Scientific and Research Activities
Riccardo Guidotti
March 27, 2017
1
Research Activities
1.1
Classification of Publications
• Clustering Algorithms for Personal Data Mining [1, 13, 22] • Individual Models for Analyzing Human Behavior [10, 14, 15, 21] • Services for Improving Individual Mobility [8, 11, 19]
• Data-Driven Carpooling Services [4, 6, 16, 17, 20] • Nowcasting Well-Being [12]
• Link Prediction Intra e Inter Community [5, 18, 23] • Mobility Analytics with complex Networks [7, 24] • Musical Listening Analytics [9]
1.2
Research Interests and Scientific Activities
During the Ph.D. the main research activity has been the study of algorithms and models for extracting the systematic patterns characterizing human be-havior, and the development of individual and collective services with the aim of ameliorating the performance of existing services besides providing to the user an improved self-awareness. Riccardo Guidotti employed the developed methodologies on various types of data such as mobility data, transactional data, musical listening data and social networks.
Personal Data Analytics. During the Ph.D., supervisioned by Prof. Dino Pedreschi and Dott.ssa Fosca Giannotti, Riccardo Guidotti has studied and de-veloped techniques for the extraction of individual patterns from personal data, and he employed such knowledge for developing novel services.
Riccardo Guidotti developed two parameter-free clustering algorithms for personal data mining which are easily usable in a context where data mus t
be treated at individual level without the need of a comparison among vari-ous datasets and without violating the privacy of the users. In particular, the algorithms have been realized to solve the following problems:
• partitioning of the stops of a user to the aim of detecting his personal locations [13];
• partitioning of shopping transactions for extracting the individual behav-ioral patterns [1].
Moreover, the individual and collective clustering approach proposed in [22] al-lows to evaluate the points of interests of the shared by the users.
The personal patterns emerging from the data can be captured by indicators and models. Guidotti has defined the following individual models for the analysis of human behavior and for the extraction of the individual knowledge:
• indicators to evaluate the predictability of the shopping purchases [15]; • a user-centric model for mobility data [14];
• a personal data model for musical preferences [10].
In addition, Guidotti has proposed an approach that by exploiting constraint programming improved an existing data mining procedure for the extraction of personal mobility data models [21].
Algorithms and models for Personal Data Analytics can be exploited to develop services for improving individual mobility. Riccardo Guidotti has used the individual mobility profiles and the vision in in [14] for defining the following services:
• a route planner exploiting the collective knowledge for suggesting better routs [19];
• a trajectory prediction system which exploits individual, collective and hybrid movements patterns to produce the prediction [8];
• a system for evaluating compatibility among users which do not know each other but which frequently attend the same places at the same time [11].
Starting from the Master Thesis [24], Riccardo Guidotti developed data-driven carpooling services to the aim of reducing the number of circulating cars. Di seguito sono riportate le diverse tipologie di carpooling:
• a carpooling service which tries to integrate private and public means of transport [17, 20];
• a carpooling service considering the compatibility among drivers and pas-sengers while producing the recommendations[4, 16].
Finally, Riccardo Guidotti developed an individual an collective analysis of the shopping transactions able to nowcast the well-being of the society [12].
Intra - Inter Comunity Link Prediction. An additional research field has been the study of link prediction techniques on dynamic networks. In particu-lar, these techniques exploit community discovery for improving the prediction. The proposed methodology is suitable for predicting the link either within the community as well as outside [5, 18, 23].
Analisi di Mobilit`a attraverso Reti Complesse. Guidotti has employed a technique of complex network analysis to the aim of extracting two indicators for evaluating the significance of a locations in relationship to the significance of the users that have visited it and vice versa [7].
Analisi di Ascolti Musicali e Tracce Audio. Finally, Riccardo Guidotti has analyzed musical listening and musical track of emerging bands and famous artists to the aim of understanding the patterns leading to musical success [9].
