A Machine Learning Approach for Complex Information Retrieval Applications
A Machine Learning Approach for Complex Information Retrieval Applications by Luo Si - Machine Learning Summer School at Purdue, 2011. The explosive growth of WWW contents and other types of digital media has created a critical problem of information overload. Information retrieval techniques provide the search capabilities of ranking documents for user queries according to degree of relevance, and more generally provides solutions of acquisition, storage, organization, retrieval and analysis of information. The old generation of information retrieval techniques used intuitive heuristics with limited justifications. More recently, formal learning techniques with more solid foundation have been applied to information retrieval applications and have obtained promising results. This talk presents some research for designing effective machine learning algorithms for different complex information retrieval applications for: 1). Modeling connected information items instead of isolated information items; 2). Modeling partially observed knowledge when full knowledge is absent; 3). Modeling evidence of information items from heterogeneous sources. These techniques will be presented with information retrieval applications of federated search, expertise search and question answering.
|A Machine Learning Approach for Complex Information Retrieval Applications|
Machine Learning Summer School at Purdue, 2011