整合复杂性建模及其与自杀之间的关系
文献类型：学位论文
| 作者 | 李东启 |
| 答辩日期 | 2024-06 |
| 文献子类 | 硕士 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院心理研究所 |
| 其他责任者 | 朱廷劭 ; 刘晓倩 |
| 关键词 | 整合复杂性 神经网络 大语言模型 迁移学习 自杀 |
| 学位名称 | 理学硕士 |
| 学位专业 | 应用心理学 |
| 其他题名 | Integrative Complexity Modeling and its Relationship to Suicide |
| 中文摘要 | Integrative complexity is a concept used in psychology to measure the structure of an individual's thinking. It involves two main aspects: differentiation and integration. Differentiation refers to an individual's ability to recognize and understand the different perspectives or elements present in a message; integration refers to an individual's ability to combine these different perspectives or elements into a logical and coherent whole. Integration complexity is measured by manually analyzing the content of texts. Although measures of integration complexity provide a powerful tool for understanding the cognitive complexity of individuals, there are some limitations to these measures. First, integrative complexity, even with detailed scoring criteria, the scoring process still involves subjective judgments, and agreement may vary across raters. In addition, the process of integrative complexity analysis is complex and time-consuming, especially for large or lengthy textual materials, which may limit the potential for largescale application. Finally, the scoring system and process may be influenced by specific cultural and linguistic contexts, and additional adjustments may be required for applications with non-English texts. Integrative complexity is currently demonstrating its interdisciplinary value and broad research potential in the fields of managerial psychology, political psychology, and cultural psychology. In the field of managerial psychology, leaders' levels of integrative complexity influence how they deal with complex managerial challenges, develop strategies, and promote team diversity. In political psychology, researchers use integrative complexity to analyze the thinking styles of political leaders, foreign policy decision-making processes, and the political attitudes and behaviors of the masses. Cultural psychology uses integrative complexity to explore the thinking patterns and information processing strategies of individuals in different cultural contexts. However, integrative complexity has not been well studied in the field of health psychology. Integrative complexity, as a measure of thinking structure, can shed some light on how individuals process information and apply stress and negative emotions, which is very important for their mental health. According to the suicide avoidance theory, individuals may escape from intolerable self-consciousness and emotional pain through suicidal behaviors, and within this theoretical framework, low integrative complexity may be a risk factor for suicidal behaviors because lower integrative complexity may lead to feelings of helplessness and despair as individuals have difficulty in seeing multiple facets of the problem and possible solutions in the face of stress and psychological pain. Aiming at the problems of high cost of manual integration complexity assessment methods, low accuracy of automated assessment methods, and the lack of Chinese text assessment schemes, this study designs an automated assessment scheme for Chinese and English texts based on the text data enhancement technique of the large language model and the model migration technique, and explores the automated assessment methods for two sub-structures of integration complexity: the fine integration complexity and the