A new high-tech personality profiling method finds that US President Barack Obama was “unfriendly” when talking about Israel and the Palestinians in a recent interview in The New York Times — a limited finding that, it should be stressed, cannot be taken to reflect the president’s overall disposition on the subject.
The computer-based method examines texts to characterize personality in a given context. It has not been fully validated and does not definitively test policies or intentions. Even so, the results would reinforce the notions of many Israelis and Palestinians who consider Obama and his administration unfriendly to their respective peoples.
“When I saw the results, I simply started laughing,” said Prof. Yair Neuman, a researcher at Ben-Gurion University of the Negev, who developed the method. “If it would have been different, fine. But it is in line with the suspicions of some Israelis.”
Neuman says, in this context, the dimension of unfriendliness means a person is uncomfortable discussing a certain subject and is trying to avoid talking about it.
‘Obama is basically saying, Guys, stop bothering me with these Israelis.’
By comparison, the analysis indicates that Hillary Clinton, considered the leader in the race for the Democratic Party’s nomination for president in 2016, didn’t express negative emotions about Israel and the Palestinians in a recent interview she gave the Atlantic magazine.
Neuman’s method, which he describes as augmented intelligence, works with any text of a person’s words – from a speech, interview, essay, or online post. The text is processed by computer software he developed. An expert must interpret the results.
Among other things, Neuman has used the program to profile political leaders and school shooters, to predict behavior on Twitter, and to evaluate online customer reviews. He did a custom analysis of Obama and Clinton’s interviews for The Times of Israel.
Excavating the unconscious
Obama and Clinton both sat down with journalists during the first week of August to discuss at length their views on Israel and the Middle East – Obama with columnist Thomas Friedman of The New York Times and Clinton with correspondent Jeffrey Goldberg of the Atlantic.
The Times of Israel provided Neuman with transcripts of the portions of both interviews that focused on Israel and the Palestinians for analysis. Neuman’s main finding is that Obama was less friendly than Clinton was when discussing the subject.
“Obama is basically saying, Guys, stop bothering me with these Israelis. Just leave me alone. I don’t want to deal with them,” said Neuman. “It is extremely different from Clinton’s speech. She is not uncomfortable speaking about the subject.”
Pundits have said of the interviews that Clinton, who served as secretary of state during Obama’s first term, was aiming to articulate a more robust foreign policy, especially in the Middle East, and to bolster her pro-Israel credentials in anticipation of a second presidential run in 2016.
Neuman partially validated his method in a study earlier this year, successfully using it to predict the personality types of some 2,500 university students based on stream-of-consciousness texts they wrote.
In soon to be published studies, he profiles and explains the actions of ousted Egyptian president Mohammed Morsi and the perpetrator of the 2007 Virginia Tech massacre based on their words. And in unpublished research, he accurately predicts whether Twitter users will retweet rumors and determines which online customer reviews to take seriously.
“The software excavates hidden, unconscious dimensions of personality,” said Neuman. “What it has done so far is only the tip of the iceberg.”
Reading between the lines
The program is highly abstract and complex. In crude terms, it works by creating a “semantic signature” of a person’s words and compares it to semantic signatures of well-established personality dimensions.
Words and phrases are given semantic signatures based on the contexts in which they appear in English-language databases, like Wikipedia. The signatures are defined by which words the words and phrases appear alongside, which words those words appear alongside, and so on. Frequency and proximity are what count.
Built into the program are semantic signatures for dozens of personality dimensions, including the ”Big Five” personality types – extroversion, agreeableness, conscientiousness, neuroticism, and openness to experience – as well as emotional and psychodynamic characteristics. The signature of each dimension is based on a handful of words that describe it in the psychological literature.
For example, the profile of unfriendliness is based on the use of these words: remote, solitary, apathetic, and indifferent. Unfriendliness is a psychodynamic characteristic, meaning it has unconscious causes. This dimension of the program has not been systematically validated.
The semantic signatures of the words from the text are compared to those of the personality dimensions. The more similar the signatures of the words are to that of a dimension, the higher the person in question scores on that dimension of personality.
Neuman says his program does not allow him to profile a person’s entire personality – just how it is expressed in context. A person’s true character, he says, becomes clear only with time.