Section x.1
Teaching network science

The perspective offered by networks is indispensable for those who wish to understand today’s interlinked world. This textbook is the best avenue I found to share this perspective, offering anyone the opportunity to become a bit of a network scientist. Many of the choices I made in selecting the topics and in presenting the material were guided by the desire to offer a quantitative yet easy to follow introduction to the field. At the same time I tried to pass on the many insights networks offer about the many complex systems that surround us. To resolve these often conflicting desires, I paired the technical advances with historical notes and boxes that illustrate the roots and the applications of the key discoveries.

This preface has two puposes. On one end, by describing the class that motivated this text, it offers some practical tips on how to best use the textbook. Equally important, it acknowledges the long list of individuals who helped move forward this textbook.

Online compedium

Network science is rich in content and knowledge that is best apprecitated online. Therefore throughout the chapters we encounter numerous ONLINE RESOURCES that point to pertinent online material - videos, software, interactive tools, datasets and data sources. These resources are available on the http://barabasi.com/NetworkScienceBook website.

The website also contains the PowerPoint slides that I used to teach network science, mirroring the content of this textbook. Anyone teaching networks should feel free to use these slides and modify them as they see it fit to offer the best classroom experience. There is no need to ask the author for permission to use these slides in educational settings.

Given the empirical roots of network science, the book has a strong emphasis on the analysis of real networks. We have therefore assembled ten network maps that are frequently used in the literature to test various network characteristics. They were chosen to represent the diversity of the networks explored in network science, describing social, biological, technological and informational systems. The Online Compendium offers access to these datasets, which are used throughout the book to illustrate the tools of network science.

Finally for those teaching the book in different languages, the website also mirrors the ongoing translation projects.

Network science class

I have taught network science in two different settings. The first is a full semester class that attracts graduate and advanced undergraduate students with physics, computer science and engineering background. The second is a three-week two-credit class for students with economics and social science background. The textbook builds on both teaching experiences: In the full semester class I cover the full text, integrating into the lectures the proofs and derivations contained in the Advanced Topics. In the shorter class I only cover the content of the main sections, omitting the Advanced Topics and the chapter on degree correlations.

In both settings a key component of the class are assignments and the research project described next.

For the longer class we assign as homework a subset of the problems listed at the end of each chapter, testing the technical proficiency of the students with the material and their problem solving ability. Two rounds of homework cover the material as we progress with the class.

We ask each student to select a concept or a term related to network science and write a Wikipedia page on it (Image x.1). What makes this assignment somewhat challenging is that the topic must not be already covered by Wikipedia, yet must be sufficiently notable to be covered. The Wiki assignment tests the students' ability to synthetize and distill material in an easy-to-understand encyclopedic style, potentially turning them into regular Wikipedia contributors. At the same time the assignment enriches Wikipedia with network science content, offering a service to the whole community. Those teaching network science in other languages should consider contributing to Wikipedia in their native language.

As a warmup to network analysis, students are asked to analyze the social network of the class. This requires a bit of preparation and the help of a teaching assistant. In the very first class the instructor hands out the class list and asks everyone to check that they are on the list or add their name if they are missing. The teaching assistant takes the final list, and during the class prints an accurate class list for each student. At the end of the class each student is asked to mark everyone they knew before coming to the class. To help students match the faces with the names, each student is asked to briefly introduce themselves - also offering a chance for the instructor to learn more about the students in the class. These lists are then compiled to generate a social network of the class, enriching the nodes with gender and the name of the program the students are engaged in. The anonymized version of this network is returned to the class halfway through the course, the assignment being to analyze its properties using the network science tools the students acquired up to that point. This allows them to explore a relatively small network that they are invested in and understand. The assignment offers a preparation for the more extensive network analysis they will perform for their final research project. This homework is assigned after the hands-on class on software, so that the students are already familiar with the online tools available for network analysis.

The final project is the most rewarding part of the class, offering the students the opportunity to combine and utilize all the knowledge they acquired. Students are asked to select a network of interest to them, map it out and analyze it. Some procedural details enrich this assignment:

