His research has made several key contributions to online popularity prediction, real-time tracking and countering disinformation campaigns, and understanding shortages and mismatches in labour markets. First, he has developed theoretical models for online information diffusion, which can account for complex social phenomena, such as the rise and fall of online popularity, the spread of misinformation, or the adoption of disruptive technologies. Second, he built a skill-based real-time occupation transition recommender system usable in periods of massive disruptions (such as COVID-19). Third, he approached questions such as “Why did X become popular, but not Y?” and “How can problematic content be detected based solely on how it spreads?” with implications in detecting the spread of conspiracy theories and disinformation campaigns. Finally, he linked social media predicted personality profiles with worker occupations, applicable in building personalising career recommendations.
Marian-Andrei's research receives funding from selective funders such as Facebook Research and Defence Science and Technology (DST). In addition, he publishes in the most selective venues, such as the PNAS, PLOS ONE, PLOS Computations Biology, WWW, NeurIPS, IJCAI, and CIKM. As a result, his work has received significant media attention—including Bloomberg Business Week, Nature Index, BBC, and World Economic Forum.
Marian-Andrei disseminates his research to the broader public by regularly contributing to The Conversation. In addition, he also leverages his research to real societal impact by, for examples, serving as an expert for the NSW government's Defamation Law Reform or providing evidence for the Australian Federal Senate inquiry into media diversity. See more at www.rizoiu.eu
My research revolves around artificial intelligence, machine learning and data mining. More specifically, I am interested in Social Network Analysis, popularity prediction, knowledge injection into non-supervised learning algorithms, data representation and temporal evolutions. I deal with large datasets of complex data (textual, image), often issued from the online social media and my main tools are modeling and simulation, clustering and topic modeling.
A little more details
My current research interest is to model theoretically popularity on online media, as well as estimate the influence of media content and network characteristics on online attention. We established a generative model that predicts online attention, based on an exogenously-driven Hawkes self-exciting processes. We also examine the geographical diffusion of media content over time and the goal is to generate statistical descriptions of content diffusion over time and geographical areas. We are handling very large Twitter datasets (the network), which relate to Youtube videos (the content).
My previous work dealt with how partial expert information can be leveraged into a non-supervised learning algorithm that treats complex data. This complex data is of different natures (text, image), it is temporal and structured, linked to knowledge repository (e.g. ontology) and/or labeled. Semi-supervised clustering is used to model the additional information (structure, labels, time) and to inject the heterogeneous information into the learning algorithm. A series of application emerge from the theoretical research: using the temporal dimension to detect temporal patters and typical evolutions, using the image labels to improve image numerical representation and an automatic topic evaluation using concept trees.
Note: This page lists publications from 2009 to 2016. For recent updates please see the lab publications page.
- M.-A. Rizoiu, J. Velcin, S. Bonnevay and S. Lallich, "ClusPath: A Temporal-driven Clustering to Infer Typical Evolution Paths," Data Mining and Knowledge Discovery, pp. 1–26, 2015.
- M.-A. Rizoiu, J. Velcin, and S. Lallich, "Semantic-enriched Visual Vocabulary Construction in a Weakly Supervised Context," Intelligent Data Analysis, vol. 19, iss. 1, pp. 161–185, 2015.
- M.-A. Rizoiu, J. Velcin, and S. Lallich, "How to use Temporal-Driven Constrained Clustering to detect typical evolutions," International Journal of Artificial Intelligence Tools, vol. 23, iss. 4, pp. 1460013, 2014.
- M.-A. Rizoiu, J. Velcin, and S. Lallich, " Unsupervised Feature Construction for Improving Data Representation and Semantics ," Journal of Intelligent Information Systems, vol. 40, iss. 3, pp. 501–527, 2013.
- C. Musat, M.-A. Rizoiu, and S. Trausan-Matu, "An Intra and Inter-Topic Evaluation and Cleansing Method," Romanian Journal of Human-Computer Interaction, vol. 3, iss. 2, pp. 81–96, 2010.
- M.-A. Rizoiu, L. Xie, T. Caetano, and M. Cebrian "Evolution of Privacy Loss on Wikipedia, " in Proc. International Conference on Web Search and Data Mining (WSDM '16), 2016.
preprint + SI: slides: poster: bibtex: Presentation page
- Y.-M. Kim, J. Velcin, S. Bonnevay, and M.-A. Rizoiu, "Temporal Multinomial Mixture for Instance-Oriented Evolutionary Clustering, " in Proc. European Conference on Information Retrieval (ECIR '15), 2015, pp. 593–604.
- M.-A. Rizoiu, "Semi-Supervised Structuring of Complex Data, " in Proc. Doctoral Consortium of the International Joint Conference on Artificial Intelligence (IJCAI '13), 2013, pp. 3239–3240.
preprint: slides: poster: bibtex:
- M.-A. Rizoiu, J. Velcin, and S. Lallich, "Structuring typical evolutions using Temporal-Driven Constrained Clustering," in Proc. International Conference on Tools with Artificial Intelligence (ICTAI '12), 2012, pp. 610–617.
preprint: slides: bibtex:
- C. Musat, J. Velcin, S. Trausan-Matu, and M.-A. Rizoiu, "Improving topic evaluation using conceptual knowledge," in Proc. International Joint Conference on Artificial Intelligence (IJCAI '11), 2011, pp. 1866–1871.
- C. Musat, J. Velcin, M.-A. Rizoiu, and S. Trausan-Matu, "Concept-based Topic Model Improvement," in Proc. International Symposium on Methodologies for Intelligent Systems (ISMIS '11), 2011, pp. 133–142.
- M.-A. Rizoiu, J. Velcin, and J.-H. Chauchat, "Regrouper les données textuelles et nommer les groupes à l'aide des classes recouvrantes," in Proc. Extraction et Gestion des Connaissances (EGC '10), 2010, pp. 561–572.
preprint: slides: bibtex:
- M.-A. Rizoiu and J. Velcin, "Topic Extraction for Ontology Learning," in book: Ontology Learning and Knowledge Discovery Using the Web: Challenges and Recent Advances, pp. 38–61, 2011.
- M.-A. Rizoiu, "Semi-Supervised Structuring of Complex Data," PhD Thesis, University Lumière Lyon 2, June, 2013.
preprint: abstract: slides: short version: abstract: bibtex: