Effective Information Retrieval for Partitioned Document Collections

点击翻译按钮获取中文摘要

Current information retrieval services make use of massive indexes in order to resolve content-based queries. Monolithic approaches like this have been effective until now because the volume of data stored has been manageable on a single machine or tightly-coupled cluster of machines, and because the data has been available for collection. But with an increasing amount of automatically generated data, and an increasing diversity of information sources, other approaches are required. In this project we will investigate mechanisms for handling retrieval tasks when the indexes to the data are stored locally with the data, and when no central index is viable.