1.3
Period Abroad
01/09/2014 – 18/12/2014 Internship at IBM Research Dublin, Eire, in the research group Smarter Urban Dynamics of Dr. Francesco Cal-abrese within the European Project FP7 PETRA. The research activity has been focused on the analysis of individual mobility with the objective of proposing novel carpooling solutions as valid alternatives as everyday means of transport.
1.4
Partecipazione Attiva a Progetti di Ricerca
SoBigData 2015-2018 (H2020 Program project under the scheme “INF-RAIA-1-2014-2015: Research Infrastructures” – Grant agreement #6540-24 “SoBigData: Social Mining & Big Data Ecosystem”1).
SoBig-Data project aims at creating the Social Mining & Big SoBig-Data Ecosystem, i.e., a research infrastructure providing an integrated ecosystem for ethic-sensitive scientific discoveries and advanced applications of social data mining on the various dimensions of social life. The participation of Ric-cardo Guidotti is related to the development of analytical methods for mobility and well-being to the aim of providing to the users an improved self-awareness.
Cimplex 2015-2018 (H2020 Project under the funding scheme “FETPRO-ACT-1-2014: Global Systems Science (GSS)” – Grant agreement #6411-912). CIMPLEX project proposes a visionary research to develop mod-eling, computational, and ICT tools needed to predict and influence dis-ease spread and other contagion phenomena in complex social systems. The participation of Riccardo Guidotti is related to the development of personal data mining methods for the exaction of individual and social behaviors.
PETRA 2014-2017 (SMARTCITIES Project under the funding scheme FP7-ICT – FET Open Grant agreement #6090423). The PETRA project
aims at developing a service platform that connects the providers and controllers of transport in cities with the travelers in a way that informa-tion flows are optimized, while respecting and supporting the individual freedom safety and security of the traveler. Cities will get an integrated platform to enable the provision of citizen-centric, demand-adaptive city-wide transportation services. Travelers will get mobile applications that facilitate them with personalized travel priorities and choices for route and modality The participation of Riccardo Guidotti is related to the devel-opment of a predictive mobility model exploiting both the individual and collective knowledge .
ICON 2012-2014 (Project under the funding scheme FP7-ICT-2011-C – FET Open Grant agreement #2847154). ICON project aims at developing
a new approach in which gathered data is systematically analyzed to dy-namically revise and adapt constraints and optimization criteria through a novel Inductive Constraint Programming paradigm that bridges the gap between the areas of data mining on one hand, and constraint program-ming on the other hand. The participation of Riccardo Guidotti is related to the development of an iterative loop optimizing the recommendations for a data-driven carpooling service.
1.5
Program Committee
Program Committee member for Workshop DyNo 2016: 2th Workshop on Dynamics in Networks5
2https://www.cimplex-project.eu 3http://www.petraproject.eu/
1.6
Attivit`
a di Revisione
He has been the reviewer for the following journals and conferences:
International Journals
FITEE: Frontiers of Information Technology and Electronic Engineering
TIST: Transactions on Intelligent Systems and Technology
DAMI: Data Mining and Knowledge Discovery
International Conferences
CIKM: Conference on Information and Knowledge Management
DyNo: Dynamics in Networks
ECML PKDD: European Conference on Principles and Practice of Knowl-edge Discovery in Databases
ICDE: International Conference on Data Engineering
ICDM: IEEE International Conference on Data Mining
SIGKDD: ACM International Conference on Knowledge Discovery in Databases
2
Didactic Activities
02/2016 – 03/2016: Didactic support (tutor) for the course “ICT: BI” - Master di secondo livello “Master MAINS: Management, Innovazione e Ingegneria dei servizi” at Scuola Superiore Sant’Anna. Lecturer: Dr. Anna Monreale.
03/2016 – ad oggi: Didactic support (tutor) for the course ‘Data Mining an Machine Learning ” - Master di secondo livello “Big Data Analytics” at University of Pisa. Lecturers: Prof. Dino Pedreschi, Dott.ssa Fosca Giannotti.
04/2016 – ad oggi: Didactic support (tutor) for the course “Mobility Data Mining” - Master di secondo livello “Big Data Analytics” at Univer-sity of Pisa. Lecturers: Dr. Mirco Nanni.
