The Couple That Fakes Their Own Paparazzi Photos

.
Sara Morsillo and Samuel Batista. Photo: @zoomthashit/Instagram
Follow Me is a weeklong series about personal brands, for better or for worse.

A few weeks ago, the Cut’s social media editor, Nana Agyemang, sent me a link to an Instagram profile. “I’m obsessed with this couple,” she wrote in all caps. “I think they live in Paris? Maybe Spain? I can’t tell.”

The account was only three months old and had a one-word bio, “Samsara,” spelled out in a cool Gothic script. It featured candid-style photos of an attractive couple wearing impeccably coordinated outfits and eating fast food that matched their clothing. In one, they’re walking through a parking lot in head-to-toe mauve, looking sleepy or possibly hungover; he’s stuffing his face with a pink-frosted donut that’s the same color as the lighter she’s just used for her cigarette. They look at once tacky and glamorous, posed and awkward, solemn and ridiculous. I loved it. And so did their 26,200 followers, who — like me — couldn’t figure out who they were.


“Are you guys actors, models — what do you do to be so perfect?” demanded one commenter. Another: “Who takes these photos and who are you people?” Samsara replied coyly with “thank u” and heart emojis, respectively.

Nana and I did some stalking of our own, but came up empty. A friend of mine thought she recognized the woman from Love & Hip Hop, but it was a false lead. Another colleague was briefly certain that the guy was a Formula 1 racer. I speculated that the whole thing was some photographer’s weird art project, a Cindy Sherman-esque riff on fame and social media.

People comment on our profile, ‘Who are you?’ But actually, I think they don’t want to know.
As it turns out, they’re two Italian fashion students, Sara Morsillo and Samuel Batista, who met a year ago at the Sapienza University of Rome. (Their bio — Samsara — is a combination of their names.) When I reached them via WhatsApp video chat, they explained they were in the middle of moving to Milan, where they will continue their studies. Sam, 21, also works full-time at a bar; Sara, 22, manages the account. They conceptualize and style the images together, and take the photos themselves using a tripod.

“On social media, it doesn’t matter who you really are,” said Sam. “People comment on our profile, ‘Who are you?’ But actually, I think they don’t want to know. They prefer the idea that you are someone. And this gives us many possibilities.”

The couple came up with the idea for the account last summer. “Our concept is to show the three top things that people like to show on social media: love, fashion, and food,” said Sara. When I asked if the idea was to parody an aspirational lifestyle, she nodded vigorously. “In Italian, it is aspirazionale — you would like to be this way.” On Instagram, they look like rich, well-dressed, in-love celebrities captured in the midst of their day-to-day lives; in reality, they are regular students in $20 pants from Asos, painstakingly composing each outfit and photo. “It is meant to be — satirico?” Sam paused, bending over his phone to look up the word. “Satire.”

Get unlimited access to The Cut and everything else New York
LEARN MORE »
We started it as sort of a joke. But now it is getting serious.
“I’m sure that if, tomorrow, I started to put my real life on social media, people would stop following me,” he continued. And that’s part of the point: “Outside of social media, this reality does not exist.” Even for actual famous people, added Sara. Their Instagram handle, @zoomthashit, was inspired by the common urge to zoom in on photos of celebrities and search for flaws. “Like cellulite,” she said gleefully, miming a pincerlike zooming motion with her thumb and forefinger.

Both are astonished by their newfound influencer status. “We started it as sort of a joke. But now it is getting serious,” said Sam. As for what’s next, they expressed interest in figuring out how to make this project their full-time job. The irony is, of course, that they might turn into the authentic version of what they originally meant to parody. In showing that you don’t have to be rich or famous to look rich and famous online, perhaps they’ll become both.

In fact, the crossover has already begun. Brands have started reaching out, and the couple now accept free merchandise, which they tag in their photos. In a recent post, they’re dressed in sharp, futuristic black-and-white outfits from the unisex brand Biskit, walking next to what appears to be a traffic circle and clutching bananas. They look like models beamed down from space, especially compared to the genuinely normal-looking person in jeans and a T-shirt who just happened to wander into the frame behind them.

.
The Couple That Fakes Their Own Paparazzi Photos
By Charlotte Cowles
https://www.thecut.com/2019/11/couple-that-fakes-their-own-paparazzi-photos-fashion.html

How The 2010s Killed The Celebrity Gossip Machine

.
In 2010, celebrities were beholden to swarms of paparazzi and the ever-present threat of TMZ. A decade later, they’re back in control.

Picture of Anne Helen Petersen
Anne Helen Petersen
BuzzFeed News Reporter
Posted on December 26, 2019, at 11:27 a.m. ET

Grosescu Alberto Mihai / Shutterstock

Charley Gallay / Getty Images
Back in the winter of 2011, I was sitting in my tiny apartment in Austin, Texas, finishing my dissertation on the history of celebrity gossip. Starting all the way back at the beginning of what we now know as Hollywood, I traced the evolution of how stars were created, packaged, sold, and consumed, from Mary Pickford through Britney Spears.

But writing a dissertation is dull, solitary, deeply unglamorous work, and when a producer emailed me to ask if I’d be in a documentary about “the business of celebrity,” I jumped at the chance. They flew me to New York, filmed me at Sarah Lawrence to give me the “aura” of an academic, and asked me questions about the past and present of celebrity. If you’ve never been part of a documentary, you likely don’t realize just how many times the “talking heads” get asked a question before landing on an answer the producer likes. “Can you answer that again, but much more condensed?” “Can you say the same thing, but with one word?”

I was deep in academia at the time, where answering with one word, about anything, felt like blasphemy. But they kept asking me leading questions about the effect of Perez Hilton, and the paparazzi, and TMZ: “How have they made life hell for celebrities?” “Can you talk about how they’ve ruined celebrity?”

My responses weren’t just too long, they were too unemotional. I viewed the rise of the digital paparazzi, and the gossip blogs built alongside them, in less moralizing terms. This was simply the latest pendulum swing in a century of oscillations in celebrity power. At the end of the 2000s, celebrities had found themselves largely beholden to the seemingly ever-growing swarms of paparazzi, forced to remain vigilant about how and when they appeared in public, terrified that a snippet of unflattering, unbecoming, or straight-up scandalous footage would make its way to TMZ.

Which is why the producers of the documentary kept asking me the same questions. They wanted something closer to the thesis of their film: that Perez, and amateur paparazzi, and TMZ, and the voracious appetites for content they both sparked and satiated, were ruining celebrity. It wasn’t until the film came out, a year later, that I realized the reasoning for the thesis: The film’s executive producer and director (absent the day of my filming) was Kevin Mazur, who’d spent decades photographing celebrities for Rolling Stone.

Mazur considers himself a “good guy” in the industry: the sort of guy celebrities trust, who they invited into their home, who’d never publish a photo that was unflattering or unsanctioned. Which is how he convinced Jennifer Aniston, Jennifer Lopez and then-husband Marc Anthony, Elton John, Kid Rock, and Salma Hayek to participate in the film as well, describing the paparazzi’s tactics, from the general hounding on the street to the use of helicopters to catch footage of Lopez and Anthony’s backyard wedding ceremony. Their argument, like that of the film, was ostensibly that the business of celebrity had become exploitative and dangerous for the celebrities themselves — which, given the rash of car crashes as celebrities fled brazen paparazzi, was true. The industry itself was broken, transformed from a system of honor and veneration into one of shame and denigration, which treated its products as little more than commodities to be bought and traded.


Spencer Platt / Getty Images
A subway newsstand in Brooklyn.

In many of the interviews for the documentary, I attempted to point out that this wasn’t the first time that this had happened. What’s more, the real anxiety wasn’t necessarily over the fact that celebrities were being treated as commodities — they always have been! — but that the celebrities, and the apparatus they pay to protect them, were feeling something they hadn’t felt in some time: out of control and, as such, out of power.

What began in the aughts came to a head in the 2010s, as celebrities, publicists, and the various outlets they collaborated with, from People to Entertainment Tonight, scrambled to combat narratives generated by those outside the established Hollywood system. They were pissed, and they were scared, and rightly so — the internet, and the masses of amateur photographers, content generators, and gossip sites were fickle in their affections, unpredictable in their turns. Someone as apparently beloved as Tom Cruise had quickly become an internet punchline overnight, simply by doing the same shtick he’d been doing for the last 20 years — only at the time, there was YouTube to remix the couch jump, and Perez Hilton, Lainey Gossip, and countless other gossip bloggers primed to lampoon his attempts at romance.

In the end, the solution was so straightforward. Celebrities simply became their own paparazzi.
And the options for defense were slim. You could play the paparazzi game — like Angelina Jolie and Brad Pitt, purportedly arranging for photographers to “accidentally” catch them with their newly adopted daughter Zahara, playing in domestic bliss on the beach. Or you could play the paparazzi, appearing in the same outfit every time you went outside, and thereby driving down the price of a shot. Or you could move out of Hollywood, as Julia Roberts did to raise her twins. But to extract yourself from the paparazzi game was to extract yourself from a certain brand of fame altogether — something that some celebrities, like, say, Jennifer Lopez, could not afford.

But if this decade’s start represented one extreme of the celebrity power pendulum, then 2020 marks a return swing: The stars haven’t been this powerful, with this much leverage, since the 1950s, when they first broke free from their studio system contracts, or the 1980s and ’90s, when the rise of the power publicist and agent granted them unprecedented leverage. In short, no could ever make that documentary, with that argument, today.

