I found the piece “Sentiment Analysis and Subjectivity” by Bing Lui interesting as it really expanded on idea of computational text analysis as a means to analyze data. When we first began to look into text mining in class, I was a little confused as to how computers were able to take something like a news article or opinion piece, something so subjective and in the “grey area”, and turn it into something so black and white without going through the human thought-process of synthesizing the information. Lui’s writing helped me to understand this as he talked about how exactly computers synthesize the information for “opinion mining”.

Opinion mining is done by looking at each individual sentence and wording within that sentence with certain subjectivity classifications that help us deem what the overall consensus of the written piece entails. After learning how the actual process works, I find it curious to compare “Sentiment Analysis and Subjectivity” with what a peer of mine (BM) said in their post on “Text Mining/Language Standardization”, that a “computer will never understand the emotional values and ever-changing expressions of human beings”. From what I read today I think that computers are getting incredibly close to understanding the emotional values of human beings due to the work that has been done with opinion mining and the call for a better way to synthesize the public’s opinion and perception of things and products on a larger scale. Although there is a long way still to go, I think that this significant headway will be the basis for future breakthroughs.

Before reading Feminist Data Visualization by Catherine D’Ignazio and Lauren F. Klein I was curious how feminism, the advocacy of women’s rights on the basis of the equality of the sexes, could be intertwined with data visualization. I saw feminism as something that ebbed and flowed and was maybe a little subjective at times. In contrast to my notions of feminism, I saw data visualization as something purely objective. I thought a mix of the two would be confusing and honestly a stretch. After reading Feminist Data Visualization, I now understand I am very wrong. Feminist thought braided into data visualization lends to a productivity in the advances of female rights at a level that is both modern and practical. What I found the most interesting was the authors focus on epistemology – who is included in dominant ways of producing and communicating knowledge and whose perspectives are marginalized. From this focus on epistemology, feminist data visualization holds six principles constant when discussing data synthesis and visualization: Rethink Binaries, Embrace Pluralism, Examine Power and Aspire, to Empowerment, Consider Context, Legitimize Embodiment and Affect, Make Labor Visible. I think these principles are a great way to evaluate data as a means to better both humans and society.  As MLC said , “critical thinking about all of these categories will allow the audience and the author to remove some of the societal inequalities that all STEM fields currently have.”At first I saw these principals as a means to skew the objective information available, but after my growing understanding that data is constantly skewed and can really never be truly objective, I believe that having concrete principals in place for when someone is working with data leads to a more positively skewed outcome.

The article “Using Metadata to Find Paul Revere by Kieran Healy was very interesting because it allowed the reader to get an understanding of how we could relate the mediums of social networking to going through historical data. Paralleling people on social media today to historical figure Paul Revere drives home the point that information stripping technologies of today can be used to find important people of the past due to repetition. The relationships between people, the interconnections between groups and the overarching figures allow historians to learn who the key players of history are just like today. Also, through chronicling and annotating every social connection between people and groups, the author suggests that you could one day be able to learn more about their social and personal life. My peer, NL, relates this article to the six degrees of Francis Bacon “because they both showed that important people can be found using unbiased network analysis”. The one thing I found a little troubling to understand was how social media analysis would work in the future. Looking at the information we have of the past, it was only the most famous who were written about. Only the most important were a part of organizations, social groups and clubs. Now, whether it be the most important or the least important person, there is so much information on every human being. How will someone know that Barack Obama is more important than say, Snooki 100 years from now?

In the article “Lynching, Visualization and Visibility” by Lincoln Mullen, Mullen highlights how data can be perceived differently without the context alongside it. In the piece, Mullen looks at historical data of lynchings, a controversial and saddening part of American history. The writer also highlights the work of another journalist, Mathews, who looked at the same given data alongside a strong religious understanding of the act. As a result of this, Mathews drew a very different conclusion on the historical data than Mullen did. Mathews visualized the data without looking only at the empirical data, but rather, the history and religious significance behind the act to find out why it was caused. This writer looked at the trends of data, and took from it not their significance, but what it said about the religious understanding of lynching. I found this wildly confusing. Matthews argued that with a lack of lynching data, it meant, essentially, that more lynchings were happening during that time period due to the understanding that lynching was “a ritual that made power visible, yet its power depended in part on its lack of visibility in the official records”. So basically, at times were there was less information available about lynchings, more lynchings were taking place. To echo what RF said in their blog post, “the context in which we read and visualize data can sometimes be just as powerful as the insights themselves.” Reading this article was incredibly eye opening as it essentially rebuffs any information taken down about lynching acts ever and that same information has given us a warped understanding of history as a result.

This study carried out by Vejune Zemaityte, Deb Verhoeven, and Bronwyn Coate discusses the general implications of the 10 percent in the Australian Movie Industry. Most movie producers use the rule to estimate the income of an American film broadcasted in Australia, where they assume that it will bring in 10% less revenue than it did in America. Zemaityte and her colleagues carried out this study to show that this rule was quite inaccurate, to essentially invalidate the use of the 10% rule for this purpose.