dialectical integration complexity. In addition, this study explored the role of integrative complexity on individuals' suicidal ideation and suicidal behavior based on suicide avoidance theory. Three studies were designed and implemented in this paper, firstly, a predictive model of integrative complexity for English text based on a large language model text data enhancement technique, secondly, a predictive model of integrative complexity for Chinese text based on a model migration technique, and finally, the role of integrative complexity on suicidal ideation and suicidal behavior was explored on social media data. The results of the study show that: In this study, it was found that the neural network model can well capture the English language features of integration complexity, dialectical integration complexity and elaborative integration complexity, which makes the neural network model outperform other machine learning models in terms of prediction performance for all three metrics on the English text dataset. The prediction performance of the neural network model was further improved by augmenting it with GPT text data. Through migration learning, it was found that the neural network in this study can be further used as a pre-trained language model for other languages, and the integration of the complexity prediction model with high accuracy for other language versions can be achieved through small-sample tuning. In addition, this study found that before an individual's suicidal behavior occurs, there is a significant decrease in integration complexity, dialectical integration complexity, and elaborative integration complexity, and dialectical integration complexity negatively moderates the relationship between an individual's negative emotions and suicidal ideation, but the positive moderating effect of fine integration complexity is only found in the English text, which is believed to be a possible result of individual cultural values in different cultures. |
| 英文摘要 | 整合复杂性是心理学中用来测量个体思维结构的一个概念。其主要涉及两个 方面：区分性和整合性。区分性是指个体能够识别和理解信息中存在的不同观点 或元素的能力；整合性是指个体能够将这些不同的观点或元素合并成一个有逻辑 性和连贯性的整体的能力。整合复杂性的测量主要依靠人工对于文本内容进行分 析。尽管整合复杂性的测量为理解个体的认知复杂性提供了有力的工具，但这些 测量方法也存在一些限制。首先，整合复杂性即使有详细的评分标准，评分过程 仍然涉及主观判断，不同评分者之间的一致性可能会存在差异。此外，整合复杂 性分析的过程复杂耗时，特别是对于大量或者长篇幅的文本资料，这可能限制了 大规模应用的可能性。 整合复杂性目前在管理心理学、政治心理学和文化心理学领域展现出了其跨 学科的价值和广泛的研究潜力。在管理心理学领域，领导者的整合复杂性水平影 响他们如何处理复杂的管理挑战、制定战略和促进团队多样性。在政治心理学中， 研究者利用整合复杂性来分析政治领袖的思维风格、外交政策决策过程以及群众 的政治态度和行为。文化心理学利用整合复杂性来探讨不同文化背景下个体的思 维模式和信息处理策略。但是在健康心理学领域，整合复杂性并没有被充分研究。 整合复杂性作为一种思维结构的测量方式，能够对个体如何处理信息和应用压力 以及负面情绪做出一定的解释，这对于个体的心理健康是非常重要的。根据自杀 逃避理论，个体可能通过自杀行为来逃避无法忍受的自我意识和情绪痛苦，这一 理论框架下，低整合复杂性可能是自杀行为的一个风险因素，因为较低的整合复 杂性可能导致个体在面对压力和心理痛苦时，难以看到问题的多个方面和可能的 解决方案，从而感到无助和绝望。 针对当前整合复杂性人工测评方法成本高、自动化评估方法精度低以及缺乏 中文文本评估方案等问题，本研究基于大语言模型文本数据增强技术和模型迁移 技术为整合复杂性的评估设计了对于中英文文本的自动化评估方案，并探索了整 合复杂性两种子结构：精细整合复杂性和辩证整合复杂性的自动化评估方法。此 外，本研究还基于自杀逃避理论探索了整合复杂性对个体自杀意念和自杀行为的 作用。本文设计并实施了三个研究，首先基于大语言模型文本数据增强技术实现 了对于英文文本整合复杂性的预测模型，其次基于模型迁移技术实现了对于中文 文本整合复杂性的预测模型，最后在社交网络媒体数据上探索了整合复杂性对于 自杀意念和自杀行为的作用。 本研究发现，神经网络模型可以很好地捕获整合复杂性、辩证整合复杂性和 精细整合复杂性的英文语言特征，这使得神经网络模型在英文文本数据集上对于这三种指标的预测性能上均优于其他机器学习模型。通过使用 GPT 文本数据增 强后，神经网络模型的预测性能得到了进一步提升。通过迁移学习发现，本研究 中的神经网络可以进一步作为其他语言的预训练语言模型，通过小样本调优实现 对于其他语言版本的高准确率整合复杂性预测模型。此外，本研究发现在个体自 杀行为发生之前，出现了整合复杂性、辩证整合复杂性和精细整合复杂性的显著 下降，且辩证整合复杂性负向调节个体负性情绪与自杀意念之间的关系，但是精 细整合复杂性的正向调节作用仅在英文文本中体现出来，本研究认为这种现象可 能是不同文化背景下个体文化价值观所导致的。 |
| 语种 | 中文 |
| 源URL | [http://ir.psych.ac.cn/handle/311026/47942]  |
| 专题 | 心理研究所_社会与工程心理学研究室 |
| 推荐引用方式 GB/T 7714 | 李东启. 整合复杂性建模及其与自杀之间的关系[D]. 中国科学院心理研究所. 中国科学院大学. 2024. |
入库方式： OAI收割
来源：心理研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。