论文信息 - Using Document Re-ranking for Chinese Information Retrieval System

Using Document Re-ranking for Chinese Information Retrieval System

For information retrieval, users hope to acquire more relevant information from the top N ranking documents. In this paper, we describe a Chinese Information Retrieval (IR) system that use a document re-ranking method based on term distribution of the top 100 ranking documents, to improve the precision of the 1000 ranking documents by reordering the initial retrieval documents. The experiment with NTCIR-3 formal Chinese test collection shows that the proposed method of document re-ranking can improve the precision of the retrieved documents.

Chen Yu

[1] Dong-Hong Ji,et al. Chinese Information Retrieval Based on Related Term Group , 2005, NTCIR.

[2] José Carlos González Cristóbal,et al. MIRACLE Retrieval Experiments with East Asian Languages , 2005 .

[3] Yang Lingpeng,et al. Document Re-ranking Based on Automatically Acquired Key Terms in Chinese Information Retrieval , 2004, COLING.

[4] Tiejun Zhao,et al. Documents Ranking Based on a Hybrid Language Model for Chinese Information Retrieval , 2006, 2006 IEEE International Conference on Information Acquisition.

[5] David M. Blei,et al. Topic segmentation with an aspect hidden Markov model , 2001, SIGIR '01.