In the article “Lynching, Visualization, and Visibility” Mullen explores how lynching as a religious history aspect is different in many way from lynching in a data collection aspect. They first look at the studies of lynchings with no stories behind them. They make charts of when and where the lynchings occur but they put no story behind them. I think this is problematic because they are looking at a piece of history that is very controversial and by just looking at the raw data I think a lot of the story aspect is lost. The line in the article that reads “lynching was a ritual that made power visible, yet its power depended in part on its lack of visibility in the official records.” As the lynchings were not recorded as legitimate data, making data visualizations of them now is somewhat inaccurate because we are unsure of what information is really accounted for. They then go on to adress how now data visualizations could also be skewed because they do not keep track of things such as when police use brute force. I think we should be learning from the past and taking into account now how people are being treated, we should be learning from our mistakes. We should be addressing racial conflicts and they showed how lynching could have done that and how now we should be addressing that with police brutality and understanding when it really goes in not just sometimes.

As RF said in their blog “the context in which we read and visualize data can sometimes be just as powerful as the insights themselves.” I think this statement is really true, but I think the Lynching article addresses a contradiction to this because it shows how sometimes their might not be enough data collected in order to create a good data visualization.  

It is baffling to me that this data is available even though lynchings were not reported by state or federal government. Even though there was a wide knowledge of lynchings, it was still illegal by law, and thus the government turned the other cheek. All of the data was collected from citizens and newspapers, who were likely related to the victims, or from the community of the victims. This data is also almost definitely under-reported which is heartbreaking in and of itself especially because the sheer number of reported lynchings was incredibly high on a weekly basis. The visualization demonstrated just how many lynchings there were by showing a near opaque red graph for 40 years until it started to dissipate around the 1920s. Once again, this was only reported data- and the graph only shows these years on the axis, but that doesn’t mean that the lynchings didn’t extend far beyond it.

Michael’s annotation that highlights the importance of whose voices are represented in the data strikes me as one of the key arguments D’Ignazio and Klein’s article Feminist Data Visualization. Because data collection is a process designed by the collector, it is inherently biased and controlled by the collector. This consequently limits the scope of the data to the resources possessed by the collector. Feminist theory in the context of data visualization aims to enlarge this scope to be as inclusive as possible by emphasizing the perspectives of many who have been marginalized and whose voices have historically been excluded. Yet, Klein and D’Ignazio recognize that even feminist theory runs into limitations when considering those who are gender non-conforming or transgender.

Michael notes how this question is a recurring theme in many of the articles we are reading.  This idea reminded me of the Han Rosling’s 200 Countries 200 Years video which briefly addressed the silencing of certain narratives through averages. For example, the video addressed the intersection of life expectancy and wealth by analyzing country averages for both categories. Rosling acknowledged that for certain countries, say China, when split into counties, had numbers that fell all over the graph and far from the average. Specifically, more rural provinces were poorer and had shorter life expectancies whereas Shanghai had higher levels of wealth and longer life expectancies. Rosling, the designer of the visualization, could have omitted dividing China into provinces and thus left out the voice of the poorer and more marginalized community. It was his choice to include such a narrative, but he only did so for China. Rosling therefore left many voices out of his data by using mostly averages per country.

Feminist Data Visualization aims to minimize the voices left out of data collection, making inclusive conclusions and visualizations.

Feminist Data Visualization Response

This was a very interesting article that had me alter the way that I thought about data. I had been exposed to feminist theory throughout my whole life whether it be articles, school, or other situations but I never had connected it to data and visualizing research. I realized also my close mindedness think that feminism was just about women but rather it draws our attention to question of epistemology. It was also very interesting that it drew upon “a set of canonical and contemporary theories from the humanities.” These theories we used to show emphasis to the nature of knowledge and perception. I also learned a lot within the section titled Feminist human computer. Specifically, about peoples work on the exploration of the implications of feminist theory and learned a lot about how the human-computer interaction and how this helps promote these visualizations. Not only this but also digital humanities as a whole. There are still some things to be critical with this article however. We have to think of the context which is from a particular social, cultural, and material context. This makes the visualization created from this feminist approach so one must think have this output might be received.

Hans Roslings’ “200 countries, 200 years, 4 minutes” video was incredibly fascinating. Roslings took thousands of pieces of data and graphed 200 countries based on their life expectancy and income per person. To begin, he showed what the graph looked like in the early 1800’s. Almost all the countries were generally in similar areas, but as he changed the time and countries went through depressions, wars, and genocides, they began to grow apart. Some countries became better off and others fell. Finally, he played the simulation all the way through to his most current data. In the long run, the countries on the map started to converge higher on the graph (meaning higher life expectancy and higher income per person). This simulation very intriguing to watch because it showed how countries would rise and fall based on wars, depressions, and other hardships. Similarly, Roslings was effective with his presentation of data because it was interesting to watch and not too long. Overall, I think this is an example of a good way to present large amounts of data to an audience, and I really enjoyed watching this video. Similar to one of my classmate who wrote about this video, I agree that the graph is worth one thousand words. Seeing the countries move around the graph is incredibly interesting, and would be hard to replicate without the use of this graph.