Document Actions
Paper on graph mining in data streams accepted at KDD 2011
The paper "Mining Frequent Closed Graphs on Evolving Data Streams" by Albert Bifet, Geoff Holmes, Bernhard Pfahringer, and Ricard Gavaldà has been accepted for oral presentation at KDD-2011 (the 17th ACM SIGKDD Conference on Knowledge Discovery and Data Mining). KDD is tied among the 2-3 top conferences in the world on data mining and related areas. This year only 56 papers out of 714 submissions (7.8%) were accepted for oral presentation.
The paper presents a framework for mining frequent subgraphs in high-speed streams whose items are graphs, with the additional ability to adapt to changes when the item distributinon changes over time. To our knowledge this is the first work able to do frequent graph mining in a time-evolving data streams. Possible applications domaing include social network analysis, chemical and biological informatics, communication networks, computer vision, video indexing, and the Web, as all of this provide examples of large-scale graphs and setes of graphs that must be analyzed online in essentially real time.
This work is part of a continuing joint research effort between LARCA and the Machine Learning group of the CS Department, U. Waikato (New Zealand), with emphasis on methods for learning and mining in Data Streams.
The paper presents a framework for mining frequent subgraphs in high-speed streams whose items are graphs, with the additional ability to adapt to changes when the item distributinon changes over time. To our knowledge this is the first work able to do frequent graph mining in a time-evolving data streams. Possible applications domaing include social network analysis, chemical and biological informatics, communication networks, computer vision, video indexing, and the Web, as all of this provide examples of large-scale graphs and setes of graphs that must be analyzed online in essentially real time.
This work is part of a continuing joint research effort between LARCA and the Machine Learning group of the CS Department, U. Waikato (New Zealand), with emphasis on methods for learning and mining in Data Streams.
