论文部分内容阅读
【目的】为降低专家检索过程中的噪声并提升用户满意度,提出构建用户主导下的专家检索可信度评测机制。【方法】在BIR模型基础上,阐述评测机制运行需要遵循的原则和假设,围绕专家特征设置参数,依次设计前后端可信度评测机制。【结果】以学术专家检索为例,说明后端可信度评测通过求解最佳专家特征向量目长来降低检索噪声,前端可信度评测将用户相关性反馈作为检索路径选择的必要参照。【局限】前端可信度评测不适用于用户提问较长的情形;后端可信度评测对专家信息组织方式要求高。【结论】综合两种可信度评测机制,该机制可提升专家检索关联资源的广度和用户参与的深度。
【Objective】 In order to reduce the noise in the process of expert retrieval and enhance the degree of user satisfaction, a mechanism for evaluating the credibility of expert retrieval under the construction of users is proposed. 【Method】 Based on the BIR model, the principles and assumptions to be followed in the operation of the evaluation mechanism are elaborated. The parameters are set around the characteristics of the experts and the evaluation mechanisms of the front and back end credibility are designed in turn. [Results] Taking the example of academic expert retrieval, it shows that the back-end credibility evaluation reduces the search noise by solving the best expert eigenvector length, and the front-end reliability evaluation takes the user relevance feedback as the necessary reference for the retrieval path selection. [Limitations] The front-end credibility evaluation does not apply to the user asking a longer case; the back-end credibility evaluation of the expert information organization requirements. 【Conclusion】 Two credibility evaluation mechanisms are integrated. This mechanism can enhance the breadth of experts to retrieve related resources and the depth of user participation.