But the celebrities didn’t vanquish the paparazzi so much as figure out how to undercut them — and the publications they fueled. In the end, the solution was so straightforward. Celebrities simply became their own paparazzi, posting all manner of details and footage of their daily lives on social media, and effectively put real paparazzi out of business.


Michel Dufour / WireImage
Gwyneth Paltrow holds her baby, Apple Martin, as they attend the ceremony for the Chevalier de la Légion d'Honneur on Sept. 5, 2004, in Paris.

Back in 2004, after Gwyneth Paltrow gave birth to her daughter, Apple, she battled paparazzi around every London corner, all desperate for a rare shot of the baby — to the extent that Paltrow threatened legal action. When she gave birth to her son, Moses, two years later, she tried a different tactic. She simply stepped out the door, Moses cradled in her arms, and waited while the sea of photographers snapped hundreds of photos, effectively devaluing the photos, and decreasing the demand for more, at least in the short term. But Beyoncé and Jay-Z had an even better plan.

In August 2011, Beyoncé performed “Love on Top” at the MTV Video Music Awards — and at the song’s end, unbuttoned her suit jacket, pivoted slightly to the side, and rubbed her slightly growing belly. It was the Beyoncé version of a pregnancy announcement, which prompted what was then a record-breaking number of tweets per second on Twitter. When she gave birth, in January 2012, she and husband Jay-Z paid $1.3 million to rent out the entire fourth floor of Lenox Hill Hospital in New York to ensure total privacy.

Just two days after their baby was born, they released a public statement announcing the birth of a “healthy 7 lbs” girl named Blue Ivy Carter. And then there was nothing: no photos, paparazzi or otherwise, until Feb. 11, when they published five photos to...Tumblr. MeetBlueIvy.tumblr.com. It was the beginning of a total-control publicity strategy that extends to this day, whether on Tumblr, on Instagram, in Beyoncé-produced documentaries, or in Beyoncé-controlled magazine interviews and photo shoots.

The only rupture in Beyoncé’s immaculate personal narrative control came in 2014, when elevator surveillance footage of Beyoncé’s sister, Solange, attacking Jay-Z was leaked to TMZ. But it was one of the last moments of TMZ’s relevance — and instead of directly addressing what happened, Beyoncé simply dropped a remix of “Flawless” a few months later that alluded to the attack — “Of course sometimes shit goes down when there's a billion dollars on an elevator” — while actually revealing nothing about its motivation.

Beyoncé is admittedly the most extreme example of a celebrity leveraging control over their own image. But the more appropriate description might be “gold standard”: there’s what Beyoncé does with her publicity, and then there’s everyone else trying to figure out their own version of it.


Christopher Polk / Getty Images
Beyoncé, Jay-Z, and their daughter Blue Ivy Carter attend the 60th Annual Grammy Awards on Jan. 28, 2018, in New York City.

The Kardashians, for example, have never shunned the paparazzi. But they’ve savvily used social media to extend the footprint of their existing reality television stardom (see: Kim posting footage of Taylor Swift okaying Kanye’s evocation of her in “Famous” on Snapchat). Kylie opted out of the Kardashian tradition of obsessive social media pregnancy documentation, refusing to even confirm that she was pregnant — until she announced the arrival of newborn daughter Stormi to Instagram, in a post that, until recently, held the record for the most-liked photo on the platform. An 11-minute video of her pregnancy, released to YouTube and titled “To Our Daughter,” has been viewed almost 88 million times.

Instead of relying on the gossip press — which, as 2000s and early 2010s coverage of the Kardashians makes clear, changes its posture toward celebrities from week to week — the extended Kardashian family figured out how to not only craft their own story, but profit from it as well. Publications often pay celebrities for “exclusive” shots and features of their newborn children, but now, celebrities can essentially bring the operation in-house — control the format, control the timing, and reap whatever YouTube advertising dollars and Instagram marketing potential that follows. (After posting the birth announcement, Kylie gained over 3 million followers in 15 days.)

Of course, the feeling of unedited, raw access is itself an illusion.
In the past, in the midst of a scandal or a breakup, celebrities would agree to interviews to tell “their side of the story.” Now, they just post an announcement online, either on their own celebrity brand website (as Paltrow did on Goop when she “consciously uncoupled” from Chris Martin), or via screenshots using the Notes app, iMessage, or whatever weird thing was happening with Channing Tatum and Jenna Dewan’s separation announcement. The Notes app, screenshotted and posted across a celebrity’s social media channels, has become the go-to method of public apology, for anything from collaborating with R. Kelly (Lady Gaga) to stranding thousands of people on an island without adequate food or shelter (Ja Rule).

The Notes app apology provides more space than a tweet, more personality than a publicist’s statement, and — just like everything else that comes out of a celebrity’s “personal” social media account — the aura of authenticity. A Notes apology, like an Instagram selfie, gives the audience a feeling of unique access to the “real” star, seemingly unmediated by the polishing hands of publicists, editors, and photoshoppers. It has the same root appeal of the paparazzi photograph: intimacy and access — only this time, controlled by the star themselves.

Of course, the feeling of unedited, raw access is itself an illusion. During the era of Classic Hollywood, press agents ghostwrote all manner of fan magazine articles “authored” by the stars; in 2009, Britney Spears and 50 Cent both openly admitted to employing “ghost tweeters” to maintain their brand. Few celebrities would confess as much today: Some get around the fact that half of the content is clearly run by their publicity team by marking self-authored tweets, like Tom Hanks does with “Hanx.” But as Lindsey Weber points out, any effective “apology” has gone through a publicist edit. Every photo has been touched up or facetuned. Every time a celebrity comments on another celebrity’s Instagram, it’s with full knowledge that it could become a story itself.

And that’s where the power dynamic has truly shifted. The gossip magazines and blogs used to manufacture their own narratives of good and evil, slights and backbiting, weight loss and gain, employing carefully selected paparazzi photos and “sources close to the star” to prove their point. Now, they simply reprint what’s happened on social media: a collection of “Celebrity Kids Meeting Santa” is actually just a roundup of Instagram photos; a story on how Jennifer Garner’s son promised to “always” be her date is just a reprint of a caption for Garner’s latest Instagram video.


Instagram: @undefined
Again, this strategy isn’t entirely new: Publicists have always seeded stories and exclusives with gossip publications, and Garner herself was known for appearing in public parks with her children following cheating rumors on the part of then-husband Ben Affleck, essentially inviting paparazzi coverage that would steer focus away from rumored infidelity. These days, Garner isn’t forced into calculated park appearances, or hoping her publicists craft an effective counternarrative in their negotiations with the gossip magazines, or that the right quotes from interviews make it into the feature-length magazine profile. Instead, she just broadcasts her “Pretend Cooking Show” on Instagram to her 7.9 million followers — a strategy that has proven far more effective in remaking her image, and disarticulating her from Affleck, than any formal publicity maneuver.

For all the control celebrities currently enjoy over their images, ruptures do still occur — usually, as before, via the release of unsanctioned paparazzi photos. But those scandals, like news of a breakup, can be readily repaired via social media, where every comment, “clapback,” and interaction is amplified through coverage in the rest of the gossip press. The entirety of a recent post on People.com, for example, reports that Justin Timberlake left a “flirty comment” on wife Jessica Biel’s Instagram, weeks after issuing a public apology (Notes app!) to Biel after he was “caught” holding hands with his costar during a night out in New Orleans. The paparazzi may have incited the so-called scandal, but its resolution plays out entirely on the star’s terms.

Gossip has always allowed us to speak the otherwise unspeakable.
When a celebrity is shamed for weight gain, they can counter, as Rihanna did in 2017, with a Gucci Mane meme that renders the entire conversation ridiculous. A celebrity can come out — as Frank Ocean did on Tumblr, back in 2012, or Lil’ Nas X did, in 2019, on Twitter — without having to appear on the front of a magazine with the headline “I’m Gay” accompanying an interview designed to appeal to mainstream America. Tabloid photos of aging stars used to cannibalize their images, transforming them into abject horrors. Now, Jane Fonda posts a photo of herself the morning after a red carpet appearance, still in her dress from the night before. Instead of “makeup-less horror can’t even take care of herself,” she gets to caption the photo herself: “Here’s me the next morning. I couldn’t get my dress unzipped so I slept in it. Never wanted a husband in my life until now.” Instead of an aging has-been, she’s “honestly so relatable.”

It’s difficult to complain, really, about any of these shifts: Once-banal celebrity images have become more interesting and delightful; there’s far less traditional gatekeeping on the part of the gossip industry and, as a result, the celebrity world in general has become less white, less straight, and less American-dominated than ever before. Hundreds of people lost a paycheck with the end of the paparazzi boom, but it’s hard to ignore that those paparazzi regularly put their subjects in danger. And while a major, old school, social-media-resistant star like Tom Cruise may still be able to open a global franchise, he’s the last of a dying breed.

Sure, celebrity profiles are far less interesting, and celebrity trainwrecks have been replaced by the trainwreck of American politics and an ever-growing number of scammy influencers. But when I look back at the peak of the ’00s gossip area, and the abject thrill of every new picture of Britney breaking down, I remember that the gossip had never been better — but it never made anyone, least of all Spears herself, feel good. Gossip has always allowed us to speak the otherwise unspeakable, working through understandings of sex and femininity and sexuality in a space “distant” from our own. But the power of the paparazzi in the ’00s, and the relative powerlessness of the stars, drove home a difficult truth: No matter how much distance between us and the stars, the intimacy we crave has very real psychological costs, especially when extracted without consent.


Medianews Group / Getty Images
Kim Kardashian West takes a selfie with a fan in Costa Mesa, California, on Dec. 4, 2018.

Celebrities of superlative talent will always be able to allow the work, and intermittent interviews about it, speak for itself. For everyone else, the current strategy is to let social media speak for you — and use it to keep the conversation going, even when the “actual” work is disappointing, or there are no more cute babies to post. It’s the way to maintain interest, to stay relevant, to launch and maintain a celebrity brand — to publicize the commodity that is the self.

It’s hard to believe now, but back in 2009, Ashton Kutcher was the most prominent celebrity on Twitter. He was the first to reach 1 million followers and the first, at least to my knowledge, to really understand what he could with the medium. On vacation in Turks and Caicos with then-wife Demi Moore, he posted a “twitpic” (lol) of Moore, taken from behind, as she leaned over in a pair of white briefs. The next month, he and Moore posted shots — and short video — from backstage at the Oscars. All of this seems so mundane now, but at the time, it was startlingly novel. The pair had spent years being hounded by the paparazzi. Now they were taking it back.

Turning the paparazzi gaze on oneself might offer the semblance of control. But the appetite it arouses remains the same: constant, nearly impossible to satiate, always wanting more. More content, more confession, more members of the family, more drama. And that hunger transforms celebrities into primary exploiters of their own lives. This will sound familiar to anyone with even a glancing relationship to fame. Surveillance by others turns into self-surveillance; anxiety over what to wear when you go out to the grocery store expands into what to wear and how to document that same trip to the grocery store in a way that seems different, and interesting, and special.

In 2010, Instagram had just barely launched; Snapchat didn’t even exist. No one — not me, with my 400-plus-page dissertation on celebrity gossip, not the stars themselves, not even the celebrity photographer who produced that documentary — could see just how powerfully the levers of power would shift in celebrities’ favor. But a decade later, the landscape of celebrity seems almost unrecognizable from the one I was asked to describe on-camera back then.

Celebrity or not, feeling a lack of control is exhausting and psychologically unnerving, if not permanently damaging. But so is its inverse. The stars who came of age in the ’00s and early 2010s are still haunted by living through a period of unyielding, unpredictable, incredibly invasive surveillance. But the celebrities of today are living through the same — and will someday have to grapple with the fact that they were forced, implicitly and explicitly, to do it to themselves. ●
.

The ‘golden years’ of paparazzi have mostly gone

.
The ‘golden years’ of paparazzi have mostly gone

By Allison Schrager

While some exclusive celebrity photographs can command huge sums, earning a steady income in this controversial industry is getting a lot harder, writes Allison Schrager.

Santiago Baez has been a paparazzo since the early 1990s. Camera in hand, he’s witnessed the fallout of extramarital affairs, new babies, deaths, new love and breakups of some of New York’s most famous residents.

For paparazzi like Baez, earning a living requires an encyclopedic knowledge of where famous people live in New York, as well as a network of drivers, and shop and restaurant workers who call in tips when they spot celebrities in the vicinity. Often, the tips are from the celebrities themselves via social media: looking to build a following, they alert the public (mostly directed at photographers) about their movements, or their publicist will call an agency to dispatch a photographer.

Most pictures aren’t worth much, but a shot of a new baby, a celebrity kissing a new paramour, or a wedding can change fortunes overnight.

But Baez’s income is not dependably constant. His success balances his training and knowledge of celebrities with the crushing awareness that his earnings are remarkably variable and unpredictable.

The ‘paparazzi gold rush’

These fortunes are determined by a handful of people like Peter Grossman, the photo editor at Us Weekly from 2003 to 2017. But Grossman didn’t work with paparazzi directly; instead, a photographer like Baez sells his pictures to an agency that has the relationship with photo editors like Grossman. A paparazzo receives anywhere between 20% and 70% of the royalties the picture earns, depending on the photographer and the deal he or she negotiated with the agency. The more senior, skilled, and talented paparazzi command better terms, which often includes exclusively selling their pictures to just one agency.


In the heyday of paparazzi photographs, magazines like Us Weekly would pay thousands for an exclusive picture of celebrities doing mundane tasks (Credit: Alamy)

Exclusive shots that make waves in the world of tabloid news can command huge sums: Grossman told me he paid “mid six figures” for a series of photographs of the actress Kristen Stewart in a passionate embrace with Rupert Sanders, the married director of Snow White and the Huntsman, a film she had starred in.

Grossman lived through the heyday of paparazzi photography: he was the man behind the rise of “Just Like Us” pictures in the early 2000s – candid shots of celebrities doing mundane tasks like getting coffee or pumping petrol that proved a hit with his magazine’s readers. Soon, lots of outlets were publishing their own “Just Like Us” pictures, kicking off what’s known in the industry as the gold rush years, coinciding with the heyday of Paris Hilton, Britney Spears, and Lindsay Lohan.

At the gold rush peak, an exclusive ‘Just Like Us’ picture would typically fetch $5,000 to $15,000
Although the price of a photograph depended on what the celebrity was doing and whether it was an exclusive, at the gold rush peak, an exclusive “Just Like Us” picture would typically fetch $5,000 to $15,000.

The gold rush era brought about gold rush mentality, with many new photographers flocking to the industry, willing to break laws and giving paparazzi an even worse reputation for going too far and harassing celebrities and even their young children. Grossman urged everyone to take a coordinated step back, pay less for pictures, and not break laws or put themselves or others in danger to get the shot, but it didn’t work.

The global financial crisis and the rise of online media finally killed the gold rush. Digital media increased the demand for celebrity photographs but decreased the price media companies were willing to pay for them. Photo agencies began to consolidate or go out of business, and the remaining ones changed their business model. Instead of making magazines pay per photo, they offered a subscription service: publishers could use as many photos as they wanted to fulfill the greater demand for cheaper shots. As a result, paparazzi are paid a small fraction of the subscription fee; how much depends on how many of their pictures are used each month. That means an exclusive “Just Like Us” photo that would have fetched $5,000 to $15,000 before, now pays only $5 or $10.

Paparazzi are earning less and less. Gone are the days when many could count on a six-figure income. Now, getting a rare exclusive shot is necessary to earn big money.


To spread the risk around, paparazzi often form alliances – but those alliances can be upended by one big exclusive (Credit: Alamy)

Risky business

Seeing a celebrity often happens by chance, which is exactly part of the reason why Baez’s income is so volatile. Not surprisingly, Baez employs risk strategies in his craft similar to what people use in financial markets.

Financial economists separate risk into two broad categories: the first is idiosyncratic risk, or the risk unique to a particular asset. Suppose Facebook changes management; the future of the company is unclear, and the price of the stock might drop based on factors unique to Facebook that don’t impact any other stock. Idiosyncratic risk is risk that applies only to one individual stock or asset.

The paparazzi face lots of idiosyncratic risk. What a celebrity does today – whether she spends time with A-list or D-list friends, for example – determines how much the paparazzi earn that week. If a celebrity stops being interesting or popular, the value of these pictures decreases. Such images are like a stock: their value varies based on a particular photographer getting the right shot at the right time.

Photographers often form teams or alliances to share tips  to increase the odds they’ll be in that place
The paparazzi manage this idiosyncratic risk by spreading it around: photographers often form teams or alliances to share tips (on sightings) and sometimes royalties to increase the odds or payoffs they’ll be in that place.

Because each photographer bears lots of risk based on how lucky he is that day, an alliance pools their luck, reducing their idiosyncratic risk.

The second kind of risk is systematic risk, or risk that affects the larger system instead of an individual asset. Systematic risk is when every stock rises or falls together because the entire market surges or crashes as it did in 2008. Systematic risk events often happen because of a big economic disruption like a recession or an election result that people think will affect business. Systematic risks are harder to manage than idiosyncratic risks, and the downsides are potentially more dangerous. If the entire stock market tanks, you risk losing your job and stock portfolio at the same time.


There are curious parallels between Wall Street traders and the paparazzi. Both try to manage financial risk every day (Credit: Getty Images)

You can see systematic risk play out with paparazzi, like the boom of the gold rush years and the crash when people stopped buying tabloid magazines during the recession. The downside of systematic paparazzi risk has become more severe in the last 10 years. It is harder for everyone to make money. Many paparazzi have left the business: after nearly 30 years of taking celebrity photographs, Baez moved back to the Dominican Republic in the summer of 2018, with his wife and son, to find new work.

Paparazzi – just like us?

The job of a paparazzo is riskier than most. But to some extent we all face some level of idiosyncratic and systematic risk in our careers, so we can learn a lot from these photographers.

The more systematic risk associated with your job, the more exposed you are
Suppose you want to change jobs from a safe, salaried support role to a sales job based on commission. Odds are you’ll earn more than you did in the salaried job because as a salesperson you will face both kinds of risk: it is a job with loads of idiosyncratic risk; for example, how much you earn will depend on your sales skills and the behavior of your clients (you can manage this risk by working in a team and having lots of clients). You will also face systematic risk because sales depend on the state of the economy.

