CAPTAIN’S LOG

I just don’t dig CVs and professional biographies.

2020 – Infinity

DataKolektiv’s website goes online. A strategic decision is made: DataKolektiv will focus the majority of its efforts on bridging Data Science with the Semantic Web where data structures that could really support AI are being built: graph-based, rich data structures + probabilistic inference in ML. I am looking for highly skilled analysts, software engineers, and research scientists to join me. On the second thought, given the scope and complexity of the work in contemporary Data Science and what DataKolekiv will be focusing on… I begin to understand that it might prove to be a long quest. The World slowly recovers from the COVID19 induced crisis and lockups. A new chapter opens up.

A ship is safe in harbor, but that’s not what ships are for.

2010 – 2020

2017. Understanding that I have a handsome of consultancy agreements in Data Science and Analytics at my fingers, I founded a legal vessel, DataKolektiv: the mothership.

A horse, a horse, my kingdom for a horse”

Not a month since the foundation of DataKolektiv passes and I begin to provide Data Science services for Wikidata and Wiktionary to Wikimedia Deutschland. The pace of learning is high again and it feels good. So very good. My stack: R, Python, Hadoop, Spark via Pyspark, SQL, SPARQL, RDF, XML, JSON; MariaDB, PostgreSQL, one nice Spark analytics cluster, several number-crunching servers, Horizon – a dedicated virtual infrastructure, Git/Gerrit. My focus: distributional semantics of Wikidata usage in Wikipedia and other Wikimedia projects, potentials for knowledge transfer across different Wiktionaries, ad hoc analytics and campaign evaluations for WMDE, studies of the possibility for the optimization of the Wikidata Query Service (WDQS) – a SPARQL query endpoint running on Blazegraph, and many more.

The position of a Data Scientist for Wikidata comes as a total surprise to me: a FOAF shares a social media post where it is mentioned that Wikimedia Deutschland is looking for an Analyst; our mutual connection just tags me. I get in touch with them, apply for the position, and in less than two weeks my onboarding begins. After studying human semantic memory for almost twenty-five years I find myself faced with the task of doing Data Science on the most complex collaborative semantic network that has ever evolved.

Consulting Data Science and Engineering companies in Serbia from time to time. Part of the Data Science Serbia initiative, acting as a member of the Executive Board. Volunteering as a lecturer in Introduction to Data Science in R course with Startit.

2015. Data Scientist for DiploFoundation, building NLP and Information Retrieval systems from scratch in R, SQL, social media analytics, the development of a novel model of Twitter analytics for the Geneva Engage Award, an initiative of DiploFoundation and the Geneva Internet Platform.

2014. Brief cooperation with the Center for the Study of Bioethics, Belgrade, and HeLEX Centre, University of Oxford.

2013. My doctoral thesis, On the Rationality of Cognition, is completed and defended at the University of Belgrade. My thesis committee comprises a mathematician (and a psychologist), a theoretical psychologist, and a philosopher of science. I present a novel model of choice under risk based on Bayesian theory and prove it to be empirically superior to the existing body of models. 500 pages of a thorough analysis of the Rationality Debate in psychology. It is a theoretical thesis essentially focused on the question of the applicability of mathematics in behavioral sciences. Professor Gordana Jovanović accepts to supervise it and I have never enjoyed more academic freedom than then.

A final and fundamental break up with the academic circles in Belgrade takes place. My conclusions: 99% of the academic environment in this country is too slow, too lazy, happy only when well-funded by the taxpayer’s money while not having any real responsibility to deliver useful results. It barely swims in a growing bureaucracy and constantly complains about something. Most important: no one has a vision. Vision zero.  

“I find your lack of faith disturbing”

2012. Greg Murphy, James Hampton, and I publish Semantic Memory Redux: An Experimental Test of Hierarchical Category Representation based on our work in NYC 2005 – 2007 in the Journal of Language and Memory. It gets cited in Stevens’ Handbook of Experimental Psychology and Cognitive Neuroscience in 2018 and continues to receive citations even nowadays. Upon finishing the work we decide to dedicate the paper to our memory of Edward E. Smith, an influential scholar in semantic memory, the author of the Semantic feature-comparison model, and Greg’s co-author, who passes away on August 17, 2012.

Again, various freelance arrangements as an analyst and researcher.

I lecture Cognitive Psychology and Learning and Higher Cognition at the Faculty for Media and Communications in Belgrade.

I begin cooperating with the Libertarian Club Libek in Belgrade, lecturing on Rational Choice at their Academy of Liberal Politics. The cooperation evolves into a lasting friendship.

2000 – 2010

Belgrade, Serbia. Less than a year as Head of Research in a well-positioned Market Research agency: campaign evaluations, brand perception, pharmaceutical industry, banking, gaming. Various freelance arrangements as an analyst and researcher. I decide to complete the PhD program at the University of Belgrade. I am reading obsessively again: from Morgenstern and von Neumann, Ramsey, and Savage, to Kahneman and Tversky, Birnbaum and Luce.

