Presenter Information

Luke Daniel Cross

Description

Our era of click bait, fake news and politicized media has cast a critical spotlight onto deceptively structured headlines and how they influence readership. Much of the existing research delves into the implications of phrasing on article headlines -- it is common to find that even minor shifts in the focus of a headline alters what aspects of the article are remembered. While this is a prime example of the need for conscientious, accurate reporting, it offers little for journalists already taking steps to ensure precise diction and clarity. Instead of analyzing how a misleading headline explicitly alters the way a reader views content, this research attempts to explain how the sentiment inherent to an otherwise fair headline implicitly alters a readers reaction to the article. Even the most objective headline likely uses language rife with implicit sentiment, often out of necessity. For example, “Senator cuts budget” is meaningless without context, but negative connotations associated with the word cut psychologically prime the reader to assume negative sentiment going into the article. Using the New York Times archives of online articles and their corresponding comments, this theory is tested by passing said comments and headlines through a Naïve Bayes classifier trained on film reviews to quantify sentiment, providing a means of comparing headline sentiment polarity to corresponding comment polarity.

Comments

Presented as part of the 2019 USFSP Undergraduate Research Symposium held April 16, 2019.

Share

COinS
 

Analyzing the Sentiment of The New York Times

Our era of click bait, fake news and politicized media has cast a critical spotlight onto deceptively structured headlines and how they influence readership. Much of the existing research delves into the implications of phrasing on article headlines -- it is common to find that even minor shifts in the focus of a headline alters what aspects of the article are remembered. While this is a prime example of the need for conscientious, accurate reporting, it offers little for journalists already taking steps to ensure precise diction and clarity. Instead of analyzing how a misleading headline explicitly alters the way a reader views content, this research attempts to explain how the sentiment inherent to an otherwise fair headline implicitly alters a readers reaction to the article. Even the most objective headline likely uses language rife with implicit sentiment, often out of necessity. For example, “Senator cuts budget” is meaningless without context, but negative connotations associated with the word cut psychologically prime the reader to assume negative sentiment going into the article. Using the New York Times archives of online articles and their corresponding comments, this theory is tested by passing said comments and headlines through a Naïve Bayes classifier trained on film reviews to quantify sentiment, providing a means of comparing headline sentiment polarity to corresponding comment polarity.