文本描述
摘 要 对于所有公司来说,确切地了解消费者真正想要什么以及消费者将要做什么 的是一项希望和绝望的任务。许多公司在消费者调查中投入巨资,以了解消费者 对产品或广告的反应。然而,传统的研究技术,例如焦点小组访谈和调查,往往 无法提供与公司投入的成本和时间相关的令人满意的表现。消费者往往不会对调 查或访谈调查说实话,而且,这些调查必然会有局限性,因为它们旨在回答消费 者自己所感知的事实。事实上,在消费者调查中,不仅消费者表达的内容是一小 部分,而且消费者隐藏或不表达的内容也是一小部分。大部分就是连消费者自己 也不知道的无意识。最近,神经营销作为一种克服传统消费者研究极限的更科学 的研究技术而备受关注。神经营销作为神经科学和市场营销的复合词,通过使用 神经科学技术来测量消费者的大脑反应,并探索消费者的心理和行为机制,最后 将其应用于企业的营销战略和活动。随着不损害人体的大脑测量设备的出现,神 经营销研究正在被激活,并且正在将研究成果应用到企业的营销活动中。 本研究的目的是通过在基于脑电波的神经营销中构建消费者偏好预测系统来 预测和分析消费者偏好。消费者偏好研究可以分析消费者对营销活动的反应,例 如公司的产品、广告和服务。从公司的角度来看,消费者偏好研究可以提前分析 消费者对产品、广告和服务的反应,以降低市场不确定性的风险,并利用分析结 果来推动企业战略。从消费者偏好研究在市场营销中的重要性出发,本文研发了 一个可以回答“消费者无意识地喜欢什么?”的消费者偏好预测系统。 本文的具体内容如下。 首先,基于神经营销研究趋势和脑电波神经营销的理论分析,提出了脑电波 神经营销中的消费者偏好预测系统。在分析神经营销技术的基础上,综合分析了 近年来对消费者购买决策、消费者广告反应、消费者品牌态度、消费者促销策略 和消费者特性的神经营销研究。此外,在神经营销技术中,还选择了具有低成本、 易操作、无线连接等特点的脑电波技术,并且进行了基于脑电波技术的神经营销 研究的详细综述。由此,提出了一种由消费者脑电波预处理模块与消费者偏好预 测模块组成的新消费者偏好预测系统。 然后,提出了消费者脑电波中的眼电伪迹去除方法和消费者脑电地形图视频 方法,将其方法用于脑电信号预处理。为了去除消费者脑电信号中的眼电伪迹, 提出了一种结合独立成分分析和自适应滤波的新方法。此外,新定义了能够反映 消费者脑电信号的时间-频率-空间特性的消费者脑电地形图视频,并提出了一种 利用短时间傅里叶变换与双调和样条插值进行计算的方法。 其次,基于对消费者脑电地形图视频的分析,设计了一种消费者偏好分类预 测模型,该模型对于消费者偏好预测更加合适且更加稳健。在分析消费者脑电地 形图视频的不稳定和非线性特性的基础上,提出了一种结合卷积神经网络和长短 期记忆神经网络的消费者偏好预测模型。在预测模型中,特别地,新定义了用于 消费者脑电地形图视频处理的面膜层,并将其插入到卷积神经网络的卷积层和池 化层之间。接着从理论上讨论了由特征学习端和预测端组成的该模型适用于反映 消费者大脑活动时间、频率和空间信息的脑电地形视频处理。 最后,对于所提出的消费者偏好预测系统进行了性能评估与应用仿真。通过 确定预测模型的结构和参数,在消费者依赖模式和消费者独立模式中评估了消费 者偏好预测系统的性能。性能评估结果表明,在消费者依赖模式和消费者独立模 式下,所提出的消费者偏好预测系统稳定、稳健且适用于实践。此外,还对该系 统在产品、广告、品牌和产品设计等营销方面的应用进行了仿真。在应用仿真过 程中,预测并分析了消费者对产品、广告、品牌和产品设计的偏好,最总推出对 营销活动有用的结果。 本研究的重点是建构一个接近神经营销实际应用的系统。本研究将脑电波作 为一种神经营销技术。这是迄今为止神经营销技术中一种较便宜的技术,并且对 神经营销操作的要求很少。此外,在实际的营销环境中,还可以测量消费者的脑 电波。本研究提出脑电信号预处理与消费者偏好预测的方法。通过考虑到消费者 脑电信号的非线性和非稳定来提出的消费者脑电波预处理方法能够改善消费者偏 好预测系统的性能。消费者偏好预测方法是一种更适合于消费者脑电信号时间、 频率和空间特性的更适合的方法。本文提出的消费者偏好预测系统可以有效地应 用于神经营销研究以及实际的营销环境。由于神经营销研究需要大量的成本,迄 今为止,神经营销的应用研究主要是由大公司主导的。本文所提出的消费者偏好 预测系统不需要太多的成本,可以很容易地引入到中小规模的企业。综上所述, 本研究结果可更有效地且成本较低地应用于神经营销的研究和商业化。 如今,企业靠消费者正在逐渐成长。因此,消费者偏好分析研究是营销策略 研究和营销管理中非常重要的问题。所提出的消费者偏好预测系统及其应用方法 可以有效地应用于市场营销策略研发和市场营销管理中的消费者偏好分析。而且, 该系统基于低成本、低维护费用的脑电波技术,可以低成本地应用于大公司和中 小型公司。此外,可以无线测量消费者脑电波信号,从而可以通过接近营销环境 而不是实验环境来分析消费者们的偏好。另外,它也适用于以消费者偏好分析为 基础的神经营销研究。摘 要 关键词:神经营销;消费者偏好;脑电地形图视频;深度学习;三维卷积神经网 络;长短期记忆神经网络哈尔滨工业大学管理学博士学位论文 - IV - Abstract It is a hope and desperate task for all companies to understand exactly what consumers really want and what consumers are going to do. So many companies invest heavily in consumer surveys to understand how consumers response to products or advertisements. Traditional research techniques such as focus group interviews and surveys, however, often fail to deliver satisfactory performance relative to the cost and time the firm has invested. Consumers tend to not tell the truth about surveys or interview surveys, and these surveys have limitations because they are designed to answer only what consumers perceive themselves. In fact, in the consumer survey, not only the content expressed by the consumer is