09/2015 – ad oggi: Didactic support (tutor) for the course “Data Min-ing: Fondamenti” at University of Pisa. Lecturer: Prof. Dino Pedreschi.
2.1
Supervision and Thesis Revision
2017, Correlator. Cono Stabile, Title: “Personalytics: Un’applicazione web per esplorare i propri dati personali”. Laurea Triennale in Informatica Umanistica, University of Pisa, February 2017. Final graduation mark: 107/110. Correlators: Dr. Anna Monreale.
2015, Correlator. Gian Felice Meloni, Title: “Analisi dei Pattern In-dividuali di Acquisto nella Grande Distribuzione”. Laurea Magistrale in Business Informatics, University of Pisa, April 2015. Final graduation mark: 107/110. Correlators: Prof. Dino Pedreschi, Dr. Diego Pennacchioli.
3
Scientific Awards
ISTI Young Researcher Award 2017, ISTI-CNR. Award for re-search activity during 2016.
ISTI Young Researcher Award 2016, ISTI-CNR. Award for re-search activity during 2015.
IBM Fellowship Award, IBM Research Dublin. Award for Ph.D. proposal and academic achievements 2014/2015.
Ph.D. Scholarship for the Ph.D. School “G. Galilei” at Dipartimento di Informatica dell’Universit`a di Pisa financed by MIUR 2013/2016.
References
3.1
International Conference Under Revision
[1] R. Guidotti, A. Monreale, M. Nanni, D. Pedreschi, F. Giannotti Clus-tering Individual Transactional Data for Masses of Users. Submitted to KDD 2017.
[2] R. Guidotti, G. Rossetti, L. Pappalardo, D. Pedreschi, F. Giannotti Next Basket Prediction using Recurring Sequential Patterns. Submit-ted to KDD 2017.
[3] R. Guidotti, L. Gabrielli, A. Monreale, D. Pedreschi, F. Giannotti Discovering Temporal Regularities in Retail Customers’ Shopping Be-havior. Submitted to PKDD 2017.
3.2
International Journals
[5] G. Rossetti, R. Guidotti, I. Miliou, D. Pedreschi, F. Giannotti A Su-pervised Approach for Intra/Inter-Community Interaction Prediction in Dynamic Social Networks. Social Network Analysis and Mining, SNAM, (Accepted to appear), 2016, Springer.
[6] R. Guidotti, M. Nanni, S. Rinzivillo, D. Pedreschi, F. Giannotti Never Drive Alone: Boosting Carpooling with Network Analysis. Information Systems, 2016, Elsevier. DOI:/10.1016/j.is.2016.03.006
[7] R. Guidotti, A. Monreale, S. Rinzivillo, D. Pedreschi, F. Giannotti Unveiling Mobility Complexity through Complex Network Analysis. So-cial Network Analysis and Mining, SNAM, 6: 59, 2016, Springer. DOI:10.1007/s13278-016-0369-2
[8] R. Trasarti, R. Guidotti, A. Monreale, F. Giannotti MyWay: Location prediction via mobility profiling. Information Systems, 2015, Elsevier. DOI:10.1016/j.is.2015.11.002
3.3
International Conferences and Workshops
[9] L. Pollacci, R. Guidotti, G. Rossetti “Are we playing like Music-Stars?” Placing Emerging Artists on the Italian Music Scene. 9th International Workshop on Machine Learning and Music at PKDD, MML 2016, Garda, Italia.
[10] R. Guidotti, G. Rossetti, D. Pedreschi Audio Ergo Sum: A Personal Data Model For Musical Preferences. 5th International Symposium “From Data to Models and Back”, DataMod 2016, Vienna, Austria, Springer.
[11] R. Guidotti, M. Berlingerio Where Is My Next Friend? Recommending Enjoyable Profiles in Location Based Services. 7th Workshop on Com-plex Networks, CompleNet 2016, Dijon France, pp. 65–78, Springer.