Systematic risk is especially dangerous. In an economic downturn, your pay may be reduced or disappear entirely, it is likely to be harder to find another job, your assets might take a hit, and your partner’s income may be at risk too. The more systematic risk associated with your job, the more exposed you are.

Why we feel so much economic anxiety

The livelihood of the average paparazzo is threatened by major changes in the publication industry. The photographers manage idiosyncratic risk by forming unstable alliances, but the larger systematic risk that could wipe out their jobs is harder to manage. They could form a union and demand better terms from the agencies, but historically they struggle to cooperate with one another. And the paparazzi are not the only ones who face the risk that their jobs will no longer be viable.

One reason people seem to worry more about their economic future than they did in the past is that they sense more systematic risk in the job market. A few decades ago, most of the employment risk was idiosyncratic: conflict with the boss, a position that was a bad fit, a poorly managed company. If you lost your job, you could probably find another one just like it. Workers formed trade unions, banded together, and demanded better pay and benefits, confident that there was a need for their skills. The job market had its ups and downs, but risk seemed to be relatively easy to manage.

In today’s economy, systematic risk is more acute. There’s a chance technology – robots and artificial intelligence – could take over your job or at least require new skills you don’t have. If you lose your job during a recession, you may never find a similar one.

It is a larger trend that threatens everyone, but for paparazzi like Baez, the threat is more immediate. It is a risky business that is only getting riskier with fewer rewards.

This article is adapted from An Economist Walks into a Brothel by Allison Schrager, published by Portfolio.
.
https://www.bbc.com/worklife/article/20190423-how-the-paparazzi-make-their-money

How social media has created a new breed of paparazzi

.
How social media has created a new breed of paparazzi

Paparazzi Miles Diggs (center) takes photos with his business partner, Cesar Peña (right).
It was the “get” of the week: the first picture of Hannah Davis’ engagement ring from Yankee icon Derek Jeter.

“I was looking for them since the engagement rumors started,” said paparazzo Miles Diggs, who runs 247Paps.TV with his partner, Cesar Peña.

They monitored Twitter and Instagram for sightings until “she finally slipped up,” Diggs said.

On Halloween night, Davis posted a photo to Instagram: her dressed as an angel in a pink wig; him in a devil costume.


Although she didn’t name the hotel, Diggs noticed distinctly patterned carpet and curtains in the background. He trawled Midtown hotel Web sites until he found rooms advertised with the same interior design.

The next morning, he posted himself outside the hotel.

“He came out first, then she finally came out and showed off that big ring,” Diggs recalled.

Snap! A $25,000 picture.

‘It’s all a hustle’
Diggs and Peña are a new breed of paparazzi, cursed and blessed by social media. Celebrity selfies and fan-snapped shots on the Internet have cut the prices they can charge for their pictures. But those tweets and Instagrams and Facebook posts allow them to track down and trail boldfaced names like never before.

By the time they get dressed, I’m already there.
 - Paparazzo Miles Diggs
“It’s all a hustle,” said Peña, 37, who lives in Harlem with his wife and 7-year-old daughter. He used to work in security at the Waldorf Astoria hotel before watching an E! network show, “Celebrity Uncensored,” which motivated him to pick up a $400 Canon video camera and start taping celebs in 2006.

Diggs, 21, joined him three years ago. Until then, he was a fan-turned-autograph hunter. He realized that while stars could turn him down for a signature, he didn’t need their permission to take their photo in public. He dropped out of studying photojournalism at NYU and has been working full time as a paparazzo ever since.

“A picture that now costs $400 was worth $15,000 to $20,000” before Instagram, Peña said. “It was harder to get pictures, harder to get stories.”


Miles Diggs (left) with his business partner, Cesar Peña (right).
When one of Nicki Minaj’s dancers posted a photo to Instagram of a rehearsal at a studio in Greenpoint, Brooklyn, with the caption “Feeling Myself” (the song Minaj recorded with Beyoncé), Diggs and Peña were able to get photos of both divas leaving the studio that night.

After Alicia Keys uploaded Instagram photos about training for the New York City Marathon, the paps headed to her New Jersey home to catch her out for a morning run.

When Kylie Jenner recently uploaded an Instagram photo of herself with model Hailey Baldwin in a fashion-shoot trailer, Diggs recognized the street and a shop in the background. He headed to the Bowery.

“We were sitting there behind the trailer before she even came out,” he said.

“Snapchat is a celebrity’s worst nightmare right now because I have an eye like a hawk,” Diggs said.

A star will post a picture of their hotel room’s view, and “by the time [they] get dressed, I’m already there,” he said.

The duo also searches Twitter for celebrity names or for phrases like “just saw Beyoncé,” “Jennifer Aniston is at” and “spotted Jennifer Aniston.”

Tricks of the trade
The practice of slipping $50 bills to doormen and bouncers for info is “a very LA mentality,” said Diggs, and not how the New York City ­paparazzi scene works.

Every day, Diggs and Peña do laps around Soho and the West Village, passing the major ­hotels — The Greenwich Hotel, The Mercer and the Trump Soho. They also have a mental map of the stars’ haunts and their homes — Taylor Swift and Beyoncé in Tribeca, Kim Kardashian and Kanye West in Soho, Rihanna in Chinatown.

Many of their tips come from what Diggs and Peña call their “little agents,” a half-dozen teens and college kids they’ve befriended who stalk celebs to take selfies with them.

No cash is exchanged, but Diggs and Peña will tip off the little agents when they spot their favorite celebrities.

“Their cut is being part of my crew. And they understand that,” Peña said.

“We have different agents for different people,” Diggs said.

“If they’re going to wait 10 hours to meet Justin Bieber, they can keep an eye on Justin Bieber while we’re on Rihanna. And if Justin Bieber happens to come out, they get their photos, and if they’re there, they’ll take video for us. Because we told them where Justin Bieber was to begin with.”

Diggs and Peña also recognize the stars’ drivers. They can guess who is in the club or restaurant by the car outside.

‘Snapchat is a celebrity’s worst nightmare right now because I have an eye like a hawk’
 - Miles Diggs
Recently, a magazine writer tipped them off that Kourtney Kardashian’s baby daddy, Scott Disick, was in town, so the pair monitored his usual haunts.

But Diggs was actually expecting singer Demi Lovato to emerge from the Trump Soho when he saw Disick step out — with an 18-year-old model .

“Boom. He just walked out the front door with a girl. As soon as they saw me, they split apart,” recalled Diggs, whose “eyes lit up with dollar signs.”

“If you’re friends with somebody, you keep walking,” Diggs said. “If you’re doing something with somebody, you split apart. It’s like an automatic guilt trip. So as soon as you see that, you know. They’re telling me that they’re sleeping together.”

Diggs and Peña followed the coy couple for five hours, waiting until Disick no longer suspected paps on their trail.

Finally, around midnight, they spotted the pair leaning on each other and play-fighting in a bar. They took their shots through their car window.

And they made sure not to create a stir, fearing it would bring attention to the couple — and spoil their exclusive.

“When you have something of that much value, after a while, it’s not about shooting more . . . it’s about protecting them from being seen,” Diggs said.

“Everybody with one of these is the enemy,” he said, holding up his iPhone. “I could shoot, with this huge camera right here, a crazy exclusive. A lady walks by — ‘Oh, this is for my grandson’ — takes a picture with her flip phone. We’re screwed.”

So Diggs and Peña waited ­until the next morning to send their work in to Splash News, their agents. That photo earned them $10,000 in the first day.

A picture’s worth
Sometimes finding celebrities is sheer luck.


Miles Diggs (left) and Cesar Peña catching up with Rihanna.
On a recent Wednesday, the pair was driving around in their tinted-windowed Chevrolet SUV when they spotted actor Bradley Cooper and his model girlfriend, Irina Shayk, walking down the street. The couple hadn’t been photographed together in months.

Diggs jumped out to follow on foot, trying to get ahead of them so he could photograph their faces.

“I got one or two frames, and they were off running,” he said.

It was a bizarre game of hide-and-seek, with the lovers trying to outrun the pap and ducking behind cars, and Diggs doing the same so they’d think he left.

Exclusives earn the most cash. One of the final video interviews with actor Philip Seymour Hoffman earned $6,000 to $7,000 for them after his death. Footage of Kristen Stewart and Robert Pattinson kissing in a car brought in $20,000. A recent exclusive of Rihanna in a Chelsea art gallery with artist Mr. Brainwash brought in $5,000.

Their biggest “get” ever was a photo and video of Beyoncé, Jay Z and Solange emerging from The Standard hotel in 2014 — right after Solange and the rapper’s elevator brawl.

After identifying Jay Z’s car by its driver, Diggs and Peña had set themselves up behind it to get the best angle. As a result, they were the only fotogs out of the dozens there to get a clear shot of all three celebs as they walked out the door.

Once leaked video of the brawl made news, Diggs’ and Peña’s shots were in high demand. Within a day, the footage brought in $35,000. Over a year, it has netted them $100,000.

What would be the greatest celebrity shot ever?

Easy: Brad Pitt and Jennifer Aniston having coffee together. They haven’t been seen together since their 2005 divorce.

“That’s like $500,000!” Diggs marveled.

“At least!” Pena laughed. “That’s the dream of all times.”
.


