most of 2007. The names of the wiki links are struck through as they do not appear in the article on February 1, 2008. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 42: Partial dashboard view of the controversial wiki links in the ‘global warming’ Wikipedia article between April 16, 2010 and April 14, 2014. In this period for example scientific opinion on climate change is more contro- versial than greenhouse gas. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
Figure 42: Partial dashboard view of the controversial wiki links in the ‘global warming’ Wikipedia article between April 16, 2010 and April 14, 2014. In this period for example scientific opinion on climate change is more contro- versial than greenhouse gas. Partial screenshot from the Contropedia Demo tool, 2015. Source: Density Design et al. 2015.
You are almost done. Please take a moment to fill out the following information*: Figure 43: Alexa toolbar installation and registration process, with a field for the user’s postal code. Partial screen- shot, 2011. Source: Alexa n.d. The Alexa Toolbar for Firefox - Demographic Information
You are almost done. Please take a moment to fill out the following information*: Figure 43: Alexa toolbar installation and registration process, with a field for the user’s postal code. Partial screen- shot, 2011. Source: Alexa n.d. The Alexa Toolbar for Firefox - Demographic Information
-org with 4 percent. The ccTLD .ir contained 3 percent of all censored hosts. I would like to add a fourth type, multiple aggregator site scraping or more concep- tually said, device cultures to the sampling techniques described above. In any case, Google Ad Planner, Alexa, Google Web Search, Likekhor (Google Reader) as well as the crowd-sourcing platforms (Donbaleh, Sabzlink and Balatarin) either through que- ry results or (dynamically generated) listings yield websites relevant for Iranians and Persian speakers. In the case at hand, with the exception of the searcher’s web (gained through .ir and generic TLD queries in Google’s region search), the percentages of .ir sites among the significant hosts outputted by the devices were relatively low (see Table 3). The crowd-sourced web references yielded the fewest .ir sites at just over 10 percent, whilst both Google Ad Planner’s web as well as Alexa’s web yielded most at about 25 percent. As noted earlier, the .ir sites in the overall collection of URLs were much less likely to be blocked than the .com sites. 80 percent of the websites tested and found blocked from inside Iran were .com, followed by .net with 6 percent and .org with 4 percent. The ccTLD .ir contained 3 percent of all censored hosts. Table 3: Percentage of .ir sites in top websites collected from Alexa, Google Ad Planner, Likekhor, Don- baleh/Sabzlink and Balatarin, July 2011. Source: Rogers et al. 2011.
-org with 4 percent. The ccTLD .ir contained 3 percent of all censored hosts. I would like to add a fourth type, multiple aggregator site scraping or more concep- tually said, device cultures to the sampling techniques described above. In any case, Google Ad Planner, Alexa, Google Web Search, Likekhor (Google Reader) as well as the crowd-sourcing platforms (Donbaleh, Sabzlink and Balatarin) either through que- ry results or (dynamically generated) listings yield websites relevant for Iranians and Persian speakers. In the case at hand, with the exception of the searcher’s web (gained through .ir and generic TLD queries in Google’s region search), the percentages of .ir sites among the significant hosts outputted by the devices were relatively low (see Table 3). The crowd-sourced web references yielded the fewest .ir sites at just over 10 percent, whilst both Google Ad Planner’s web as well as Alexa’s web yielded most at about 25 percent. As noted earlier, the .ir sites in the overall collection of URLs were much less likely to be blocked than the .com sites. 80 percent of the websites tested and found blocked from inside Iran were .com, followed by .net with 6 percent and .org with 4 percent. The ccTLD .ir contained 3 percent of all censored hosts. Table 3: Percentage of .ir sites in top websites collected from Alexa, Google Ad Planner, Likekhor, Don- baleh/Sabzlink and Balatarin, July 2011. Source: Rogers et al. 2011.
Table 4: Metrics commonly used in national web characterization studies, 2011. Source: Baeza-Yates, Castillo, and Efthimiadis 2007.
Table 4: Metrics commonly used in national web characterization studies, 2011. Source: Baeza-Yates, Castillo, and Efthimiadis 2007.
Languages on the Iranian web Figure 44: The distribution of languages on the Iranian webs. Likekhor, Alexa, Balatarin, Google Ad Planner, Google Web, Donbaleh and Sabzlink were queried to retrieve (ranked) lists of relevant URLs. As platforms such as Alexa only provide a ranked list of hosts, all URLs were chopped to their (sub-)domain part, in order to facilitate compari- son. The data were collected on July 7, 2011 and resulted in a list of URLs per web device. To automatically detect the different languages used in the Iranian webs, a custom tool was used that makes use of AlchemyAPI. The graph shows the percentage of URLs in a given language, per web device. The languages are color-coded. Likekhor’s web is mainly in Persian, while the Donbaleh and Sabzlink’s web has most diverse languages. Visualization created in Adobe Illustrator, 2011. Source: Rogers et al. 2011.
Languages on the Iranian web Figure 44: The distribution of languages on the Iranian webs. Likekhor, Alexa, Balatarin, Google Ad Planner, Google Web, Donbaleh and Sabzlink were queried to retrieve (ranked) lists of relevant URLs. As platforms such as Alexa only provide a ranked list of hosts, all URLs were chopped to their (sub-)domain part, in order to facilitate compari- son. The data were collected on July 7, 2011 and resulted in a list of URLs per web device. To automatically detect the different languages used in the Iranian webs, a custom tool was used that makes use of AlchemyAPI. The graph shows the percentage of URLs in a given language, per web device. The languages are color-coded. Likekhor’s web is mainly in Persian, while the Donbaleh and Sabzlink’s web has most diverse languages. Visualization created in Adobe Illustrator, 2011. Source: Rogers et al. 2011.
The health of the Iranian web Figure 45: The liveliness of the Iranian webs measured by HTTP response codes retrieved from the Netherlands. The graph shows color codes indicating the response codes and is subdivided in pie charts per web device. Data was collected in July and August 2011. The crowd-sourced web had most blocked pages. Visualization created in Adobe Illustrator, 2011. Source: Rogers et al. 2011.
The health of the Iranian web Figure 45: The liveliness of the Iranian webs measured by HTTP response codes retrieved from the Netherlands. The graph shows color codes indicating the response codes and is subdivided in pie charts per web device. Data was collected in July and August 2011. The crowd-sourced web had most blocked pages. Visualization created in Adobe Illustrator, 2011. Source: Rogers et al. 2011.
YouTube, Iran Traffic Divided by Worldwide Traffic and Normalized
YouTube, Iran Traffic Divided by Worldwide Traffic and Normalized
across a country, considering that both the implementation and user experience of censorship may vary by city, ISP, or even by computer (Wright, De Souza, and Brown 2011, 5). Taking this concern into account, we selected proxies from different cities and ISPs, and subsequently considered the response code returned by the majority. Table 5 details which proxies were used for this research. Table 5: Details of the proxies used to test for censorship in Iran. URLs were queried through the vari- ous proxies between August 22, 2011 and September 8, 2011. Proxies retrieved from the Censorship Explorer tool.
across a country, considering that both the implementation and user experience of censorship may vary by city, ISP, or even by computer (Wright, De Souza, and Brown 2011, 5). Taking this concern into account, we selected proxies from different cities and ISPs, and subsequently considered the response code returned by the majority. Table 5 details which proxies were used for this research. Table 5: Details of the proxies used to test for censorship in Iran. URLs were queried through the vari- ous proxies between August 22, 2011 and September 8, 2011. Proxies retrieved from the Censorship Explorer tool.
Related topics:
Engineering
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved