PKUspace: A collaborative platform for scientific researching

Abstract

This Paper introduces PKUSpace - a collaborative platform for scientific learning and researching. Supported by efficient automated metadata extraction and powerful text classification modules, PKUSpace provides users with a very helpful and interactive procedure to upload e-resources. We use heuristic methods and regular expression matching techniques to mine different patterns and extract relevant metadata. Upon the well-organized e-sources, we builds a uniform retrieval interface with multi-function to provide taxonomy navigation, full-text search, structural search in the metadata level and metasearching module with novel result-merging method. Combining two on-line scoring methods: scoring with exponential distribution and metadata scoring, we achieve an effective merging result of matched documents. Furthermore, PKUSpace provides an academic forum for students and professional users in every community to discuss their ideas and research work freely. © Springer-Verlag 2004.