新型信息检索模型发展研究

Research on Development of New Information Retrieval Models

介绍了3个新型信息检索模型——信念网络模型、粗糙集理论检索模型和遗传算法检索模型。认为信念网络模型以概率推理为基础,推理结果说服力强,并采用图形化网络结构直观地表达变量的联合概率分布及其条件独立性,能大量节约概率推理计算;粗糙集理论检索模型通过不可分辨关系确定问题的近似域,对问题不确定性的描述和处理具有客观性;遗传算法检索模型模仿自然界中生物的遗传和进化机理,可以方便地应用遗传操作算子,具有独特的优越性。这3种检索模型由于理论独特、方法新颖和效果良好,已经得到了一定的应用,逐渐成为研究热点。

The paper introduces new information retrieval models including belief network model, rough set model, and genetic algorithm model. The paper thinks that belief network model take probability theory as the foundation, its reasoning results are persua- sive, it adopt graphical network structure to express variable＇s joint probability distribution and conditional independence visually, which reduce probabilistic inference computation greatly; rough set model determine approximate domain by indistinguishable relation- ship, so that it can describe and handle with the uncertainty objectively; genetic algorithm model simulate organism＇s genetic and evo- lutional mechanism in nature, so that it can apply genetic operators conveniently and have unique superiority. The three models have been applied to a certain extent and become research hotpots gradually for their unique theories, new methods and good results.