By Amber JamiesonNovember 8, 2015 | 6:03am
https://nypost.com/2015/11/08/paparazzi-reveal-secrets-of-tracking-stars-on-social-media/

Big data and sentiment analysis: A comprehensive and systematic literature review

 

Abstract

Sentiment analysis can extract information from many text sources such as reviews, news, and blogs; then it classifies them based on their polarity. 

Moreover, big data is produced via mobile networks and social media. 

Applications of sentiment analysis on big data are used as a way of classifying the opinions into diverse sentiment. 

Accordingly, performing sentiment analysis on big data can be helpful for a business to take useful commercial insights from text-oriented content. 

However, there are very few comprehensive investigations and profound argument in this context. 

The goal of this paper is to provide a comprehensive and systematic investigation of the state-of-the-art techniques and highlight the directions for future research. 

In this paper, we used systematic literature review method and in the first step, we obtained 15 351 articles; then, based on different filters, 48 related articles were attained. 

We have selected 23 articles based on the year of publication, the relevance of the journal, the completeness of the text, the nonrepeatability of the title, and the page number. 

Also, we have categorized big data and sentiment analysis into two classifications: centralized and distributed platforms. 

Furthermore, the disadvantages and advantages of the investigated techniques are studied and their key issues are emphasized. 

Consequently, this study shows that a better analysis of textual big data in terms of sentiment increases efficiency, flexibility, and intelligence. 

By providing comparative information and analyzing the current developments in this area, this paper will directly support academics and practicing professionals for better handling of big data in the field of sentiment analysis. 

This study sheds some new light on using sentiment analysis and big data for public opinion estimation and prediction.

REFERENCES

1. Dwivedi A, Pant R, KhariM, Pandey S,Mohan L, PandeM. E-governance and big data framework for e-governance and use of sentiment analysis. Available at SSRN 3382731. 2019.

2. Ramanujam RS, Nancyamala R, Nivedha J, Kokila J. Sentiment analysis using big data. Paper presented at: 2015 International Conference on Computation of Power, Energy, Information and Communication (ICCPEIC); 2015:0480-0484.

3. Darbandi M, Haghgoo S, Hajiali M, Khabir A. Prediction and estimation of next demands of cloud users based on their comments in CRM and previous usages. Paper presented at: 2018 International Conference on Communication, Computing and Internet of Things (IC3IoT); 2018:81-86.

4. Banic L, Mihanovi ́ c A, Brakus M. Using big data and sentiment analysis in product evaluation. Paper presented at: 2013 36th International Convention ́on Information and Communication Technology, Electronics and Microelectronics (MIPRO); 2013:1149-1154.

5. Jinturkar M, Gotmare P. Sentiment analysis of customer review data using big data: a survey. Int J Comput Appl. 2016;975:8887.

6. Sehgal D, Agarwal AK. Real-time sentiment analysis of big data applications using Twitter data with Hadoop framework. Soft Computing: Theories and Applications. Singapore: Springer; 2018:765-772.

7. Karimkhan M, Bhatia JB. Sentiment Analysis and Big Data Processing. Haryana, India: IJCSC; 2014.

8. Bohlouli M, Dalter J, Dornhöfer M, Zenkert J, Fathi M. Knowledge discovery from social media using big data-provided sentiment analysis (SoMABiT). J Inf Sci. 2015;41:779-798.

9. Kurian D, Vishnupriya S, Ramesh R, et al. Big data sentiment analysis using hadoop. Int J Innov Res Sci Technol. 2015;1:92-96.

10. Pashazadeh A, Navimipour NJ. Big data handling mechanisms in the healthcare applications: a comprehensive and systematic literature review. J Biomed Inform. 2018;82:47-62.

11. Sehgal D, Agarwal AK. Sentiment analysis of big data applications using Twitter Data with the help of HADOOP framework. Paper presented at: 2016 International Conference System Modeling and Advancement in Research Trends (SMART); 2016:251-255.

12. Wang H, Xu Z, Fujita H, Liu S. Towards felicitous decision making: an overview on challenges and trends of big data. Inform Sci. 2016;367:747-765.

13. Chandana R, Harshitha D, Ramachandra A. Big data migration and sentiment analysis of real time events using hadoop ecosystem. Paper presented at: International Conference on Intelligent Data Communication Technologies and Internet of Things; 2018:764-770.

14. Roshanfekr B, Khadivi S, Rahmati M. Sentiment analysis using deep learning on Persian texts. Paper presented at: 2017 Iranian Conference on Electrical Engineering (ICEE); 2017:1503-1508.

15. Han Z, Wu J, Huang C, Huang Q, Zhao M. A review on sentiment discovery and analysis of educational big-data. WIRES Data Min Knowl Discov. 2019;10(1):e1328.

16. Kim K, Lee J. Sentiment visualization and classification via semi-supervised nonlinear dimensionality reduction. Pattern Recogn. 2014;47:758-768.

17. Kaur P, Rupal N. A perspective: sentiment analysis and recent trend in opinion mining of Big Data. Int J Adv Computron Manag Studies. 2016;1(3):1-5.

18. Shinde-Pawar M. Formation of smart sentiment analysis technique for Big Data. Int J Innovat Res Comput Commun Eng. 2014;2:7481-7488.

19. Keshavarz H, AbadehMS, AlmasiM. A new lexicon learning algorithm for sentiment analysis of big data. Paper presented at: 2017 IEEE 15th International Symposium on Intelligent Systems and Informatics (SISY); 2017:000249-000254.

20. Minanovic A, Gabelica H, Krstic Ž. Big data and sentiment analysis using KNIME: online reviews vs. social media. Paper presented at: 2014 37th ́ International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO); 2014:1464-1468.

21. Tromp E, Pechenizkiy M, Gaber MM. Expressive modeling for trusted big data analytics: techniques and applications in sentiment analysis. Big Data Analytics. 2017;2:5.

22. Pourghebleh B, Navimipour NJ. Data aggregation mechanisms in the Internet of Things: a systematic review of the literature and recommendations for future research. J Netw Comput Appl. 2017;97:23-34.

23. Tsai C-W, Lai C-F, Chao H-C, Vasilakos AV. Big data analytics: a survey. J Big Data. 2015;2:21.

24. Sharef NM, Zin HM, Nadali S. Overview and future opportunities of sentiment analysis approaches for big data. JCS. 2016;12:153-168.

25. Graham G, Meriton RF, Hennelly P. Sentiment Analysis Using KNIME: a Systematic Literature Review of Big Data Logistics. Leeds: The University of Leeds; 2016.

26. Benedetto F, Tedeschi A. Big data sentiment analysis for brand monitoring in social media streams by cloud computing. Sentiment Analysis and Ontology Engineering. Cham: Springer; 2016:341-377.

27. Dandannavar P, Mangalwede S. Sentiment analysis of real world big data—a review of general approaches. Int Sci Press. 2017;10:185-192.

28. Sharma S, Bansal M, Kaushik A. A survey on sentiment analysis for big data. Int J Adv Res Sci Eng. 2017;6:412-416.

29. Balaji SN, Paul PV, Saravanan R. Survey on sentiment analysis based stock prediction using big data analytics. Paper presented at: 2017 Innovations in Power and Advanced Computing Technologies (i-PACT); 2017:1-5.

30. Wamba SF, Akter S, Edwards A, Chopin G, Gnanzou D. How 'big data' can make big impact: findings from a systematic review and a longitudinal case study. Int J Prod Econ. 2015;165:234-246.

31. Qiu J, Wu Q, Ding G, Xu Y, Feng S. A survey of machine learning for big data processing. EURASIP J Adv Sig Process. 2016;2016:67.

32. Russell S, Norvig P. Intelligence Artificielle: Avec plus de 500 Exercices. France: Pearson Education; 2010.

33. Nilsson NJ. Artificial Intelligence: a New Synthesis. Burlington, MA, United States: Morgan Kaufmann; 1998.

34. Cheng OK, Lau R. Big data stream analytics for near real-time sentiment analysis. J Comput Commun. 2015;3:189-195.

35. Fang Y, Chen X, Song Z,Wang T, Cao Y. Modelling propagation of public opinions on microblogging big data using sentiment analysis and compartmental models. Int J Seman Web Inform Syst. 2017;13:11-27.

36. Ragini JR, Anand PR, Bhaskar V. Big data analytics for disaster response and recovery through sentiment analysis. Int J Inf Manag. 2018;42: 13-24.

37. Troisi O, Grimaldi M, Loia F, Maione G. Big data and sentiment analysis to highlight decision behaviours: a case study for student population. Behav Inform Technol. 2018;37:1111-1128.

38. Lau RYK, Zhang W, Xu W. Parallel aspect-oriented sentiment analysis for sales forecasting with big data. Prod Oper Manag. 2018;27: 1775-1794.

39. Chen CIP, Zheng J. Improved big data analytics solution using deep learning model and real-time sentiment data analysis approach. Paper presented at: International Conference on Brain Inspired Cognitive Systems; 2018:579-588.

40. Martínez-Castaño R, Pichel JC, Gamallo P. Polypus: a big data self-deployable architecture for microblogging text extraction and real-time sentiment analysis. arXiv Preprint arXiv:180103710; 2018.

41. Khezr SN, Navimipour NJ. MapReduce and its application in optimization algorithms: a comprehensive study. Majlesi J Multimedia Process. 2015;4:1-5.

42. Dean J, Ghemawat S. MapReduce: a flexible data processing tool. Commun ACM. 2010;53:72-77.

43. Almeer MH. Cloud Hadoop map reduce for remote sensing image analysis. J Emerg Trend Comput Inform Sci. 2012;3:637-644.

