Experimental Platform for Big Data and Text Mining

点击翻译按钮获取中文摘要

Our group of researchers from the University of Ottawa's Text Analysis and Machine Learning lab is involved in a research program on novel techniques for data mining and text mining from very large datasets available on the Web or from other sources (so-called "big data"). Examples include mining social media texts (twitter messages, blogs, etc.), webpages, databases, and collections of scientific articles; predicting molecular interactions from very large datasets in bioinformatics; and analyzing graphs which represent enormous social networks. Our research objectives share a pattern of supporting experiments: it is necessary to explore the settings for multiple parameters used by different algorithms. The established evaluation methods also call for repeated runs with different training data sets and different test data sets. A multi-processor compute server and an accompanying data server are needed in order to enable efficient and feasible experiments of this scale.