1. The project is carried out in pairs. If the class composition allows, the students are asked to form professionally heterogenous pairs: undergraduate students are asked to pair up with graduate students, or students from different programs are asked to work together, like a physics student with a biology student. This forces the students to collaborate outside their expertise level and comfort zone, a common ingredient of interdisciplinary research. The instructor does not do the pairing, but students are encouraged to find their partners.
2. A few weeks into the course one class is devoted to preliminary project presentations. Each group is asked to offer a five minute presentation with no more than five slides, offering a preview of the dataset they selected (Image x.2). Students are advised to collect their own data - simply downloading a dataset already prepared for network analysis is not acceptable. Indeed, one of the goals of the project is to experience the choices and compromises one must make in network mapping. Manual mapping is allowed, like looking up the ingredients of recipes in a cookbook or the interaction of characters in a novel or a historical text. Digital mapping is encouraged, like scrapping data from a website or a database that is not explicitly organized as a network map, but the students must reinterpret and clean the data to make it amenable for network analysis. For example one can systematically scrap data from Wikipedia to identify relationships between writers, scientists or concepts.
3. It is important to always emphasize that the purpose of the final project is to test a student's ability to analyze a network. Consequently students must stay focused on exploring the network aspect of the data, and avoid being carried away by other tempting questions their dataset poses that would take them away from this goal.
4. The course ends with the final project presentations. Depending on the size of the class, we devote one or two classes to this (Image x.3).

We devote one class to various network analysis and visualization software, like Gephi, Cytoscape, or NetworkX. In the longer class we devote another one to other numerical procedures, like fitting, log-binning or network visualization. We ask students to bring their laptops to these classes, so that they can try out these tools immediately.

We devote one night, typically outside the class time, to a movie night, where we screen the documentary Connected by Annamaria Talas. The one-hour documentary features many contributors to network science, and offers a compelling narrative of the field's importance. Movie Night is advertised university wide, offering a chance to reach out to a wider community.

In the full semester class we invite researchers from the area to give research seminars about their work pertaining to networks. This offers the students a sense of what cutting edge research looks like in this area. This is typically (but not always) done towards the end of the class, by which point most theoretical tools are covered and the students are focusing on their final project. Such talks, advertised and open to the local research community, often inspire additional perspectives and ideas for the final project.

To aid the planning of the class, Image x.4 offers the schedule of the full semester class I co-taught before this book went to print.

Section x.2
Acknowledgements

Writing a book, any book, is an exercise in lonely endurance. This project was no different, dominating all my free time between 2011 and 2015. It was mostly time spent alone, working in one of the many coffeehouses I frequent in Boston and Budapest, or wherever in the world the morning found me. Despite this the book is far from being a lonely achievement: During these four years a number of individuals have donated their time and expertise to help move forward the project, offering me the opportunity to discuss the subject with colleagues, friends and lab members. I also shared the chapters on the internet for everyone to use, receiving valuable feedback from many individuals. In this section I wish to acknowledge the professional network that stepped in to help at various stages of this long journey.

Formulas, graphs, simulations

A textbook must ensure that everything works as promised. That one can derive the key formulas, and that the measures described in the text, when applied to real data, work as the theory predicts. There is only one way to achieve this: One must check and repeat each calculation, measurement and simulation. This was a heroic job, most of it done by Márton Pósfai, who joined the project when he was a visiting student in my lab in Boston and stayed with it throughout his PhD work in Budapest, Hungary. He checked all derivations, if needed helped re-derive key formulas, performed all the simulations and measurements and prepared the book’s figures and ta- bles. Many figures and tables amounted to small research projects, their outcome forcing us to de-emphasize some quantities because they did not work as promised, or helped us appreciate and understand the importance of others. His deep understanding of the network science literature and his careful work offered many subtle insights that enriched the book. There is no way I could have achieved this depth and reliability if it wasn’t for Márton’s tireless dedication to the project.

The design

The ambition to create a book that had a clear aesthetic and visual appeal was planted by Mauro Martino, a data visualization expert in my lab. He created the first face of the chapters and many visual elements designed by him stayed with us until the end. After Mauro moved on to lead a team of designers at IBM Research, Gabriele Musella took over the design. He standardized the color palette and designed the basic elements of the info-graphics appearing throughout the book, also redrawing most images. He worked with us until the fall of 2014, when he too had to return to London to take up his dream job. At that time the design was taken over by Nicole Samay, who tirelessly and gently retouched the whole book as we neared the finish line. The website for the book was designed by Kim Albrecht, who currently collaborates with Mauro to design the online experience that trails the book.

An important component of the visual design are the images included at the beginning of each chapter, illustrating the interplay between networks and art. In selecting these images I have benefited from advice and discussions with several artists and designers, academics and practicing artists alike. Many thanks to Isabel Meirelles and Dietmar Offenhuber from the Art and Design Department at Northeastern, Mathew Ritchie from Columbia University, and Meredith Tromble from the San Francisco Art Institute, for helping me navigate the boundaries of art, data and network science.