44. Nirmal VJ, Amalarethinam DG. Parallel implementation of big data pre-processing algorithms for sentiment analysis of social networking data. Int J Fuzzy Math Arch. 2015;6:149-159.

45. Povoda L, Burget R, Dutta MK. Sentiment analysis based on support vector machine and big data. Paper presented at: 2016 39th International Conference on Telecommunications and Signal Processing (TSP); 2016:543-545.

46. Htet H, Khaing SS, Myint YY. Tweets sentiment analysis for healthcare on big data processing and IoT architecture using maximum entropy classifier. Paper presented at: International Conference on Big Data Analysis and Deep Learning Applications; 2018:28-38.

47. Dwivedi A, Pant R, Pandey S, Pande M, Mittal AK. Benefits, of using big data sentiment analysis and soft computing techniques in E-governance. Int J Recent Technol Eng. 2019;8:3038-3044.

48. Zhang Y, Ren W, Zhu T, Faith E. MoSa: a Modeling and sentiment analysis system for Mobile application big data. Symmetry. 2019;11:115.

49. Rahnama AHA. Distributed real-time sentiment analysis for big data social streams Paper presented at: 2014 International Conference on Control, Decision and Information Technologies (CoDIT); 2014:789-794.

50. Liu B, Blasch E, Chen Y, Shen D, Chen G. Scalable sentiment classification for big data analysis using naive bayes classifier. Paper presented at: 2013 IEEE International Conference on Big Data; 2013:99-104.

https://onlinelibrary.wiley.com/doi/abs/10.1002/cpe.5671

Sentiment Analysis of Textual Content in Social Networks From Hand-Crafted to Deep Learning-Based Models [Problem Statement]

 

.

...

 Most of the existing systems on sentiment analysis rely heavily on a rich set of sentiment resources (such as text corpora with manually annotated sentiment polarity, sentiment lexicons, and word embeddings). However, the distribution of sentiment resources is very imbalanced among languages. Thus, building a sentiment analysis system in low-resource languages requires tremendous human effort to construct such resources, which is a time-consuming and expensive task.

.

Although opinions are the most common shared content in online social networks and forums, people also tend to share their emotions which are the keys to their feelings and thoughts. Emotion analysis is the task of determining the attitude towards a target or topic. The attitude can be the polarity (positive or negative), or an emotional state such as joy, anger or sadness [101–103].

.

1.2 Objectives

In this thesis, we aim to develop methods to automatically analyse textual content shared on social networks and identify people’ opinions, emotions and feelings at different level of analysis and in different languages. As such, we introduce the following set of goals:

.

• To develop efficient sentiment analysis systems based on new features that can be used with traditional Machine Learning (Machine Learning) methods. Towards this objective, we aim to use pre-built resources such as sentiment lexicons and word embeddings to extract new features.

.

• To move sentiment analysis beyond sentence-level and polarity-based analysis. Towards this objective, we are interested in:

– Proposing ensemble systems that combine classical Machine Learning models with Deep Learning to analyse emotions expressed on Twitter.

– Developing a Deep Learning based model to solve the problem of multi-label emotions classification.

– Utilising Deep Learning based models to analyse opinions at the aspect level.

.

• To move sentiment analysis beyond a single language.

We aim to utilise the concept of transfer learning to develop a system that can transfer sentiment knowledge from high resources languages to low resources languages. The final goal of this objective is to obtain a universal sentiment analysis system that works with low resource languages and does not require machine translation.

.

• To test the developed systems on real cases of analysis

.

• The idea of this objective is to test our developed systems in real applications and different domains. Hence, towards this objective, we define the following sub-goals:

– To collect tweets of local people, visitors and official brand destination offices from different tourist destinations and analyse the opinions shared in these tweets.

– To combine the aspect-based sentiment analysis with multi-criteria decision aid systems to improve the decision-making process.

.

https://deim.urv.cat/~itaka/itaka2/PDF/acabats/PhD_Thesis/TESI_Mohammed_Jabreel.pdf

.

.

Sentiment and position-taking analysis of parliamentary debates: A systematic literature review

 

Abstract

Parliamentary and legislative debate transcripts provide access to information concerning the opinions, positions and policy preferences of elected politicians. 

They attract attention from researchers from a wide variety of backgrounds, from political and social sciences to computer science. 

As a result, the problem of automatic sentiment and position-taking analysis has been tackled from different perspectives, using varying approaches and methods, and with relatively little collaboration or cross-pollination of ideas. 

The existing research is scattered across publications from various fields and venues. 

In this article we present the results of a systematic literature review of 61 studies, all of which address the automatic analysis of the sentiment and opinions expressed and positions taken by speakers in parliamentary (and other legislative) debates. 

In this review, we discuss the available research with regard to the aims and objectives of the researchers who work on these problems, the automatic analysis tasks they undertake, and the approaches and methods they use. 

We conclude by summarizing their findings, discussing the challenges of applying computational analysis to parliamentary debates, and suggesting possible avenues for further research.

REFERENCES:

1. Abercrombie, G., & Batista-Navarro, R. (2018). ‘Aye’ or ‘no’? Speech-level sentiment analysis of

Hansard UK parliamentary debate transcripts. In: Proceedings of the eleventh international confer-

ence on language resources and evaluation (LREC-2018). European Languages Resources Associa-

tion (ELRA), Miyazaki, Japan. https://www.aclweb.org/anthology/L18-1659.

2. Abercrombie, G., & Batista-Navarro, R.T. (2018). Identifying opinion-topics and polarity of parlia-

mentary debate motions. In: Proceedings of the 9th workshop on computational approaches to sub-

jectivity, sentiment and social media analysis. Association for Computational Linguistics, Brussels,

Belgium (pp. 280–285). https://doi.org/10.18653/v1/W18-6241. https://www.aclweb.org/anthology/

W18-6241.

3. Ahmadalinezhad, M., & Makrehchi, M. (2018). Detecting agreement and disagreement in political

debates. In R. Thomson, C. Dancy, A. Hyder, & H. Bisgin (Eds.), Social, cultural, and behavioral

modeling (pp. 54–60). Cham: Springer.

4. Akhmedova, S., Semenkin, E., & Stanovov, V. (2018). Co-operation of biology related algorithms

for solving opinion mining problems by using diferent term weighting schemes. In: K. Madani,

D. Peaucelle, O. Gusikhin (Eds.) Informatics in control, automation and robotics: 13th international

conference, ICINCO 2016 Lisbon, Portugal, 29-31 July, 2016 (pp. 73–90). Cham: Springer. https://

doi.org/10.1007/978-3-319-55011-4_4.

5. Allison, B. (2008). Sentiment detection using lexically-based classifers. In P. Sojka, A. Horák, I.

Kopeček, & K. Pala (Eds.), Text, speech and dialogue (pp. 21–28). Berlin: Springer.

6. Balahur, A., Kozareva, Z., & Montoyo, A. (2009). Determining the polarity and source of opinions

expressed in political debates. In A. Gelbukh (Ed.), Computational linguistics and intelligent text

processing (pp. 468–480). Berlin: Springer.

7. Bansal, M., Cardie, C., & Lee, L. (2008). The power of negative thinking: Exploiting label disagree-

ment in the min-cut classifcation framework. In: Coling 2008: Companion volume: Posters (pp.

15–18). Coling 2008 Organizing Committee, Manchester, UK. https://www.aclweb.org/anthology/

C08-2004.

8. Baturo, A., Dasandi, N., & Mikhaylov, S. J. (2017). Understanding state preferences with text as

data: Introducing the un general debate corpus. Research and Politics, 4(2), 2053168017712821.

https://doi.org/10.1177/2053168017712821.

9. Bhatia, S., P, D. (2018). Topic-specifc sentiment analysis can help identify political ideology. In:

Proceedings of the 9th workshop on computational approaches to subjectivity, sentiment and social

media analysis (pp. 79–84). Association for Computational Linguistics, Brussels, Belgium. https://

doi.org/10.18653/v1/W18-6212. https://www.aclweb.org/anthology/W18-6212.

10. Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. Journal of Machine

Learning Research, 3(Jan), 993–1022.

11. Bonica, A. (2016). A data-driven voter guide for US elections: Adapting quantitative measures of

the preferences and priorities of political elites to help voters learn about candidates. Journal of the

Social Sciences, 2(7), 11–32. https://doi.org/10.7758/RSF.2016.2.7.02. https://www.rsfournal.org/

content/2/7/11.

12. Budhwar, A., Kuboi, T., Dekhtyar, A., & Khosmood, F. (2018). Predicting the vote using legis-

lative speech. In: Proceedings of the 19th annual international conference on digital government

research: governance in the data age, dg.o ’18 (pp. 35:1–35:10). ACM, New York, NY, USA. https

://doi.org/10.1145/3209281.3209374.

13. Burfoot, C. (2008). Using multiple sources of agreement information for sentiment classifcation of

political transcripts. In: Proceedings of the Australasian language technology association workshop

2008 (pp. 11–18). Hobart, Australia. https://www.aclweb.org/anthology/U08-1003.

14. Burfoot, C., Bird, S., & Baldwin, T. (2011). Collective classifcation of congressional foor-debate

transcripts. In: Proceedings of the 49th annual meeting of the association for computational linguis-

tics: Human language technologies (pp. 1506–1515). Association for Computational Linguistics,

Portland, Oregon, USA. https://www.aclweb.org/anthology/P11-1151.

