We Know How You Feel – Online Sentiment Analysis and the Budding Field of “Content Activism”

Posted: August 23, 2011 in Advertising, Branding, Industry Research, Marketing, Public Relations (PR), Search, Social Media
Tags: , , , , ,

In an article on Mashable which I read today, the author’s argument is that how we search can tell us a lot about how we feel about the search object or subject. Basically, researchers are now conducting rhetorical analyses of the most commonly searched terms and how they are written in order to understand better how a country’s internet users skew on a number of diverse subjects: products, brands, services; and more recently now: race, ethnicity, socio-economics, politics, etc. The article is entitled, Search Stereotypes: What Web Content Reveals About Cultural Biases, and it most closely examines how Latinas are depicted in the context of search terminology.

This is very interesting to me, and it’s a large reason why I still believe a career in search marketing or search analytics would be incredibly edifying and interesting, especially at the sociological level. What new more socially and culturally oriented organizations are trying to do is to manually re-adjust those perceptions through the same media that the information is coming to them on.

As the author, Sandra Ordonez, writes “The online stereotype of the hyper-sexualized Latina is simply not true. Statistically, most Latinas are the exact opposite: smart moms with families. This is exactly why we launched Mamiverse — to fill the websphere with more content that is truly reflective of who we are,” Martinez said. It’s basically fighting fire with fire- analyzing search engine content for racial, sexual and cultural bias and then manipulating search results in the name of promoting a social cause, rather than for a product or company, and they’re calling it “Content Activism.”

Many months ago now I wrote about sentiment analysis and the Affective Norms for English Language index that allows for this type of rhetorical analysis. This approach has been applied to Twitter and the search engine optimization/marketing fields for use in better targeting and improving public brands in the public view.

Basically it all comes down to diction- word choice. Whether we each think about it every day or not, the choices we make in vocabulary to describe things can tell our audiences a lot about how we feel about the things we are describing. Additionally, whether we think about it or not, every time we type a word or search term into a Google box, we are sending a transmission to an audience that will never disappear. The ongoing record of data and information that Google represents is like a sandbox for academics, and a historical record for the rest of us. That, in itself, is a striking image.

The article lists a number of racialized and gendered search terms and their most commonly returned meta search terms- a table which I found very interesting:

Top Three Search Results and Suggested Group Information


The following descriptions are based solely on meta site descriptions found after each title. The actual sites returned were not visited. This is only a sample of searches conducted.

  • Latinas: All three results are pornographic. Descriptors include “hot,” “young” and “legs wide open.” Seems to hint that women on 8th street, a historic street in an internationally-known Latin neighborhood in Miami, are sexy and “doable.” (Search for Latin women results in various dating sites for men seeking Latin women).
  • Latin Men: Since “Latinos” is used to describe an entire group of people, we used the term “Latin men.” One link is for a site that helps you secure a stripper or exotic dancer. The other two links are porn sites. Descriptors are a bit too pornographic to list in this article.
  • Asian Men: Two links for the “Angry Asian Man” blog and an article exploring whether Asian men are good in bed. Only descriptors are found in one sentence that directly addresses stereotype: “We all know the stories about Asian men’s sexual prowess, or the lack thereof, and the age old jokes about the ‘size issue.’”
  • Asian Women: Two links for organizations that provide Asian women with domestic violence support (NYAWC and SAKHI) and a dating/marriage site for single men looking for both Asian women and Asian girls. Only descriptor includes the word “survivor.”
  • Black Men: A Wikipedia article describing the term “black men,” an article explaining why white women prefer black men, and a link to Black Men Magazine, which seems to focus on pop culture and sexy women. No descriptors, but phrases include “racial” (comes up twice), “mugger button” and “Ink Candy Party.”
  • Black Women: A link to a “black women’s interracial marriage site,” a link to “Black Voices News and Opinion” on The Huffington Post and a Wikipedia article describing the term “black people.” Descriptors include “slaves” and “enslavers,” with a sentence describing them as “surviving.”

I highly recommend checking out the article, which does a good job of acknowledging that the search results analysis is not an entirely accurate snapshot of the world’s views simply because it self-selects according to a higher echelon socio-economically. In other words, not everyone in the world has access to the Net, so the sample bias is definitely a source of some anguish in this budding field.

Still, the area of “Content Activism” connected with sentiment analysis is an interesting one- food for thought!

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s