Agents of Change

Tong Sun: a history in analysis

"My dream was to become an engineer so I could help make our country better and stronger - I was really influenced by the times."

A history in analysis

Tong Sun was a child of China’s Cultural Revolution — born during the notorious period of state-sponsored civil unrest in the 1960s and 1970s that saw millions of lives turned upside down, schools shut, and the study of science and engineering largely abandoned.

Despite this disruption to her early life, Tong has gone on to become a pioneer in computer science and a renowned technology innovator — recently founding Xerox’s Scalable Data Analytics Research Lab at PARC, a Xerox company.

Her work in the fields of big-data computing, machine learning, social network analysis and artificial intelligence during more than two decades at Xerox has led to the delivery of highly successful technologies to market. Now her lab’s data analytics work in the healthcare sector is positively influencing the health and quality of life of many thousands of people.

The story of Tong Sun’s early education and subsequent progression through the fields of science and engineering can appear as paradoxical as it is impressive. Tong was 10 years old when China’s Cultural Revolution — a period of chaos fermented by the then-Communist Party president Mao Zedong — finally came to an end.

Tong was among a generation of Chinese students that missed out on much of their early education. It was a generation also starved of scientific and technological inspiration due to the dismantling of the country’s further education system and science capabilities.

Conversely, across the Pacific, Tong’s future science and engineering peers in the United States were enjoying a more settled education against a backdrop of scientific and technological hope. Medical breakthroughs, the emergence of computers, and successful missions to the moon had highlighted the potential of science and technology to cure disease, free humans from mundane work, and explore new frontiers.

Despite the apparent disadvantages of her early education, Tong is convinced that growing up in the period known as China’s Lost Decade was pivotal to her emergence as an innovator.

 

A new hope

“It was a terrible time for the country, but the struggles we faced were nothing compared to what our parents’ generation went through,” she recalls. “And we all benefitted from a nationwide campaign in the early 1980s to promote science and technology.” The program helped renew interest in engineering and science among school-aged children.

“When I look back, it was harsh for young kids to have to focus so hard on academia alone, but the long hours and discipline provided a good foundation for science. My dream was to become an engineer so I could help make our country better and stronger — I was really influenced by the times.”

 

Working at Xerox

By the middle of the 1980s, science had become the great hope for China’s future. Yet over in the United States, technological and scientific optimism seemed to be waning. The war on cancer wasn’t being won and space exploration had stalled. Meanwhile, the world seemed powerless in the face of the emerging threat of global warming.

Yet a quiet revolution was starting to take root in Silicon Valley. Nearby at PARC, where Tong would one day work, a team of researchers was addressing the challenges of early personal computing, eventually redefining the entire relationship of humans, work and technology for the post-PC era.

But for the time being, Tong — who was studying electrical engineering in college — was unaware even of the existence of Silicon Valley.

“We didn’t know anything really about the outside world,” she says. “But Apple II and IBM PCs were available at my college, and they really fascinated me.” Tong soon realized she enjoyed writing software more than looking at circuits. “That was a turning point for me,” she says. “I went on to study logic programming and symbolic reasoning, a subfield of artificial intelligence, for my master’s degree.”

Tong recalls how the artificial intelligence (AI) field back then focused on building computing systems that were hand-coded by humans. Unlike today, the idea of machines performing cognitive tasks was impractical. “Like many others, we knew about Moore’s Law [which states that processing power would double every two years], but of course we could not fully comprehend the implications of that pace of change.” This change, explains Tong, includes incredible advances in computing and Big Data that have pushed AI into multiple directions — directly impacting businesses, and the health and wellbeing of people.

As reform began to take hold in communist China during the late 1980s, the government began easing regulations and protectionist policies. Travel to and from China became possible for a number of citizens. Tong was one of them.

“When the country started opening up, I had the opportunity to come to the United States to do a Ph.D. This was in the early 90s. At the time, AI research had been put on a back-burner by the scientific community, so I focused my research on parallel supercomputing.”

Tong joined Xerox in 1995 after finishing her Ph.D. “The world-wide web had just started to become available, and Xerox was looking to improve high performance printing and to build intelligent document solutions.” Tong and her colleagues began to realize how the internet would disrupt business and people’s lives.

 

The social media era

By 2005, social media was becoming a real point of interest for Tong. “I was fascinated by a website called Digg.com, which was a social forum for sharing news and political opinions. I have to say this had no direct commercial relevance at the time as Xerox still had a big focus on printing technologies. But this is one of the things that’s great about being part of Xerox research. It provides the freedom for researchers like me to explore emerging technologies and think about things ahead of time.”

“Before Facebook and Twitter emerged, my group started developing state-of-the-art natural-language processing and social-graph mining technologies by examining social interactions on Diggs.com,” Tong recalls. “We found that when people discussed the political news it usually became a dynamic debate that included both opinions and sentiment. We started asking ourselves, ‘Hey, we want to see if there’s a dynamic in this political social forum. Who influences who? What are they debating about? What is the sentiment and how does the sentiment change as the debate progresses?’ Who is really driving the conversation?