15. Burford, C., Bird, S., & Baldwin, T. (2015). Collective document classifcation with implicit inter-

document semantic relationships. In: Proceedings of the fourth joint conference on lexical and com-

putational semantics (pp. 106–116). Association for Computational Linguistics, Denver, Colorado.

https://doi.org/10.18653/v1/S15-1012. https://www.aclweb.org/anthology/S15-1012.

16. Chen, W., Zhang, X., Wang, T., Yang, B., & Li, Y. (2017). Opinion-aware knowledge graph for

political ideology detection. In: Proceedings of the 26th international joint conference on artifcial

intelligence, pp. 3647–3653.

17. Diermeier, D., Godbout, J. F., Yu, B., & Kaufmann, S. (2012). Language and ideology in congress.

British Journal of Political Science, 42(1), 31–55.

18. Duthie, R., & Budzynska, K. (2018). A deep modular rnn approach for ethos mining. In: Proceed-

ings of the twenty-seventh international joint conference on artifcial intelligence, (IJCAI-18), pp.

4041–4047.

19. Dzieciątko, M. (2019). Application of text analytics to analyze emotions in the speeches. In E.

Pietka, P. Badura, J. Kawa, & W. Wieclawek (Eds.), Information Technology in Biomedicine (pp.

525–536). Cham: Springer.

20. Frid-Nielsen, S. S. (2018). Human rights or security? Positions on asylum in european parliament

speeches. European Union Politics, 19(2), 344–362. https://doi.org/10.1177/1465116518755954.

21. Glavaš, G., Nanni, F., & Ponzetto, S.P. (2017). Unsupervised cross-lingual scaling of political texts.

In: Proceedings of the 15th conference of the European chapter of the association for computa-

tional linguistics: Volume 2, short papers (pp. 688–693). Association for Computational Linguis-

tics, Valencia, Spain. https://www.aclweb.org/anthology/E17-2109.

22. Glavaš, G., Nanni, F., & Ponzetto, S.P. (2019). Computational analysis of political texts: Bridging

research eforts across communities. In: Proceedings of the 57th annual meeting of the association

for computational linguistics: Tutorial abstracts (pp. 18–23). Association for Computational Lin-

guistics, Florence, Italy. https://doi.org/10.18653/v1/P19-4004. https://www.aclweb.org/anthology/

P19-4004.

23. Grimmer, J., & Stewart, B. M. (2013). Text as data: The promise and pitfalls of automatic content

analysis methods for political texts. Political Analysis, 21(3), 267–297.

24. Hirst, G., Riabinin, Y., & Graham, J. (2010). Party status as a confound in the automatic classifca-

tion of political speech by ideology. In: Proceedings of 10th international conference on statistical

analysis of textual data/10es Journées internationales d’Analyse statistique des Données Textuelles

(JADT 2010), Rome, pp. 731–742.

25. Honkela, T., Korhonen, J., Lagus, K., & Saarinen, E. (2014). Five-dimensional sentiment analysis

of corpora, documents and words. In T. Villmann, F. M. Schleif, M. Kaden, & M. Lange (Eds.),

Advances in self-organizing maps and learning vector quantization (pp. 209–218). Cham: Springer.

26. Hopkins, D. J., & King, G. (2010). A method of automated nonparametric content analysis for

social science. American Journal of Political Science, 54(1), 229–247. https://doi.org/10.111

1/j.1540-5907.2009.00428.x.

27. Iliev, I. R., Huang, X., & Gel, Y. R. (2019). Political rhetoric through the lens of non-parametric sta-

tistics: Are our legislators that diferent? Journal of the Royal Statistical Society Series A (Statistics

in Society), 182(2), 583–604. https://doi.org/10.1111/rssa.12421.

28. Iyyer, M., Enns, P., Boyd-Graber, J., & Resnik, P. (2014). Political ideology detection using recur-

sive neural networks. In: Proceedings of the 52nd annual meeting of the association for computa-

tional linguistics (Volume 1: Long Papers) (pp. 1113–1122). Association for Computational Lin-

guistics, Baltimore, Maryland. https://doi.org/10.3115/v1/P14-1105. https://www.aclweb.org/antho

logy/P14-1105

29. Jensen, J., Naidu, S., Kaplan, E., Wilse-Samson, L., Gergen, D., Zuckerman, M., & Spirling, A.

(2012). Political polarization and the dynamics of political language: Evidence from 130 years of

partisan speech [with comments and discussion]. Brookings Papers on Economic Activity, pp. 1–81.

30. Ji, Y., & Smith, N.A. (2017) Neural discourse structure for text categorization. In: Proceedings of

the 55th annual meeting of the association for computational linguistics (Volume 1: Long Papers)

(pp. 996–1005). Association for Computational Linguistics, Vancouver, Canada. https://doi.

org/10.18653/v1/P17-1092. https://www.aclweb.org/anthology/P17-1092.

31. Kaal, B., Maks, I., & van Elfrinkhof, A. (2014). From text to political positions: Text analysis across

disciplines (Vol. 55). Philadelphia: John Benjamins Publishing Company.

32. Kapočiūtė-Dzikienė, J., & Krupavičius, A. (2014). Predicting party group from the Lithuanian par-

liamentary speeches. Information Technology and Control, 43(3), 321–332.

33. Kaufman, D., Khosmood, F., Kuboi, T., & Dekhtyar, A. (2018). Learning alignments from legisla-

tive discourse. In: Proceedings of the 19th annual international conference on digital government

research: Governance in the data age, dg.o ’18 (pp. 119:1–119:2). ACM, New York, NY, USA.

https://doi.org/10.1145/3209281.3209413.

34. Kim, I. S., Londregan, J., & Ratkovic, M. (2018). Estimating spatial preferences from votes and text.

Political Analysis, 26(2), 210–229.

35. Lapponi, E., Søyland, M. G., Velldal, E., & Oepen, S. (2018). The talk of norway: A richly anno-

tated corpus of the norwegian parliament, 1998–2016. Language Resources and Evaluation, 52(3),

873–893. https://doi.org/10.1007/s10579-018-9411-5.

36. Laver, M., Benoit, K., & Garry, J. (2003). Extracting policy positions from political texts using

words as data. American Political Science Review, 97(2), 311–331.

37. Lefait, G., & Kechadi, T. (2010). Analysis of deputy and party similarities through hierarchical

clustering. In: 2010 fourth international conference on digital society (pp. 264–268). https://doi.

org/10.1109/ICDS.2010.49.

38. Li, X., Chen, W., Wang, T., & Huang, W. (2017). Target-specifc convolutional bi-directional lstm

neural network for political ideology analysis. In L. Chen, C. S. Jensen, C. Shahabi, X. Yang, & X.

Lian (Eds.), Web and Big Data (pp. 64–72). Cham: Springer.

39. Liu, B. (2012). Sentiment analysis and opinion mining, synthesis lectures on human language tech-

nologies (Vol. 5). San Rafael: Morgan & Claypool Publishers.

40. Lowe, W., & Benoit, K. (2013). Validating estimates of latent traits from textual data using human

judgment as a benchmark. Political Analysis, 21(3), 298–313.

41. Martineau, J., Finin, T., Joshi, A., & Patel, S. (2009). Improving binary classifcation on text prob-

lems using diferential word features. In: Proceedings of the 18th ACM conference on information

and knowledge management, CIKM ’09 (pp. 2019–2024). ACM, New York, NY, USA. https://doi.

org/10.1145/1645953.1646291.

42. Menini, S., Nanni, F., Ponzetto, S.P., & Tonelli, S. (2017). Topic-based agreement and disagreement

in US electoral manifestos. In: Proceedings of the 2017 conference on empirical methods in natu-

ral language processing (pp. 2938–2944). Association for Computational Linguistics, Copenhagen,

Denmark. https://doi.org/10.18653/v1/D17-1318. https://www.aclweb.org/anthology/D17-1318.

43. Menini, S., & Tonelli, S. (2016). Agreement and disagreement: Comparison of points of view in the

political domain. In: Proceedings of COLING 2016, the 26th international conference on computa-

tional linguistics: Technical papers (pp. 2461–2470). The COLING 2016 Organizing Committee,

Osaka, Japan. https://www.aclweb.org/anthology/C16-1232.

44. Mikhaylov, S., Laver, M., & Benoit, K. (2008). Coder reliability and misclassifcation in compara-

tive manifesto project codings. In: 66th MPSA annual national conference.

45. Mohammad, S. M., Sobhani, P., & Kiritchenko, S. (2017). Stance and sentiment in tweets. ACM

Transactions on Internet Technology, 17(3), 26:1–26:23. https://doi.org/10.1145/3003433.

46. Moher, D., Liberati, A., Tetzlaf, J., & Altman, D. G. (2009). The PRISMA group: Preferred report-

ing items for systematic reviews and meta-analyses: The PRISMA statement. Annals of Internal

Medicine, 151(4), 264–269. https://doi.org/10.7326/0003-4819-151-4-200908180-00135.

47. Monroe, B. L., Colaresi, M. P., & Quinn, K. M. (2008). Fightin’words: Lexical feature selection and

evaluation for identifying the content of political confict. Political Analysis, 16(4), 372–403.