The daily drill: typing, editing

I remain an old-fashioned writer, who writes with a pencil rather than a computer. I am lost, therefore, without editors and typers, who integrate my hand-written notes, corrections and recommendations into each chapter. Sabrina Rabello and Galen Wilkerson have helped get this project started. Yet, the bulk of editing fell on the shoulders of three individuals. Payam Parsinejad worked with me during first year of the project. After he had to refocus on his research, Amal Al-Husseini, a former student from my network science class, joined us, and stayed until the very end. Equally defining was the help of Sarah Morrison, my former assistant, who joined the project after she moved to Lucca, Italy. Her timely and accurate editing were essential to finish the book.

Each chapter, before it was released on our webpage, has undergone a final check by Phillipp Hoevel, who joined the project while visiting my lab, and continued to work with us even after he returned to Berlin to run his own lab. Philipp methodically reviewed everything, from the science to notations, becoming our first reader and final filter.

Brett Common has worked tirelessly to secure all the permissions for the visual materials used throughout the textbook. This was a major project on its own, whose magnitude and difficulty was hard to anticipate.

Homework

The homework at the end of each chapter were conceived and curated by Roberta Sinatra. As a research faculty affiliated with my lab, Roberta has co-taught the network science class with me in the fall of 2014, helping also catch and correct many typos and misunderstanding that surfaced while teaching the material.

Science input

Throughout the project I have received comments, recommendations, advice, clarifications, and key materials from numerous scientists and students. It is impossible to recall them all, but I will try.

Chaoming Song helped estimate the degree exponent of scale-free net- works and helped me uncover the literature pertaining to cascading failures. The mathematician Endre Csóka helped clarify the subtle details of the Bollobás model. I have benefited from a great discussion with Raissa D’Souza on optimization models, with Ginestra Bianconi on the fitness model, and with Erzsébet Ravasz Reagan on the Ravasz algorithm. Alex Vespignani was a great resource on spreading processes and degree correlations. Marian Boguña has snapped the picture for the Karate Club Trophy. Huawei Shen calculated the future citations of research papers. Gergely Palla and Tamás Vicsek helped me understand the CFinder algorithm and Martin Rosvall pointed us to some key material on the InfoMap algorithm. Gergely Palla, Sune Lehmann andSanto Fortunato offered critical comments on the community detection chapter. Yong-Yeol Ahn helped me develop the early version of the material on spreading phenomena, and Kate Coronges has helped improve the clarity of the first four chapters. Ramis Movassagh, Hiroki Sayama and Sid Redner have provided careful feedback on several chapters.

Publishing

Simon Capelin, my longtime editor at Cambridge University Press, has been encouraging this project even before I was ready to write it. He also had the patience to see the book to its completion, through many missed deadlines. Róisín Munnell has helped move the book through production within Cambridge.

Institutions

This book would not have been possible if several institutions did not offer inspiring environments and a supporting infrastructure. First and foremost I need to thank the leadership of Northeastern University, from its President, Joseph Aoun, its Provost, Steve Director, my deans, Murray Gibson and Larry Finkelstein, and my department chair, Paul Champion, who were true champions of network science, turning it into a major cross-disciplinary topic within Northeastern. Their relentless support has lead to the hiring of several superb faculty focusing on networks, spanning all domains of inquiry, from physics and mathematics to social, political, computer and health sciences, turning Northeastern into the leading institution in this area. They have also urged and supported the creation of a network science PhD program and helped found the Network Science Institute lead by Alessandro Vespignani.

My appointment at Harvard Medical School, through the Network Medicine Division at Brigham and Women's Hospital and Center for Cancer Systems Biology at Dana Farber Cancer Institute, offered a window on the applications of network science in cell biology and medicine. Many thanks to Marc Vidal from DFCI and Joe Loscalzo from Brigham, who, as colleagues and mentors have defined my work in this area, an experience that found its way into this book as well.

My visiting appointment at Central European University, and the network science class I teach there in the summer, have exposed me to a student body with economics and social science background, an experience that has shaped this textbook. Balázs Vedres had the vision to bring network science to CEU, George Soros convinced me to get involved with the university and President John Shattuck and Provosts Farkas Katalin and Liviu Matei, with their relentless support, have smoothed the path toward CEU's superb program in this area, giving birth to CEU's PhD program in network science.

Thanks to the place where it all began: As a young assistant Professor, University of Notre Dame offered me the support and the serene environment to think about something different. And big thanks to Suzanne Aleva, who followed my lab from Notre Dame to Northeastern, and worked tirelessly for over a decade to foster an environment where I can focus, uninterrupted, on science.

Finally, I am truly indebted to my children, Dániel, Izabella and Lénárd, and my wife, Janet, who accepted that I devote countless hours to this book. Without their understanding and patience I could have never finished this book.