“However, unlike conventional text documents, social media content is short, noisy and often filled with new acronyms, sarcasm and emojicons. These impose great challenges for the traditional linguistic-centric methods where rigid rules and statistical models are required. So this is where artificial intelligence really started to help. We used machine learning, which is a form of artificial intelligence, to begin solving these problems. This technique involves humans teaching a computer to discern sentiment, sarcasm and shifting topics.

“As Twitter and Facebook became popular, the volume of user-generated content grew tremendously. So we started sampling and annotating social media conversations to build a large dataset that could be used to train computers. Our sentiment analysis algorithms can use this training data set (in addition to the linguistic structures and rules) to achieve much better accuracy."

 

Customer care

Tong explains: “A few years ago, while monitoring Twitter one morning, we picked up a lot of negative sentiments about a telecom company’s phone signals in the Chicago area. Later that evening, there were news reports about weak phone signals caused by the company upgrading their cell towers in Chicago. We had spotted the problem as it was occurring, and realized that with the real-time knowledge we detected from social media, the company could have resolved the problem more quickly. They could have proactively informed their customers about this situation, updated the support website, or increased the number of customer care agents to handle this incident more effectively.

“We took the lesson and helped turn it into an offering for Xerox customers. Today, Xerox’s customer-care business, provides sentiment analysis over social channels so that our clients can more quickly serve their customers.”

Since then, Tong and her team have moved beyond just listening to everyday chatter and sentiment on social media. They’ve developed novel methods of extracting deeper insights on people’s interests, behaviors and lifestyles based on online activities (e.g. what post they liked, what content they posted, and when they posted), social ties (e.g. who they interacted with, who they followed) and community structures (e.g. what kind of groups they belong to). These insights can be aggregated and correlated with other data sets, such as geo-locations, weather data, traffic flow data, and medical claims. The potential use cases are considerable and constantly growing.

 

Health interventions

“For example, in the healthcare sector, we can use medical claims to define an at-risk population for Type 2 diabetes in a certain geo-region. We use social media combined with public survey and census data to add a layer of non-medical factors that can also affect this population. For example, we may detect underlying behaviors such as the amount of physical activity pursued, smoking, alcohol usage, diet and nutrition intakes. With this type of information, Xerox helps healthcare insurance payers, hospitals and State Medicaid agencies design and deploy targeted intervention programs.”

Tong and her team have found that community wellness campaigns, smoking cessation programs and care coordination programs can impact health outcomes across entire populations — as long as they are tailored to the specific needs of a population.

According to Tong, Xerox’s work in this area can create a “ripple effect” across healthcare systems. “Combining social behavior factors with patient electronic health records provides insights that doctors and nurses can use to help individual patients make better choices. Insurance providers also can make better underwriting decisions by more deeply understanding the populations they serve. It’s a huge business and really important for our nation’s healthcare system.”

What started in the early part of the last decade as some early research into social media by Tong and her colleagues has developed into an incredibly valuable service for society. Tong explains: “What makes me excited about Xerox is the paradigm shifting, and the new opportunities that follow. From my point of view, I’m constantly learning new things, and technology will keep shifting so I will keep learning.

“Computers through the advent of artificial intelligence are enabling people to do their jobs better. Instead of robots replacing humans, I see a future that is based on humans collaborating with computers to gain better outcomes.”

 

Making sacrifices

Tong’s career journey — from her disrupted elementary education in Chairman Mao’s China, to setting up Xerox’s Scalable Data Analytics Lab — has required plenty of sacrifice: “When I look back, we worked really hard,” she recalls. “It was far too much for young children.”

In fact, Tong — who has a 19-year-old son, a freshman at college, and a 16-year-old daughter — would offer cautionary encouragement to those who want to follow the ever-increasing popularity in artificial intelligence. “To succeed in the field of machine learning or artificial intelligence, which is still dominated by males, you need very robust mathematical understanding, analytical thinking and advanced programming skills. And this can be challenging to develop without a focus on science and engineering at the middle and high-school level. We have to tell young people this. I wouldn’t push my kids into artificial intelligence. But if young people are prepared for the hard work, I’d tell them it’s so exciting to be involved in.”

 

We’ve all changed the world. Every one of us. With every breath we take, our presence endlessly ripples outwards.

But few of us have the opportunity to change many lives for the better. And even fewer are challenged to do so every day. That’s the gauntlet thrown daily at Xerox research scientists — to try and effect change.

In return, we give them time and space to dream. And then the resources to turn dreams into reality — whether they’re inventing new materials with incredible functions, or using augmented reality to bolster the memory of Alzheimer’s patients.

We’re proud of our Agents of Change in Xerox research centers across the world.
Here are some of their stories.