基于改进TF-IDF算法的牛疾病智能诊断系统  

CATTLE DISEASE INTELLIGENT DIAGNOSIS SYSTEM BASED ON IMPROVED TF-IDF ALGORITHM

作　　者：杜永兴[1] 牛丽静 秦岭 李宝山[1] Du Yongxing;Niu Lijing;Qin Ling;Li Baoshan(School of Information Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,Inner Mongolia,China)

机构地区：[1]内蒙古科技大学信息工程学院,内蒙古包头014010

出　　处：《计算机应用与软件》2021年第2期50-53,57,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61661044);内蒙古科技大学创新基金项目-优秀青年科学基金项目(2017YQL10);内蒙古自治区高等学校青年科技英才计划项目(NJYT-19-A15)。

摘　　要：传统的TF-IDF(Term Frequency&Inverse Documentation Frequency)算法提取的关键词不能合理地代表某疾病的症状,降低智能诊断系统的性能。对此,提出一种改进的TF-IDF算法,并将其应用在牛疾病诊断系统中。系统将用户描述的文本内容转换成向量的形式,用TF-IDF算法提取关键症状词,利用余弦定理和可信度计算给出可靠的疾病推荐和治疗方案。实验结果表明,该算法在疾病诊断中准确率和可信度两方面都具有更好的效果。与传统TF-IDF算法相比,平均可信度提高约4%。The of the keywords extracted by the traditional TF-IDF(Term Frequency&Inverse Documentation Frequency)algorithm can not reasonably represent the symptoms of disease,thus reducing the performance of intelligent diagnostic systems.In response to this situation,an improved TF-IDF algorithm is proposed and applied in the cattle disease diagnosis system.The system converted the text content described by the user into a vector form,extracted the key symptom words by TF-IDF algorithm,and used the cosine theorem and credibility calculation to give a reliable disease recommendations and treatment plans.The experimental results show that the algorithm has better effects in both disease accuracy and credibility.The average credibility is improved by about 4%compared with the traditional TF-IDF algorithm.

关 键 词：智能诊断 TF-IDF 余弦相似度 VSM 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]