基于可能性条件偏好网络的交互式遗传算法及其应用

发布时间：2018-03-03 06:58

本文选题：交互式遗传算法　切入点：不确定性　出处：《郑州大学学报(工学版)》2017年06期 　论文类型：期刊论文

【摘要】：根据用户实施的人机交互行为而隐式地获取用户偏好的交互式进化优化算法,可有效减轻用户疲劳,提高个性化搜索或推荐的效率.但是,已有研究没有考虑用户交互行为和偏好的不确定性,影响了对用户偏好的拟合精度以及基于该偏好表达的进化搜索.针对该问题,提出基于可能性条件偏好网络的交互式遗传算法,以刻画用户交互行为和偏好的不确定性,并提高算法的搜索性能.首先,采用交互时间表示交互行为,考虑交互行为的不确定性,给出交互时间可信度的定义,并基于该定义给出了用户不确定偏好的表达函数;其次,利用可信交互时间和偏好函数,定义了用户对评价对象的偏好权重,并利用该权重,设计(更新)可以定量表示用户不确定偏好的可能性条件偏好网络,以更好地拟合用户偏好;然后,结合评价不确定性和可能性条件偏好网络,提出了改进的个体适应值估计策略,以更好地引导搜索;最后,将所提算法应用于图书个性化搜索中,结果表明了算法搜索的可靠性和高效性.
[Abstract]:The interactive evolutionary optimization algorithm, which acquires user preferences implicitly according to the user's human-computer interaction behavior, can effectively reduce user fatigue and improve the efficiency of personalized search or recommendation. Previous studies have not considered the uncertainty of user interaction behavior and preference, which has affected the fitting accuracy of user preference and the evolutionary search based on the preference expression. An interactive genetic algorithm based on possibility-conditional preference network is proposed to describe the uncertainty of user's interaction behavior and preference, and to improve the search performance of the algorithm. Firstly, interactive time is used to represent the interaction behavior. Considering the uncertainty of interaction behavior, the definition of interactive time credibility is given, and the expression function of user's uncertain preference is given based on the definition. Secondly, the trusted interaction time and preference function are used. This paper defines the preference weight of user to the object of evaluation, and designs (updates) a preference network which can quantitatively represent the possibility condition of user's uncertain preference, so as to better fit the user's preference. Combined with the preference network of evaluation uncertainty and possibility condition, an improved individual fitness estimation strategy is proposed to guide the search better. Finally, the proposed algorithm is applied to personalized book search. The results show that the algorithm is reliable and efficient.
【作者单位】：中国矿业大学信息与控制工程学院;
【基金】：国家自然科学基金资助项目(61473298)
【分类号】：TP18

【相似文献】