有序加权平均算子用于情感调节素材的选择

发布时间：2018-08-27 12:28

【摘要】：根据Gross情感调节过程理论,情感调节主要是调节者通过情境选择、情境修正、注意分配、认知重评、表达抑制五个阶段对自己的不良情感进行自我调节,自我消化的过程。主要的调节策略有认知重评和表达抑制两种,研究表明认知重评相对于表达抑制更能有效地调节负面情感带来的不良影响。根据认知重评的调节策略,人们通过其它有正面影响的事件如看电影,玩游戏,听音乐等来转移对引起负面情感的事件的注意力,最终实现对负面情感的调节。而要实现这种调节,用户必须要花费大量的时间来寻找相应的调节素材如电影、音乐、图片等等。这对于一些时间很紧张的用户是很不方便的。特别是需要长时间面对计算机进行工作的用户,一方面他们极易因为长期劳累而产生焦虑、心烦等负面情感,影响他们的工作效率；另一方面又需要长期有效地工作,没有多余的时间去寻找调节素材。针对这些用户,建立一个能够根据用户的喜好和素材的特点,为用户选择合适的调节素材,从而实现对用户的负面情感进行调节的机制是非常重要的。基于以上分析,本文提出一种以计算机为媒介的情感调节方法,主要是建立一个情感调节素材评价、选择模型。此模型首先从主客观两方面对情感调节素材进行有效地评价,建立一个情感调节素材评价矩阵。然后根据用户对素材的评价运用一种多属性决策算法,为用户智能的从评价好的素材库中选择一个合适的调节素材,以帮助用户更好地调节负面情感。最后结合基于生理信号的情感识别的相关研究成果,建立一个集情感识别和情感调节为一体的情感调节系统。本文的主要研究内容如下： 1.建立情感调节素材库。基于人机交互的情感调节特点,从调节素材的情感导向、类型、时间长度以及呈现方式四个方面,选择35个调节素材建立了一个情感调节素材库。 2.建立素材属性评价矩阵。根据心理学素材评价方法-PAD模型和皮肤电生理信号的一阶差分,确定了评价情感调节素材的四维属性,建立了一个情感调节素材评价模型。并利用大样本统计方法建立了一个对35个情感调节素材的评价矩阵。 3.建立基于有序加权平均算子(OWA)的情感调节素材选择模型。提出利用一种多属性选择算法-有序加权平均算子,来完成调节素材的选择过程。详细介绍了OWA算子的概念和特性。最后建立了一个基于OWA算子的素材选择模型,并引出了模型中需要解决的情感调节素材属性的赋权问题。 4.为素材属性赋权,建立与素材属性对应的权值向量。根据已建立的基于OWA算子的素材选择模型,为素材属性赋权。通过深入研究基于层次分析的主观赋权和基于熵值法的客观赋权方法,分别从主观和客观两方面对素材属性进行赋权。然后利用基于离差平方和的组合赋权法和基于BP神经网络的交互式赋权方法,分别对基于主观和客观的赋权权值进行优化。最终确定了基于以上四种赋权方法所得到的针对35个调节素材评价矩阵的权值向量。 5.验证OWA算子在情感调节素材选择中的有效性。根据建立的素材评价矩阵和权值向量,运用OWA算子对各个素材的属性进行加权集结得出每个调节素材的OWA算子集结值,作为最终选择调节素材的决策值。通过Matlab仿真实验验证,针对于四种赋权方法,有序加权平均算子选择出调节素材都一样,并都能完全剔除没有调节效果的5个调节素材。从而验证OWA算子用于情感调节素材中的有效性。另外设计实验验证了基于BP神经网络的交互式赋权的OWA算子决策选择的调节素材的调节效果较优于基于离差平方和组合赋权的OWA算子决策选择的调节素材的调节效果。 6.建立一个基于人机交互的情感调节系统。基于生理信号情感识别技术和情感调节素材选择模型,构建了一个以计算机为媒介,能够实时监测用户当前的情感状态,并能通过提供一个用户喜欢的调节素材来帮助用户管理自己的情感,降低负面情感体验,增加正向情感体验的具有“情感支持”功能的情感调节系统。并详细说明了基于生理信号的情感识别的理论、过程、方法以及研究成果。
[Abstract]:According to Gross's theory of emotion regulation process, emotion regulation is a process of self-regulation and self-digestion through five stages: situational selection, situational modification, attention distribution, cognitive reappraisal and expression inhibition. According to the strategy of cognitive reappraisal, people use other positive events such as watching movies, playing games and listening to music to divert their attention from the events that cause negative emotions and eventually achieve the regulation of negative emotions. Users have to spend a lot of time looking for adjustable materials such as movies, music, pictures and so on. This is inconvenient for some time-stressed users. On the other hand, they need to work effectively for a long time without any extra time to find adjustment materials. For these users, it is very important to establish a mechanism to adjust the negative emotions of users by choosing suitable adjustment materials according to their preferences and characteristics. Based on the above analysis, this paper proposes a computer-mediated emotional regulation method, which mainly establishes a model for evaluating and selecting emotional regulation materials. The model first evaluates emotional regulation materials from both subjective and objective aspects, and then establishes an evaluation matrix for emotional regulation materials. A multi-attribute decision-making algorithm is used to select a suitable adjustment material for the user to help the user adjust the negative emotion better from the evaluated material library.
The main contents of this paper are as follows:
1. Establishment of emotion regulation material library. Based on the characteristics of emotion regulation in human-computer interaction, an emotion regulation material library was established by choosing 35 regulation materials from four aspects: emotion orientation, type, time length and presentation mode.
2. Establish the Matrix of Material Attribute Assessment. According to the PAD model and the first-order difference of skin electrophysiological signals, determine the four-dimensional attributes of emotional regulation materials, establish an evaluation model of emotional regulation materials, and establish an evaluation matrix of 35 emotional regulation materials by large sample statistical method.
3. Establish the model of emotion regulation material selection based on OWA operator. Propose a multi-attribute selection algorithm-ordered weighted average operator to complete the selection process of emotion regulation material. Introduce the concept and characteristics of OWA operator in detail. The problem of empowerment of emotional adjustment material attribute needs to be solved.
4. To assign weight to material attributes, the weight vectors corresponding to material attributes are established. According to the established OWA operator-based material selection model, the material attributes are assigned weight. Then, the combination weighting method based on the sum of squares of deviations and the interactive weighting method based on BP neural network are used to optimize the subjective and objective weighting values respectively.
5. Verify the validity of OWA operator in the selection of emotional adjustment materials. According to the evaluation matrix and weight vector, the OWA operator is used to weighted aggregate the attributes of each material to get the aggregation value of OWA operator for each adjustment material, which is used as the final decision-making value for the selection of adjustment materials. In order to validate the validity of OWA operator in affective adjustment materials, an ordered weighted averaging operator is proposed to select the same adjusting materials, and all the five adjusting materials with no adjusting effect can be eliminated completely. The adjustment effect is better than that of the adjustment material selected by OWA operator decision-making based on the sum of squares of deviation and combination weighting.
6. Establish an emotion regulation system based on human-computer interaction. Based on physiological signal emotion recognition technology and emotion regulation material selection model, construct a computer-mediated, real-time monitoring of user's current emotional state, and provide a user's favorite adjustment material to help users manage their emotions, reduce The emotion regulation system with the function of "emotional support" with low negative emotional experience and positive emotional experience is added. The theory, process, method and research results of emotion recognition based on physiological signals are explained in detail.
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.4;B842

【相似文献】