Design and implementation of a Chinese full-text retrieval system based on a probabilistic model

Full-text retrieval has the advantages of searching documents over a wide range of words and accessing original text directly. However, because of the inadequate processing speed and insufficient memory capacity, there has been little work done on full text retrieval systems in the past. With the emergence of very large capacity compact disk ROM (CD ROM), research in full-text retrieval systems is becoming increasingly important. Practical systems based on Boolean retrieval models have appeared. However, in spite of its ability to process structured queries, the Boolean model has been criticised for its inability to provide ranked output as all retrieved documents are considered equally important. Probabilistic models have been proposed by some scholars to solve this problem and have successfully been implemented on a small scale trial basis in indexing retrieval systems. This paper focuses on applying the probabilistic model into the full-text retrieval system.<>