Amazon as Issue Engine: Islands of Weird

Team Members

Anne Helmond, Erik Hekman, Kate Miltner, Molly Kalan, Bazilah Talip, Sarietha Engelbrecht (thanks Bernhard Rieder & Erik Borra!)

Introduction

Most recently, Mizuno sneaker product page emerged as a site on which Amazon reviewers shared their opinions via snarky or sarcastic reviews - other reviewers have done this before, on both politically charged products or products that may be found outrageous or ironic (i.e. the Hutzler banana slicer, or the 3 moon wolf t-shirt). We were interested in exploring mock reviews on Amazon, which - according to Amy Johnson, "challenge traditional constructions of the public sphere, for they re-imagine it with regard to discourse, participants, and venue" (2013). Mock reviews on Amazon work within the space created by Amazon.com itself; community members engage in snark and sarcasm within this system. Within the scope of the “snark comment” community, we considered 3 products that focus on specific issues, each touching on a political or gendered issue (Mizuno sneakers, Avery binder, Bic pen for her).

Case Studies

Mizuno Wave Rider 16

Wendy Davis prepared for a 13 hour filibuster to prevent the passing of a bill that seriously restricted abortion laws in Texas; she wore a pair of pink Mizuno Wave Rider 16 sneakers. During the filibuster she wasn’t allowed to sit, or even lean on anything, for the entire time. Comments on the Amazon product page for the Mizuno sneakers rally around Wendy (#StandWithWendy and #istandwithwendy were being used on Twitter) and her “taking down of the patriarchy.” The Amazon review of the Mizuno sneakers received a lot of press coverage. It appeared on Jezebel and major news sites including The Washington Post,Time, ABC News and the NY Daily News. Within 48 hours of the filibuster, upwards of 40 Davis-related reviews were submitted to the Mizuno Wave Rider Amazon page. The product page of the shoes became a space in which people commented on this political moment - generally supporters of Davis’ actions.

Avery Binder

During the second debate in the 2012 US presidential election, the phrase “binders full of women” was used by Mitt Romney in reference to women in the workplace and gender equality more broadly. There was an immediate response to what was largely regarded as a political gaffe, as image macros featuring the phrase circulated on Twitter and a Tumblr blog was created cataloguing the images. Reddit and Facebook quickly became other places where the phrase was used. The same day of the election, snarky Amazon reviews began being posted online.

Bic for Her

The Bic For Her pen was released in 2011, but was widely covered in the news after Amazon reviews began to accumulate. Jezebel wrote an article in 2011 that was in the same snarky tone as the reviews. Reviews started to pick up at the end of August 2012 and they received a great deal of coverage online. In October Ellen Degeneres spoofed the idea of a pen for women on her show. In April the following year George Takei shared the link to the review on his Facebook page which lead to another huge spike in reviews and comments.

Research Questions:

  1. To what degree is there overlap between the various mock commenters in Amazon? (Is there a mock review culture?)

  2. Is this activity being generated organically or through organized activity? Can we trace this?

Tools

  • To collect and process Amazon reviews and corresponding data we built our own scraping software (node.js). This tool collected information about 31 Amazon products which were later used for various data analyses.

  • We used Gephi for data visualisation

  • Google Refine was used to filter data.

Data Set

We worked with 31 products stored in 1,915 separate html files. Amazon product list spreadsheet.

Methodology

Step 1: Select products

  • The products that we focused on were manually identified based on mentions and consistent coverage in news articles. Core review article within this space, snowballed through recommendation engine itself by other people have viewed. We looked at amazon.com, lolreviews.com and knowyourmemes.

Step 2: Scrape data

  • For each product page, reviewer names and attributes or the review and reviewer were scraped.

  • We went into the comment history of reviewers to see if they have also reviewed other products. For each reviewer we scraped other products they have reviewed too.

  • List of products + reviewers per product.

  • List of reviewers + list of products they have reviewed

  • The reviews on Amazon were collected on Wednesday (3th of July). For every product the amount of reviews (and thus subpages) were determined. Each product and corresponding subpages were collected and stored locally for further processing. The data was exported to CSV for later further analyses and correlations.

Step 3: Create Networks

  • We combined the data and created network using Table2Net

  • We then imported this data into Gephi to show us the main commenters and connections between products and commenters to see whether product reviews were issue-based or user-based.

Step 4: Plot reviews over time to indicate spikes and find sources of spikes

From 31 products, we used three products for deeper analysis: The Mizuno Wave Rider 16 sneaker, the Avery Binder and the Bic for Her pen. The following steps were used to create a line graphs of reviews on certain days in order to show spikes in reviews.

  1. The CSV files for these products were imported to Google Refine. Selections were then exported into Excel for graph development.

  2. In Excel: Go to date > edit cells >common transforms >to date

  3. Copy all dates. Paste dates into a separate empty column on the right of your document

  4. Go to DATA menu > Advanced Filter > Unique records only

  5. Once all unique dates have been filtered, copy list

  6. Add a new column (“UNIQUE DATES”), paste into sheet

  7. In new column next to UNIQUE DATES, type = COUNTIF (range, criteria)

  8. Range = ALL dates (2nd column)

  9. Criteria = First date

  10. Then drag formula all the way down to the final unique date for autofill

  11. Copy unique dates and date counts and Paste Special -> VALUES AND NUMBER FORMATS

  12. Select all data: CREATE CHART

Visualizations

Amazon_BigPen_OverTime.jpg

Amazon_RevOverview.jpg

Amazon_RevOverviewZoom.jpg

Findings

Amazon product recommendation:

  • There are clusters of products (or “Islands of Weird”) being recommended by category. The only interesting connection that stood out was Bic For Her and the Banana Slicer which were connected to each other.
Amazon reviewers community:
  • Mock commenters look for humorous/ironic products on which to comment. Some commenters do this on multiple funny products, and even reference other products. The level of referentiality suggests identification with this community.
  • Mock commenters gather during political events and take on the style of the mock comments in this "snark community".
  • Amazon's Top Reviewers are those reviewers whose reviews are found to be most helpful.

Two mock review spaces in Amazon:

  1. Political issue space
    1. One-off comments
    2. Largely driven by news
  2. Amazon Reviewer Community (“LOLReview”) space
    1. Performance of community through core product corpus and intertextual references
    2. Heavy involvement from Top Reviewers

Case studies - timeline:

  • Reviews for Mizuno sneakers spiked (and engagement level remained high) when Wendy Davis filibustered and received news coverage.
  • The Avery binder received a lot of coverage during the American presidential debate when Mitt Romney referred to “binders full of women”. News about this spread on Reddit, news sites and a Facebook group formed. We see a huge spike in comments around this time that then tapers out.
  • The Bic Cristal For Her pen received a great deal of coverage and reviews when influential celebrities such as Ellen Degeneres and George Takei mentioned the product and review page.

Project Slides

References

'Imagined Connections: Mock Amazon Reviews & the Public Sphere' by Amy Johnson

http://mashable.com/2013/02/15/amazon-review-bombs/

http://www.geekosystem.com/takei-amazon-reviews/

http://www.dailydot.com/entertainment/funniest-amazon-reviews-pens-uranium-wolf-shirt/

http://www.smosh.com/smosh-pit/photos/20-hilarious-amazon-reviews

http://www.apartmenttherapy.com/laugh-break-five-of-the-funniest-amazon-product-reviews_1-183577
Topic revision: r2 - 06 Jan 2014, AnneHelmond
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback