Dr. Georgiana Ifrim is an Assistant Professor in the School of Computer Science, University College Dublin, and an SFI Funded Investigator with the Insight Centre for Data Analytics, Ireland. She holds a PhD (2009) and MSc (2005) in Computer Science (research area: Machine Learning), from Max-Planck Institute for Informatics, Germany and a BSc (2003) in Computer Science from University of Bucharest, Romania. Until 2014, Dr. Ifrim held a senior research fellow position with the Insight Centre for Data Analytics, University College Dublin, Ireland. Prior to that, she held postdoctoral positions with the Cork Constraint Computation Centre (2013), Ireland, and the Bioinformatics Research Centre (2010), Denmark.
Dr. Ifrim’s research focuses on scalable machine learning and data mining, in particular, large scale sequence learning, time series forecasting and information extraction/retrieval. She has worked in application domains ranging from Web mining, news and social media, energy and biology. Her current research interests include scalable sequence learning and real-time prediction for streaming data.
- Sequence Learning (Classification and Regression for Symbolic Sequences & Time Series)
- Sequence learning with all-subsequences (ECMLPKDD17, ICDE17, PlosOne14, KDD11, KDD08)
- Product review rating using opinion mining (COLING10)
- Learning from Massive News and Social Streams
- Connecting news and Twitter streams (TKDE17, WWW16, ECML16, ECML14)
- Twitter event detection (Winner of SNOW@WWW14 Data Challenge)
- Energy-Efficiency Applications
- Energy price forecasting for cost-aware scheduling design (SUSCOM14, CP13, Opt4Smartcities@CP13, CP12)
- Extracting and Exploiting Knowledge Graphs (aka ontologies)
- WordNet/Yago (knowledge graphs), Naga (graph search) (ICDE08, PKDD06, KDD06)
Open Source Software
- Twitter-Topics: Twitter Event Detection (winner of SNOW@WWW14 Data Challenge, twitter-topics)
- SLR: Structured Logistic Regression (KDD08, please use SEQL, it extends SLR and is better maintained). Fast Logistic Regression for Text Categorization: learning phrase-classifiers (ngram-classifiers) with variable-length phrases (ngrams).