I am Head of Data Scientist at Jobandtalent and Research Collaborator in the Web Research Group at UPF.

Previous experiences: STUART, YAHOO LABS, TWITTER and IRISA/INRIA.

My background includes Predictive Models, Data Mining, Information Retrieval, Machine Learning, User Modeling, Recommendation Systems and Online Computational Advertising.

I have worked with many different large data collections such as Flickr, Yahoo News, Yahoo query- and web-logs, and as well on Twitter data.

I have also done various researches on credit card user's transactions, Sentiment Analysis with Yahoo and Yelp data, and geographic localization of images and videos.

I am a Mac/Unix user, and my specialties are Apache Hadoop/PIG, Java, Python and R.

About Me

michele.trevisiol (at) upf.edu
@trevi



Head of Data Scientist
@ Jobandtalent (Barcelona)

Places where I have been

Selected Presentations

Publications

2015

Michele Trevisiol, Luca Maria Aiello, Paolo Boldi, Roi Blanco “Local Ranking Problem on the BrowseGraph”, August 9 – 13 2015, Santiago, Chile, ACM, Special Interest Group on Information Retrieval, SIGIR2015

Michele Trevisiol “Exploiting Implicit User Activity for Media Recommendation”, 2015, SIGIR Forum 49(1): 70, SIGIRForum2015

2014

Michele Trevisiol, Luca Maria Aiello, Rossano Schifanella, Alejandro Jaimes “Cold-start News Recommendation with Domain-dependent Browse Graph”, October 6 – 10 2014, Foster City, Silicon Valley, USA, ACM Recommender System conference, RecSys2014

Michele Trevisiol, Luca Chiarandini, Ricardo Baeza-Yates “Buon Appetito - Recommending Personalized Menus”, September 1 – 4 2014, Santiago, Chile, ACM conference on Hypertext and Social Media, HT2014

Miriam Redi, Michele Trevisiol, Neil O'Hare, Rossano Schifanella, Alejandro Jaimes “6 Seconds of Sound and Vision: Creativity in Micro-Videos”, June 24 – 27 2014, Columbus, Ohio, USA, Computer Vision and Pattern Recognition, CVPR2014

2013

Claudia Hauff, Bart Thomee, Michele Trevisiol, “Working Notes for the Placing Task at MediaEval 2013”, Oct 18 – 19 2013, Barcelona, Spain, MediaEval Benchmarking Initiative for Multimedia Evaluation (Placing Task), MediaEval2013

Luca Chiarandini, Przemyslaw A. Grabowicz, Michele Trevisiol, Alejandro Jaimes, “Leveraging Browsing Patterns for Topic Discovery and Photostream Recommendation”, July 8 – 10 2013, Boston, USA, International AAAI Conference on Weblogs and Social Media, ICWSM2013

Michele Trevisiol, Guillaume Gravier, Jonathan Delhumeau, Hervé Jégou, “Retrieving geo-location of videos with a divide & conquer hierarchical multimodal approach”, April 16 – 19 2013, Dallas, Texas, USA, [Best Paper Runner-up] ACM International Conference on Multimedia Retrieval, ICMR2013

Marek Lipczak, Michele Trevisiol, Alejandro Jaimes, “Analyzing Favorite Behavior in Flickr”, January 7 – 9 2013, Huangshan China, Springer International Conference on Multimedia Modeling, MMM2013

2012

Michele Trevisiol, Jonathan Delhumeau, Hervé Jégou, Guillaume Gravier, “How INRIA/IRISA identifies Geographic Location of Videos”, October 4 – 5 2012, Pisa, Italy, MediaEval Benchmarking Initiative for Multimedia Evaluation (Placing Task), MediaEval2012

Michele Trevisiol, Luca Chiarandini, Luca Maria Aiello, Alejandro Jaimes, “Image Ranking Based on Users Browsing Behavior”, August 12 – 16 2012, Portland, USA, ACM, Special Interest Group on Information Retrieval, SIGIR2012

Luca Chiarandini, Michele Trevisiol, Alejandro Jaimes, “Discovering Social Photo Navigation Patterns”, July 9 – 13 2012, Melbourne, Australia, IEEE International Conference on Multimedia and Expo, ICME2012

2011

Yannis Kalantidis, Lluis Garcia Pueyo, Michele Trevisiol, Roelof van Zwol, and Yannis Avrithis, “Scalable Triangulation-based Logo Recognition”, April 17 – 20 2011, Trento, Italy, ACM International Conference on Multimedia Retrieval, ICMR2011

2010

Michele Trevisiol, “Logo Detection Application: Extraction, Matching, Segmentation and Classification”, April 13 2010, Padova, Italy, Master Thesis

Patents and Defensive Publications

Patents

Michele Trevisiol, Gabriele Tolomei, Nicola Barbieri, Mounia Lalmas, Fabrizio Silvestri and Puneet Mohan Sangal. “Generating Actionable Suggestions for Improving User Engagement”. US SN 14/954,118, Filled: Nov 2015.

Michele Trevisiol, Luca Maria Aiello, Luca Chiarandini and Alejandro Jaimes. “User Behavior Models Based On Source URL”. US 2014/0189525, Filled: Dec 2012, Published: July 2014.

Defensive Publications

Michele Trevisiol, Olivier van Laere and Gianmarco de Francisci Morales. “A Method and System for Retrieving Geo-Location of Media Content with a Hierarchical Divide and Conquer Approach”. Published in IP.com, Aug 2014.

Michele Trevisiol, Luca Maria Aiello, Luca Chiarandini and Alejandro Jaimes. “Method for Identifying Interests of Users in Online Content and Categorizing the Online Content”. Published in IP.com, Dec 2013.

Michele Trevisiol, Lluis Garcia Pueyo, Yannis Kalantidis and Roelof van Zwol. “Scalable Triangulation Based Logo Recognition”. Published in IP.com, April 2011.

Press

MIT TechReview presents “Yahoo Labs' Algorithm Identifies Creativity in 6-Second Vine Videos”, MIT TechReview - Emerging Technology From the arXiv, Nov. 24, 2014.

UPF Actualitat presents the winners of the Innova Challenge, “Investigadors del Grup de Recerca en Web guanyen l'Innova Challenge Big Data”, UPF Actualitat - Recerca i innovació, Jan. 10, 2014

Gnip Blog presents an interview about Image Analysis and Flickr by Elaine Ellis, “Data Story: Michele Trevisiol on Image Analysis”, Gnip Blog - Data Story, Nov. 12, 2013

Challenges

BBVA Innova Challenge

I participated with Luca Chiarandini at the Innova Challenge, the first time the bank has opened his real trade data for a challenge. We participated in two of the three categories.

Qkly

Qkly - Winner of the First Edition Innova Challenge Big Data by BBVA Bank

“Application for Customers” category, December 2013.


What is it?

How many times did you get stuck in the queue at a supermarket or at a grocery store? How do you feel to go shopping during the sales period? And when you're visiting some attractions or museums? Have you ever wondered if there is a way in which you could have avoided it?

Qkly automatically produces optimized schedules of all the business, attractions, or places based on your preferences. Using historical data provided by the BBVA Open API, Qkly is able to estimate how crowded a place is at a certain time. Based on this information, it helps you to pick the time in which less people are present, so as to save time and avoid stress.

Links

Official Website
Google Play App

Award

Innovacion BBVA

Press

UPF Actualitat

Qkly
Qartz Screenshot

Quartz - 2nd Winner of the First Edition Innova Challenge Big Data by BBVA Bank

““Visualization”” category, December 2013.


The Interface

Quartz allows you to explore how people spend their money. The pie charts on the map represent the amount of money spent by people at a particular location, split by customer gender or business category.

The Filtering

The legend on the bottom right corner allows you to filter out genders or categories you are not interested in.

The Quartz Wheel

If you want more information, you can click on a pie chart to open the Quartz Wheel. If you move the mouse on a segment of the wheel, you can see the number of payments in the center. Also, you can see additional segments on the wheel representing how the payments in the category distribute in sub-segments. If you move from a segment to one of its sub-segment, you will see the percentage of payments in the segment which refer to it, as well as the percentage on the total amount for the location.

Services

PC Member, 7th International Conference on Social Informatics (SocInfo2015), December 2015

PC Member, 9th ACM Conference on Recommender Systems - Demos and Posters (RecSys2015), November 2015

Keynote Speaker, International Symposium on Web Algorithms (iSWAG2015), June 2015

PC Member, International Conference on Social Informatics (SocInfo2014), November 2014

PC Member, Workshop on Human Centered Event Understanding from Multimedia (HuEvent2014) co-located with ACM Multimedia, November 2014

Interviewed by Elaine Ellis at Gnip Blog, November 2013

Posters Chair at Placing Task, MediaEval 2013 (MediaEval2013), October 2013

Co-Organizer of Placing Task, MediaEval 2013 (MediaEval13), October 2013

PC Member, Multimedia Grand Challenge Solutions at ACM Multimedia (ACMGC2013), October 2013

Web Chair, ACM Multimedia (MM2013), October 2013