2005 – 2007: first two years in the doctoral program in Cognition and Perception at the New York University, under supervision of Greg Murphy. I cannot recall a time of my life when I have learned more and faster: statistics from Laurence T. Maloney (who is probably the best lecturer that I have met in my life), statistics and causal induction from Bob Rehder (responsible for my conversion to Bayesian theory), optimality and evolution from Gary Marcus. My comeback to programming happens in NYC through Matlab for Lary’s famous Simulation & Data Analysis graduate course on statistics, simulation, and estimation theory in experimental psychology and psychophysics. I begin to learn R. My new interest: Decision Theory, especially Choice Under Risk, the empirical problem closest to the fundamental questions of representational measurement.

NYC is beautiful but orders of magnitude faster and more stressful than I find acceptable. I leave NYU in good standing and head back to Belgrade in 2007.

2004. A Diploma in Psychology from the University of Belgrade, finally. My thesis is a study of conceptual coherence, an attempt at the measurement of the consistency of the semantic structures in human memory. Supervised by Aleksandar Kostić, a longtime mentor in the Laboratory for Experimental Psychology during the 90s.

2001/2002. I drafted a proposal and a project for the foundation of the Center for Research on Information Technologies (CePIT) in the Belgrade Open School, a Belgrade based NGO whose educational programs I have attended in the previous years. The Olof Palme International Center recognizes this effort and provides funding 2002 – 2005. In three years, 2002 – 2005, I form and lead a team of young researchers who conduct the first empirical research on Information Society development and Internet Behavior in the region of South-Eastern Europe, spanning local, national, and international samples of respondents. I edit, co-author, and publish five books on the topics of Information Society and Internet Behavior, most of them written before I turned thirty. I still do not even have a diploma from the University, which I will earn only in 2004. The media coverage of the Center in three years surpasses the media coverage of the whole Belgrade Open School – among the most prominent Serbian NGOs – in the previous ten years. The success of this research program is not dear to everyone’s heart. You can feel it in the air that they want my head at the University and I begin to actually have fun in the whole situation while ignoring the background inter-personal and pseudo-political games. Headhunters for major international corporations begin to contact me and offer top research managerial positions. I turn them down.

“What matters most is how well you walk through the fire”

Lessons learned on the physics of academia: In a small system, academia tends to attract any researcher into its hierarchical structure where inventions take place only rarely.

Around 2002/03: began cooperation with DiploFoundation. The diplomatic environment is natural to me because of my family background, however I do not see myself in politics. Jovan Kurbalija, the Founding Director of DiploFoundation, is pushing new and innovative approaches in the field of diplomatic studies and practice of diplomacy as well. He is among the first scholars to recognize the importance of the study of Internet Governance in international relations and diplomacy, but he is surrounded by so much inertia in the traditionally slow-evolving context of diplomacy and international politics. I engage as a researcher with Diplo, and in the following decade or so we will complete several interesting research and Data Science projects together in the field of Internet Governance, including research cooperations with stakeholders such as AT&T and the Commonwealth Internet Governance Forum. I take part in the first World Summit on the Information Society (WSIS) in Geneva, 2003, and also join the initiative for the Wikimedia movement in Belgrade.

1995 – 2000

1999/2000: I attend two International Summer Schools in Cognitive Science at the New Bulgarian University (NBU), Sofia: courses on neural networks from James L.  McClelland (the co-editor of 1986. Vol I-II of Parallel Distributed Processing: Explorations in the Microstructure of Cognition, a book that played a pivotal role in the comeback of neural networks in Cognitive Science following the early criticism by Minsky and Papert in 1969), Jeff Elman (see: Elman network, i.e. RNN), and Elizabeth Bates, founders of the PDP group. The great International Schools at the NBU have been organized thanks to the efforts and the reputation of prof. Boicho Kokinov, who manages to gather some of the finest scholars in Cognitive Science of the time in Sofia. A great scholar in analogy-making, judgment, and memory, Kokinov dies while he was only 53. I remember him as a very kind person.

I am deeply involved in my experimental work in the Laboratory for Experimental Psychology at the Faculty of Philosophy. My focus: semantic memory, concepts and categorization, reasoning, measurement in behavioral sciences, multidimensional scaling (MDS). I read obsessively: journal papers, monographs, books… Very soon I begin to understand that the boundaries between social sciences (if there are any) – psychology, sociology, economics, political science, anthropology – are a product of power relations in the academia only. They are all addressing a set of similar – and most of the time the same – problems. I approach them with the mind of a mathematician and face many misunderstandings from colleagues of a humanistic background who still claim to be doing science. Science is measurement + mathematics, but the interaction between the two in behavioral and social sciences is extremely complicated, I try to explain. For them, science is telling stories that something is a science because it uses a Likert-type scale or a reaction time measurement in a behavioral experiment. Only a few of them understand the fundamental importance of the derivation and justification of data structures from experimental observations in building a theory of behavior.

