| As an important part of agricultural information resources, agricultural knowledge resources are widely used in agricultural expert systems and other agricultural informational applications. The development of agricultural integrate and intelligent technologies results in the requirements of effective integration of agricultural knowledge resources. The infrastructure of agricultural informatization is insufficient and the existing agricultural information systems are short of unified standards and operational guidance, so the performance, reliability, availability and various attributes have significant differences. It is necessary to describe, quantify and evaluate the attributes effectively, offer the essential data for valid management of agricultural knowledge resources, and promote the effective sharing of these resources. This paper analyses the structure features and attribute characteristics of agricultural knowledge base, builds the evaluation index system of agricultural knowledge base based on multi-dimension QoS, brings up the comprehensive evaluation model and recommendation method, and develops a comprehensive evaluation system for agricultural knowledge base. The main points of this thesis are as follows:(1) By analysing the connotation and representation of agricultural knowledge, the application effect and evaluation methods of agricultural knowledge base, an evaluation index system came out. In the index system, three functional comprehensive indexes are knowledge quantity, renewable difficulty and user feedback of agricultural knowledge bases, and four non-functional comprehensive indexes are service performance, service reliability, service availability and service cost of agricultural knowledge bases, and there are 10 secondary indexes. Indexes’ quantization equations for agricultural knowledge base were also acquired in the paper.(2)The attributes and change features of QoS metrics were researched to define the suitable quality functions and complete the standardization of these indexes. A normalized description model was built based on the index system and standardization methods. By using the description model, entropy weights of QoS metrics were calculated and revised via user preference and a weight vector was obtained. Utilizing a description matrix and the weight vector, a modified TOPSIS algorithm was used to evaluate agricultural knowledge bases comprehensively. The model was used to calculate the evaluation ranking of 7 agricultural knowledge bases in Beijing agricultural data center. And the result was compared with other evaluation solutions.The comparison reveals that the accuracy rate is improved 8%~14% and 6% by using this method compared to using single dimension methods and the Domain-Adaptive QoS Evaluation Model, respectively. The results prove that the model is effective.(3) By the above-mentioned research, a comprehensive evaluation model of agricultural knowledge bases based on multi-dimension QoS was built. A set of the same location users was built by their position features. A neighbor set was established from the user set based on the similarity of users’ QoS preferences. Then a recommendation list was acquired by the neighbor set and the recommendation of agricultural knowledge bases based on former evaluation results was realized. The test in which the evaluation and recommendation model were used approved that the length of time of recommendation model reduced about 90% compared with the evaluation model, and they had similar accuracy rate. It proves that the recommendation method of agricultural knowledge base based on evaluation results is efficient.(4)A comprehensive evaluation system was developed by using the aforesaid evaluation index system, models and method. The system includes a QoS measuring module, a QoS parameter setting module, an evaluation module and a recommendation module of agricultural knowledge bases. A graphical user interface of these modules was designed and developed. A demonstration of system running was succeeded and it worked well. The system was used to evaluate 7 agricultural knowledge bases in Beijing agricultural data center, and the process contains index actual value acquiring, evaluation value calculating, user preference setting, ranking result calculating, user feedback and recommendation. It is used to erify that the evaluaton model and recommendation method carry out well to realize the comprehensive evaluation of agricultural knowledge bases.The paper researched the QoS index system, index quantification method, QoS description model, comprehensive evaluation model and recommendation algorithm. A prototype system for agricultural knowledge bases evaluation was developed and tesed well in Beijing agricultural data center. The paper offers theoretical basis and actual useful tools for integrated management and efficient sharing of agricultural knowledge bases. |
