Skip to Main Content
A Topic-based Document Retrieval Framework (TDRF) is proposed in this paper to resolve the topic-based document retrieval. The TDRF includes nine parts, of which Corpus Topic Learning, Query Topic Learning and Relationship Sorting are the core. Experiments on similar document retrieval showed that TDRF's instance outperforms the Vector Space Model (VSM) in average precision, recall and f-measure. The value of TDRF may lie in that it provides a simple, universal and novel methodology for document retrieval.