ISSN 0439-755X
CN 11-1911/B
主办：中国心理学会
　　　中国科学院心理研究所
出版：科学出版社

心理学报 ›› 1997, Vol. 29 ›› Issue (3): 240-249. cstr: 32110.14.1997.00240

• • 上一篇    下一篇

基于特征的联想

刘述忠;程虎   

  1. 北京大学计算机系,中国科学院软件研究所
  • 发布日期:1997-09-25 出版日期:1997-09-25

FEATURE BASED ASSOCIATION

Liu Shuzhong(CASE Lab.,Department of Computer Science and Technology,Beijing University,100871)Cheng Hu(Institute of Software,Academia Sinica,Beijing,100080)   

  • Online:1997-09-25 Published:1997-09-25

摘要： 联想是认知科学中十分重要的现象，许多学者进行过不同角度的研究，获得了许多有意义的结果。根据有关对象（０ｂｊｅｃｔ）描述的基本概念，从联接机制的角度对联想问题进行一些探讨，并提出一个基于特征的联想模型ＦＢＣＡＭ（ＦｅａｔｕｒｅＢａｓｅｄＣｏｎｎｅｃｔｉｏｎｉｓｔＡｓｓｏｃｉａｔｉｏｎＭｏｄｅｌ）。在ＦＢＣＡＭ中，对象由其特征及特征的伴随值描述，从而可以获得更为精确的描述。对象的全部特征构成描述模式，伴随值则相应地构成伴随向量。在此基础上可以细致地刻画概念之间的相似性，并可对联想及注意机制进行细致的模拟与分析。文中利用联接机制的思想实现了ＦＢＣＡＭ并给出该系统的部分运行结果，表明ＦＢＣＡＭ可以较合理地解释联想过程

关键词: 相似性, 联想, 联接机制, 认知

Abstract: Association is very important in cognitive psychology and cognitive science.Researchers study it from different aspects and have got some significant results.Inthis paper the association is discussed from connectionist association modes,FBCAM(Featured-Based Connectionist Association Model),is described.In FBCAM,and object is described with its features and theaccompanying values of the features.All featuresof an object from the description pattern of the object and the accompanying valuesof the features form the accompanying vector,so that objects can be described precisely.Therefore the similarity between twoconcepts can be analyzed in detail,andthe association and attention can be analyzed and simulated with details.FBCAM isimplemented in the computer(SUN SPARC Workstation, C programming language).

Key words: similarity, association, connectionism, cognition