a small part, but also the content hidden or not expressed by the consumer is only a small part. Most of them are unconscious that even consumers do not know. Recently, neuromarketing has attracted attention as a more scientific research technique to overcome the limit of traditional consumer research. Neuromarketing, as a compound word of a neuro meaning neuroscience and a marketing, explores consumer psychology and behavioral mechanisms by using neuroscience technology to measure the brain response of consumers, and applies it to marketing. With the emergence of brain measurement devices that do not damage the human body, neuromarketing research is being activated and research results are being introduced into the marketing activities of companies. The purpose of this research is to predict and analyze consumer preference by constructing a consumer preference prediction system in neuromarketing based on electroencephalograph (EEG). Consumer preference researches allow companies to analyze consumer responses to marketing activities such as products, advertisements and services. From point of view of a company, consumer preference research can analyze the consumer response to products, advertising, and services in advance to reduce the risk of market uncertainty, and use the analysis results to drive corporate strategy. From the importance of consumer preference research in marketing, this thesis has designed a consumer preference prediction system that can answer What do consumers prefer unconsciously. The details of this thesis are as follows. First, based on the theoretical analysis of the neuromarketing research trend and the EEG based neuromarketing, the consumer preference prediction system in EEG based neuromarketing was proposed. Based on the analysis of neuromarketing techniques, recent neuromarketing researches on consumer purchase decision,Abstract - V - advertising, consumer brand attitude, consumer promotion strategy and consumer characteristics were analyzed comprehensively. In addition, neuromarketing based on EEG technique which has the characteristics of low cost, easy operation and wireless portability was selected in the neuromarketing techniques, and a more detailed review was given. From this, a new consumer preference prediction system consisting of consumer EEG preprocessing and consumer preference classifier or predictor was proposed. Next, as consumer EEG preprocessing methods, the method of removing ocular artifacts in EEG signals and the EEG topographic video method were proposed. In order to remove ocular artifacts which distort consumer EEG signals, the new method combining independent component analysis and adaptive filter was proposed. In addition, the consumer EEG topographic video which can reflect the time-frequency-space characteristics of consumer EEG signals was newly defined, and the method of calculating it using the short time Fourier transfo