Lessons learned on the epistemology of academia: scientific problems can be divided into fundamental and publishable.

I have published four journal papers already and presented my work at numerous conferences and invited talks. Still, my diploma is far, far away. 

Science is beautiful but relations in academia are not. Everything seems to be about positions and privileges, while only a minority are truly focused on fundamental research.

“I think God, in creating man, somewhat overestimated his ability.”

1990 – 1995

I begin working on my first scientific publication – on the process of semantic verification in human memory – when I was nineteen. It was published in an internal publication of the Laboratory of Experimental Psychology, Faculty of Philosophy, when I was twenty, and later in the Yugoslavian psychology journal “Psihologija” in 1996. I am completely into the naturalistic, the empirical work, the experimental method, in sharp contrast with my previous interest in pure logic and metamathematics.

1993. Prof. Predrag Ognjenović asks me to start working as a TA in General Psychology and study cognitive psychology in the Laboratory for Experimental Psychology, Faculty of Philosophy, as soon as I passed this (considered to be notoriously difficult, back at the times) exam. The offer comes as a consequence of a single question that I have posed during one of his lectures on cognitive information processing in the second semester, after which he invited me to conduct an experimental study on it – while I was still a first-year student with no coursework completed at all. In the Lab, I start learning from Aleksandar Kostić, an expert in psycholinguistics and one of the best scientists that I have met in my life (note: the competition is tough). It is him to whom I own the understanding of what empirical work in science truly means and who influenced my interest in probability and information theory.

In 1992, I have enrolled in the Department of Mathematics, Uni. Belgrade: besides genuine interest in Computer Science it was the easiest way for me to avoid being recruited and sent somewhere to fight in the Yugoslav civil war – a war that was everything but not mine. I study logic and algebra with a passion only to begin to understand how human minds are immensely more interesting to me than computers. In 1993, I enroll in the Department of Psychology. Sometime in 1994 or 1995, I don’t remember anymore, I am enrolled in the Department of Philosophy and the Department of Psychology in parallel, but I give up realizing how mad it is.

Late 80s and early 90s: I am programming faster and better than any of my informatics high school teachers in the XIII Belgrade Gymnasium: Pascal, Fortran, C, even Prolog. I enroll in the scientific programs of the Petnica Science Center for young, talented pupils in Yugoslavia, but got bored pretty soon: most of the young teaching associates in computer science from the Department of Mathematics and the School of Electrical Engineering of the University of Belgrade are all about showing off, mystifying knowledge instead of sharing it. However, there are some brilliant guys too. Some of us begin to study on our own; I am reading the Niklaus Wirth Bible, Algorithms + Data Structures = Programs. A friend and later a future colleague at the Department of Mathematics introduces me to the music of Nick Cave. At the campus, we drink and party a lot together in the evenings and code in the mornings and afternoons. I listen to the carefully tailored power stories told by the “elders” in a wannabe elite institution and begin to understand the essence of the Yugoslavian and future Serbian system: the narrative about the reality is stronger and more important than the reality itself, while the hierarchical rank is certainly more important than creativity and intelligence.

The Yugoslav Civil War begins. I immediately choose to be a moral universalist – a position that I will hold on strongly to for the rest of my life, alongside anarcho-liberalism that I will discover only later. However, nationalism proves to be invincible. A tragedy begins to unfold in front of my eyes.

“I dislike people who get out of things unscraped. No scars, no scratches.
Agnosceo veteris vestigia flamme. Refined through a scar.”

1980 – 1990

In late 1983, my father handles me the first issue of a newly started Yugoslavian microcomputer magazine. “Take a look” he says. He didn’t even open it: it was meant as a present, something to spawn my curiosity.  In 1984, my first computer programs in BASIC and pseudo-code are written out in a checkered notebook. One cannot still find any microcomputers on a socialist Yugoslavian market, and in 1986 my mother brings me a Commodore 64 as a present from her visit to France. I program in BASIC, assembler, and Pascal. In 1988, while still in the elementary, I started publishing game reviews for popular computer magazines in former Yugoslavia. I am fifteen when I publish a review of a Pascal compiler for Commodore 64 in the Slovenian computer magazine Moj mikro. I switch to Commodore Amiga and then quickly to a monochromatic Atari ST, better for programming.

1974 – 1980

I was born in Belgrade in 1974. When I was three, my family moved to Rome, Italy, where my father took the post of the military attaché of the Socialist Federative Republic of Yugoslavia in Italy. The first video game I remember playing is a heavy, robust Pong-like tennis console with two simple bars moving up and down to bounce a pixelized ball: huge red pixels on a black background. I am growing up peacefully and protected, my Italian is better than my Serbo-Croatian since I am in a public kindergarten and later enrolled in a public elementary school. I remember having many friends and endlessly enjoying myself in the first anime television series: Goldrake, Mazinga Z.

“It’s not the size of the dog in the fight, it’s the size of the fight in the dog.”