伴随着当下生活节奏加快，人们更容易出现消极情绪以及无意义感，对身体各个方面产生不良影响。得益于互联网的广泛使用，大量的微博用户在平台上发布内容以记录当下的困惑和抒发内心的想法，这些自然发布的内容信息为心理学的研究提供了大量真实有效且可追溯的数据。与传统问卷调查的方式相比，微博文本记录了更加真实的用户情况。因此，本研究提出基于微博文本构建生命意义感检测模型，并探究生命意义感与睡眠的关系。

       研究 1 首先采用睡眠相关关键词爬取微博用户所发布的文本内容，爬取了超过 700 万睡眠相关微博的数据，经过一系列数据清洗处理后，构建了包含超过 300 万可用微博的数据集。通过建立生命意义感相关关键词列表，对该数据集进行筛选，构建睡眠_生命意义感子数据集，并招募 20 名心理学学生对随机抽取的微博文本进行生命意义感多个指标的交叉标注。基于预训练的中文BERT（Bidirectional Encoder Representation from Transformers）模型构建文本分类器，对微博是否与生命意义感相关进行判断；在此基础上构建子模型，分别从多维度（寻找意义感与拥有意义感）评估生命意义感的高低。

       研究 2 应用依存句法分析与语义依存分析技术，对微博文本中体现的个体睡眠情况进行检测与溯源。本研究建立了睡眠关键词、负面情绪词和程度词等相关词典，基于 LTP（Language Technology Platform）环境，对微博文本进行分词、词性标注处理。基于睡眠关键词典，构建睡眠相关语义依存路径图，并设计负面情绪和情绪累积程度在语义依存路径图中的权重算法，以量化单条微博文本中所体现的用户睡眠质量，根据睡眠质量得分对当前微博是否体现了睡眠困扰进行评估。此外，本研究尝试对微博文本中的语义角色和语义依存进行精准定位和溯源，抽取当前睡眠困扰存在紧密关联的时间信息以及影响用户睡眠质量的关键事件。
       研究 3 基于微博文本大数据集对用户的生命意义感进行自动化多维度评估，同时对睡眠质量进行溯源与结构化分析，并在此基础上分析生命意义感与睡眠的关系。结果表明生命意义感相关的微博文本与生命意义感不相关的微博文本相比睡眠分数更低，其所体现的睡眠质量更好；高水平寻找意义感体现的微博文本相较于低水平寻找意义感体现的微博文本睡眠分数更低，睡眠质量更好；高水平拥有意义感体现的微博文本相较于低水平拥有意义感体现的微博文本睡眠分数更低，睡眠质量更好。

    With the accelerated pace of life, people are more likely to have negative emotions and meaninglessness, which have adverse effects on all aspects of the body. Thanks to the widespread use of the Internet, a large number of Weibo users publish content on the platform to record the current confusion and express their inner thoughts, and these naturally released content information provides a large amount of real, valid and traceable data for psychological research. Compared with traditional questionnaires, Weibo texts record more realistic user situations. Therefore, this study proposes to construct a sense of meaning detection model based on microblog text, and explore the relationship between sense of meaning and sleep.

    Study 1：First used sleep-related keywords to crawl the text content posted by Weibo users, crawled the data of more than 7 million sleep-related microblogs, and after a series of data cleaning processing, constructed a dataset containing more than 3 million available microblogs. By establishing a list of keywords related to the sense of meaning of life, the dataset was screened, the sleep_sense of life sub-dataset was constructed, and 20 psychology students were recruited to cross-label the randomly selected Weibo text for multiple indicators of sense of meaning of life. Based on the pre-trained Chinese BERT (Bidirectional Encoder Representation from Transformers) model, a text classifier is constructed to determine whether Weibo is related to the sense of meaning of life. On this basis, a sub-model is constructed to evaluate the level of meaning of life from multiple dimensions (finding a sense of meaning and having a sense of meaning).

    Study 2: Application of dependency syntactic analysis and semantic dependency analysis techniques to detect and trace the individual sleep status reflected in Weibo text. In this study, a dictionary of sleep keywords, negative emotional words and degree words was established, and based on the LTP (Language Technology Platform) environment, word segmentation and part-of-speech annotation were processed for Weibo text. Based on the sleep keyword dictionary, a sleep-related semantic dependency pathway map is constructed, and a weight algorithm for negative emotions and emotion accumulation in the semantic dependency
pathway map is designed to quantify the user's sleep quality reflected in a single Weibo text, and evaluate whether the current Weibo reflects sleep distress according to the sleep quality score. In addition, this study attempts to accurately locate and trace the semantic roles and semantic dependencies in microblog texts, and extract the time information of the current sleep disturbance that is closely related and the key events affecting the user's sleep quality.

    Study 3: Based on the Weibo text big data set, the user's sense of meaning of life is automatically evaluated in multiple dimensions, and the traceability and structural analysis of sleep quality are carried out, and the relationship between sense of meaning of life and sleep is analyzed on this basis. The results showed that Weibo texts related to the sense of meaning of life had lower sleep scores than Weibo texts that were not related to the sense of meaning of life, and the sleep quality reflected was better. Compared with the low-level Weibo text that seeks the sense of meaning, the sleep score is lower and the sleep quality is better. Compared with Weibo text with a high level of meaning, the sleep score is lower and the sleep quality is better than that of the Weibo text with a low level of meaning.