[12] R. Guidotti, M. Coscia, D. Pedreschi, D. Pennacchioli Going Beyond GDP to Nowcast Well-Being Using Retail Market Data. International Conference and School on Network Science, NetSciX 2016, Wroclaw, Poland, pp. 29–42,Springer.
[13] R. Guidotti, R. Trasarti, M. Nanni TOSCA: TwO-Steps Clustering Al-gorithm for Personal Locations Detection. 23rd International Confer-ence on Advances in Geographic Information Systems, SIGSPATIAL 2015, Seattle, USA, pp. 1–10, ACM.
[14] R. Guidotti, R. Trasarti, M. Nanni Towards user-centric data manage-ment: individual mobility analytics for collective services. 4th Interna-tional Workshop on Mobile Geographic Information Systems, SIGGIS 2015, Seattle, USA, pp. 80–83, ACM.
[15] R. Guidotti, M. Coscia, D. Pedreschi, D. Pennacchioli Behavioral Entropy and Profitability in Retail. 2nd International Conference on Data Science and Advanced Analytics, DSAA 2015, Paris, France, pp. 1–10, IEEE.
[16] R. Guidotti, A. Sassi, M. Berlingerio, A. Pascale, B. Ghaddar Social or green? A data-driven approach for more enjoyable carpooling. 18th International Conference on Intelligent Transportation Systems, ITSC 2015, Gran Canaria, Spain, pp. 842–847, IEEE.
[17] M. Berlingerio, V. Bicer, A. Botea, S. Braghin, N. Lopes, Nuno, R. Guidotti, F. Pratesi Managing travels with PETRA: the Rome use case. PKDD Nectar Track at PKDD. PKDD Nectar Track 2015, Porto, Portugal.
[18] G. Rossetti, R. Guidotti, D. Pennacchioli, D. Pedreschi, F. Giannotti Interaction Prediction in Dynamic Networks exploiting Community Discovery. International conference on Advances in Social Network Analysis and Mining, ASONAM 2015, Paris, France, pp. 553–558, IEEE.
[19] R. Guidotti, P. Cintia Towards a Boosted Route Planner Using In-dividual Mobility Models. International Workshop on Modelling and Knowledge Management applications Systems and Domains at SEFM, MoKMaSD 2015, York, UK, pp. 108–123, Springer.
[20] A. Botea, S. Braghin, N. Lopes, Nuno, R. Guidotti, F. Pratesi Manag-ing travels with PETRA: the Rome use case. Data MinManag-ing And Smart Cities Applications Workshop a ICDE. DAMASCA 2015, IEEE.
[21] L. Kotthoff, M. Nanni, R. Guidotti, B. O’Sullivan Find Your Way Back: Mobility Profile Mining with Constraints 21st International Conference on Principles and Practice of Constraint Programming, CP 2015, Cork, Ireland, pp. 638–653, Springer.
[22] R. Guidotti, A. Monreale, S. Rinzivillo, D. Pedreschi, F. Giannotti Re-trieving Points of Interest from Human Systematic Movements. Inter-national Workshop on Modelling and Knowledge Management appli-cations Systems and Domains at SEFM, MoKMaSD 2015, Grenoble, France, pp. 294–308, Springer.
3.4
Poster
[23] G. Rossetti, R. Guidotti, D. Pennacchioli, D. Pedreschi, F. Giannotti Time-Aware Interaction Prediction in Dynamic Social Networks ex-ploiting Community Discovery. IC2S2 2015, Helsinki, Finland.
3.5
Master and Bachelor Thesis
[24] R. Guidotti Mobility Ranking: Human Mobility Analysis using Rank-ing Measures Tesi di Laurea Magistrale in Informatica, Universit`a di Pisa, Pisa, Luglio 2013.
[25] R. Guidotti Shortest Path on Multigraph with Cost Modifiers – Sviluppo di misure su reti multidimensionali e integrazione delle stesse in framework Java Tesi di Laurea Triennale in Informatica, Universit`a di Pisa, Pisa, Ottobre 2010.