48. Naderi, N., & Hirst, G. (2016). Argumentation mining in parliamentary discourse. In M. Baldoni,

C. Baroglio, F. Bex, F. Grasso, N. Green, M. R. Namazi-Rad, M. Numao, & M. T. Suarez (Eds.),

Principles and practice of multi-agent systems (pp. 16–25). Cham: Springer.

49. Nanni, F., Zirn, C., Glavaš, G., Eichorst, J., & Ponzetto, S.P. (2016) Topfsh: topic-based analysis of

political position in us electoral campaigns. In: PolText 2016: The international conference on the

advances in computational analysis of political text: proceedings of the conference.

50. Nguyen, V.A., Boyd-Graber, J., Resnik, P., & Miler, K. (2015). Tea party in the house: A hierarchi-

cal ideal point topic model and its application to republican legislators in the 112th congress. In:

Proceedings of the 53rd annual meeting of the association for computational linguistics and the

7th international joint conference on natural language processing (Volume 1: Long papers) (pp.

1438–1448). Association for Computational Linguistics, Beijing, China. https://doi.org/10.3115/v1/

P15-1139. https://www.aclweb.org/anthology/P15-1139.

51. Nguyen, V. A., Ying, J. L., & Resnik, P. (2013). Lexical and hierarchical topic regression. In C. J.

C. Burges, L. Bottou, M. Welling, Z. Ghahramani, & K. Q. Weinberger (Eds.), Advances in neural

information processing systems 26 (pp. 1106–1114). Curran Associates Inc. http://papers.nips.cc/

paper/5163-lexical-and-hierarchical-topic-regression.pdf.

52. Onyimadu, O., Nakata, K., Wilson, T., Macken, D., & Liu, K. (2014). Towards sentiment analysis

on parliamentary debates in hansard. In W. Kim, Y. Ding, & H. G. Kim (Eds.), Semantic technology

(pp. 48–50). Cham: Springer.

53. Owen, E. (2017). Exposure to ofshoring and the politics of trade liberalization: Debate and votes on

free trade agreements in the US house of representatives, 2001–2006. International Studies Quar-

terly, 61(2), 297–311. https://doi.org/10.1093/isq/sqx020.

54. Pang, B., & Lee, L. (2008). Opinion mining and sentiment analysis. Foundations and Trends® in

Information Retrieval, 2(1–2), 1–135. https://doi.org/10.1561/1500000011.

55. Plantié, M., Roche, M., Dray, G., & Poncelet, P. (2008). Is a voting approach accurate for opinion

mining? In I. Y. Song, J. Eder, & T. M. Nguyen (Eds.), Data warehousing and knowledge discovery

(pp. 413–422). Berlin: Springer.

56. Proksch, S. O., Lowe, W., Wäckerle, J., & Soroka, S. (2019). Multilingual sentiment analysis: A

new approach to measuring confict in legislative speeches. Legislative Studies Quarterly, 44(1),

97–131. https://doi.org/10.1111/lsq.12218.

57. Proksch, S. O., & Slapin, J. B. (2010). Position taking in European parliament speeches. British

Journal of Political Science, 40(3), 587–611.

58. Proksch, S. O., & Slapin, J. B. (2015). The politics of parliamentary debate. Cambridge: Cambridge

University Press.

59. Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A comprehensive grammar of the eng-

lish language. London: Longman.

60. Rauh, C. (2018). Validating a sentiment dictionary for german political language—a workbench

note. Journal of Information Technology and Politics, 15(4), 319–343. https://doi.org/10.1080/19331

681.2018.1485608.

61. Rheault, L. (2016) Expressions of anxiety in political texts. In Proceedings of the frst workshop on

nlp and computational social science (pp. 92–101). Association for Computational Linguistics, Aus-

tin, Texas. https://doi.org/10.18653/v1/W16-5612. https://www.aclweb.org/anthology/W16-5612.

62. Rheault, L., Beelen, K., Cochrane, C., & Hirst, G. (2016). Measuring emotion in parliamentary

debates with automated textual analysis. PLoS One, 11(12), 1–18. https://doi.org/10.1371/journ

al.pone.0168843.

63. Richards, L. (2005). Handling qualitative data: A practical guide. London: Sage Publications.

64. Rudkowsky, E., Haselmayer, M., Wastian, M., Jenny, M., Emrich, Š., & Sedlmair, M. (2018). More

than bags of words: Sentiment analysis with word embeddings. Communication Methods and Meas-

ures, 12(2–3), 140–157. https://doi.org/10.1080/19312458.2018.1455817.

65. Sakamoto, T., & Takikawa, H. (2017). Cross-national measurement of polarization in political dis-

course: Analyzing foor debate in the US the japanese legislatures. In 2017 IEEE international con-

ference on big data (Big Data) (pp. 3104–3110). https://doi.org/10.1109/BigData.2017.8258285.

66. Salah, Z. (2014). Machine learning and sentiment analysis approaches for the analysis of parlia-

mentary debates. Ph.D. thesis, University of Liverpool.

67. Salah, Z., Coenen, F., & Grossi, D. (2013). Extracting debate graphs from parliamentary transcripts:

A study directed at uk house of commons debates. In Proceedings of the fourteenth international

conference on artifcial intelligence and law, ICAIL ’13 (pp. 121–130). ACM, New York, NY, USA.

https://doi.org/10.1145/2514601.2514615.

68. Salah, Z., Coenen, F., & Grossi, D. (2013). Generating domain-specifc sentiment lexicons for opin-

ion mining. In H. Motoda, Z. Wu, L. Cao, O. Zaiane, M. Yao, & W. Wang (Eds.), Advanced data

mining and applications (pp. 13–24). Berlin: Springer.

69. Schwarz, D., Traber, D., & Benoit, K. (2017). Estimating intra-party preferences: Comparing

speeches to votes. Political Science Research and Methods, 5(2), 379–396.

70. Seligman, M. E. P. (2012). Flourish: A visionary new understanding of happiness and well-being.

New York: Simon and Schuster.

71. Sim, Y., Acree, B.D.L., Gross, J.H., & Smith, N.A. (2013). Measuring ideological proportions in

political speeches. In Proceedings of the 2013 conference on empirical methods in natural language

processing (pp. 91–101). Association for Computational Linguistics, Seattle, Washington, USA.

https://www.aclweb.org/anthology/D13-1010.

72. Sokolova, M., & Lapalme, G. (2008). Verbs speak loud: Verb categories in learning polarity and strength

of opinions. In S. Bergler (Ed.), Advances in artifcial intelligence (pp. 320–331). Berlin: Springer.

73. Taddy, M. (2013). Multinomial inverse regression for text analysis. Journal of the American Statisti-

cal Association, 108(503), 755–770.

74. Thomas, M., Pang, B., & Lee, L. (2006). Get out the vote: Determining support or opposition from

congressional foor-debate transcripts. In Proceedings of the 2006 conference on empirical methods

in natural language processing (pp. 327–335). Association for Computational Linguistics, Sydney,

Australia. https://www.aclweb.org/anthology/W06-1639.

75. van der Zwaan, J.M., Marx, M., & Kamps, J. (2016). Validating cross-perspective topic modeling

for extracting political parties’ positions from parliamentary proceedings. In Proceedings of the

twenty-second European conference on artifcial intelligence, ECAI’16 (pp. 28–36). IOS Press,

Amsterdam, The Netherlands, The Netherlands. https://doi.org/10.3233/978-1-61499-672-9-28.

76. Vilares, D., & He, Y. (2017). Detecting perspectives in political debates. In Proceedings of the 2017

conference on empirical methods in natural language processing (pp. 1573–1582). Association for

Computational Linguistics, Copenhagen, Denmark. https://doi.org/10.18653/v1/D17-1165. https://

www.aclweb.org/anthology/D17-1165.

77. Yadollahi, A., Shahraki, A. G., & Zaiane, O. R. (2017). Current state of text sentiment analy-

sis from opinion to emotion mining. ACM Computing Surveys, 50(2), 25:1–25:33. https://doi.

org/10.1145/3057270.

78. Yessenalina, A., Yue, Y., & Cardie, C. (2010). Multi-level structured models for document-level

sentiment classifcation. In Proceedings of the 2010 conference on empirical methods in natural

language processing (pp. 1046–1056). Association for Computational Linguistics, Cambridge, MA.

https://www.aclweb.org/anthology/D10-1102.

79. Yogatama, D., Kong, L., & Smith, N.A. (2015). Bayesian optimization of text representations. In

Proceedings of the 2015 conference on empirical methods in natural language processing (pp.

2100–2105). Association for Computational Linguistics, Lisbon, Portugal. https://doi.org/10.18653/

v1/D15-1251. https://www.aclweb.org/anthology/D15-1251.

80. Yogatama, D., & Smith, N. (2014). Making the most of bag of words: Sentence regularization with

alternating direction method of multipliers. In International conference on machine learning, pp.

656–664.

81. Yogatama, D., & Smith, N.A. (2014). Linguistic structured sparsity in text categorization. In Pro-

ceedings of the 52nd annual meeting of the association for computational linguistics (Volume 1:

Long Papers) (pp. 786–796). Association for Computational Linguistics, Baltimore, Maryland. https

://doi.org/10.3115/v1/P14-1074. https://www.aclweb.org/anthology/P14-1074.

https://link.springer.com/article/10.1007%2Fs42001-019-00060-w