• Home
• Newsletters
• People of ACM
• People of ACM - Nuria Oliver

People of ACM - Nuria Oliver

July 24, 2018

You have spent a large portion of your career overseeing data science initiatives at large telecommunications companies. How can data science improve telecommunications and what is an exciting area of research you are working on now in this area?

Data is a core asset for telecommunications companies. Large-scale datasets of diverse natures are generated as a result of the operation of the telecommunication networks (both mobile and fixed). Data is also generated by the provision of services, interactions with customers via contact centers, chatbots and web services, and billing processes. All this doesn’t include the Internet of Things infrastructure and services, where the expectation is that there will be over 20 billion connected devices by 2020.

This data is held securely in state-of-the-art secured data platforms; any data that we use is always with the explicit permission of customers and it is always in nonymized/pseudonymized form. All this data is not only large in terms of scale (petabytes to exabytes); it is also non-structured, diverse and generated at high speeds. Therefore, we (the telecommunications providers) need to rely on sophisticated machine learning algorithms in order to make sense of it and be able to personalize services; understand, anticipate and respond to customer needs; optimize our networks; create new data-driven services; and help make better decisions that have positive social impact. I am particularly passionate about this last topic.

In terms of an exciting area of research, we have found that telco data is valuable in many important areas. These include public health (such as for infectious disease modeling and pandemic management); transportation (such as traffic estimation and prediction); natural disasters (such as rapid assessment of how many people have been affected by a natural disaster and population displacements); and financial inclusion (such as automatic inference of socioeconomic status). Vodafone is working on all these topics in countries that have both developed and developing economies.

What is the most significant way that data science has impacted human behavior modeling and prediction?

We can model human behavior both at an individual and at an aggregate level. At an individual level, the vast majority of services that we use today include some level of personalization (that is, adaptation to our taste, needs, traits, etc.), which is typically done through the analysis and modeling of past usage data via machine learning techniques. On this front, two areas that I am particularly passionate about are personalized medicine/wellbeing and personalized education.

At an aggregate level, we’re living in an unprecedented moment in our history with the availability of large-scale human behavioral data. For the first time since we have existed as a species, we can use this data to model and quantify large-scale human behavior, which enables us to validate existing social science theories, propose new ones, and make better decisions in areas such as resource allocation, public health, and transportation. I have been working for the past eight years on a topic that we call “Big Data and AI for Social Good,” where the goal is to leverage large-scale, anonymized data to help us make better decisions in important use cases such as public health, natural disasters or socioeconomic development.

You have done research in both intelligent user interfaces and mobile computing. How do you think interfaces on mobile devices will change in the coming years?

An existing challenge in mobile devices is their small screen, which forces us to look down and makes it difficult to both enter and read information. However, significant progress in speech recognition is enabling us to talk to our mobile devices for the first time, which is a more natural way of interaction. I would anticipate a distribution of different sensors, actuators and displays throughout the body such that the mobile phone is still the main computing device, but with distributed capabilities for input and output.

I also hope that human-to-human communication through our mobile phones will be more multi-sensorial beyond text, voice and video, such that we would feel the person that we are talking to next to us. Finally, our smartphones are starting to deserve their name, given the increased capabilities of artificial intelligence—they have the potential of becoming useful companions and personal assistants with the ability to process contextual information, make sense of it and make intelligent decisions that will help us in our daily lives.

One of Data Pop Alliance’s major initiatives is to use data science to understand and prevent crime in Latin America and the Caribbean. Will you tell us a little more about this effort?

We found that aggregate urban dynamics are an effective way to automatically identify crime hotspots in cities. We found empirical evidence with data from the London metropolitan area to support Jane Jacobs’ theories of urban diversity. Follow-up work in Colombia also empirically found that the structural characteristics of a city–namely Jane Jacobs’ diversity conditions—are a better predictor of the number of homicides and robberies in Bogotá than socioeconomic indicators, such as unemployment and deprivation. Mobility information–related to the routine activity theory—improved the prediction of the model by 15%. The combination of structural and socioeconomic variables provided better predictions than if they were considered individually, in both the London and Bogotá projects. While the findings are limited to these cities and the specific time period of analysis, we are excited about the value they might bring to help define crime control and prevention action plans in cities.

You have been a champion for increasing the role of women in computing. What advice would you offer a younger female colleague who is deciding between a career in computing and other options?

Above all, my recommendation is to study what she would feel passionate about. The ideal situation is when your passion is your job. My second recommendation is to get as much information as possible about different career options, ideally by talking with professionals from different fields. In addition, she could use online services—like Vodafone’s Future Jobs Finder—that help young people identify which roles match their skills and preferences.

Computer science is one of the best career choices for many reasons: you work at the forefront of technology; you have the opportunity to invent the future and have positive impact on the lives of millions of people; you have the opportunity to work in almost any field (as computer science skills are needed across most disciplines); you have a lot of flexibility in terms of working hours; usually you are able to work from home as needed, which is wonderful to help find a sustainable work/life balance; you are always learning, as computer science is a very active field; and you have many professional opportunities with high salaries, given that there aren’t enough computer science experts (especially in areas such as artificial intelligence, data science, and security). It is a very rewarding, exciting and stimulating career choice.