基于“学评融合”的评价新理念，评价活动是当前教育评价改革背景下促进学生核心素养发展的重要方式，随着在线学习的不断发展，网络学习空间中的评他活动也得到了广泛应用。在线平台中小学生生成的评他文本数量大且具备过程性数据的特点，但并未得到充分的挖掘与分析，而评他文本是反映小学生评他能力的直观体现。因此，本研究基于“教客网”平台中产生的评他文本语料，对小学生评他文本的评价行为序列进行研究，探究小学生的评价行为情况，提供相关建议。本文主要的三个研究内容如下：

（1）评他文本写作现状及发展趋势探索。对评他文本进行行为序列编码，以此作为本研究的数据基础，并进行描述性统计分析。结果表明，学生参与评他活动持续性较低，评价行为丰富度不足，但评价行为序列长度随参与持续性有所增加。出现最多的评价行为是对他人作品表达赞美、支持或鼓励的积极情感，其次是关注到字词标点的运用。

（2）评他文本的评价行为类型探索。运用SGT算法，将评价行为序列转换为特征向量，运用k-means算法，将评他文本的评价行为类型分为7类，分别命名为“简洁表达类”“字词支持阐释类”“句子阐释类”“赞美字词类”“修改条理类”“支持总结类”“单一评价类”。通过滞后序列分析，发现了7种评他文本的评价行为序列存在较为显著的差异，体现了评他文本中多元的评价行为模式。

（3）学期水平的评价行为类型及变化趋势研究。运用潜在剖面分析，将学期水平的评价行为类型分为5类，分别命名为“一致赞美字词型”、“简短评价型”、“简洁修改型”、“字词句质量阐释型”、“字词句质量赞美型”。并通过滞后序列分析对学期水平的评价行为类型的转换情况进行研究，发现小学生评价行为类型在学期间保持一致类型的情况较多，只有部分字词句质量赞美型在下一学期能提升至字词句质量阐释型。

最后，依据研究结果提出建议。通过推进评价新理念的深化推广与应用，深化对评价标准与评价过程的指导，强化网络学习空间中评他模块的开发及使用，进而提升小学生核心素养。

Based on the new evaluation concept of "integration of learning and evaluation", evaluation activities are an important way to promote the development of students' core literacy in the context of current education evaluation reform. With the continuous development of online learning, evaluation activities in the online learning space have also been widely applied. The number of peer review texts generated by primary school students on online platforms is large and has the characteristics of process data. However, the online peer evaluation text of primary school students has not been fully explored and analyzed, and the evaluation text is an intuitive reflection of their ability to evaluate others. Therefore, this study will based on the corpus of evaluation texts generated on the "Jiaoke Network". The evaluation behavior sequence reflected in the text of primary school students' evaluation of others is studied, and based on the current evaluation behavior situation of primary school students and finally provide relevant suggestions. The three main research contents of this article are as follows:

    (1) The current situation and development trend of peer evaluation text . Conduct behavioral sequence coding on the review text as the data basis for this study, and conduct descriptive statistical analysis. The results indicate that the duration of students' participation in evaluative activities is low, and the richness of evaluative behavior is insufficient. However, the length of evaluative behavior sequence increases with the duration of participation. The most common evaluative behavior is to express positive emotions of praise, support, or encouragement towards others' works, followed by attention to the use of word punctuation.

    (2) Explore the type of the peer evaluation text. Using the SGT(Sequence graph transform) algorithm, the evaluation behavior sequence is transformed into feature vectors. Using the k-means algorithm, the evaluation behavior types of evaluation texts are divided into 7 categories, named "concise expression type", "word support and interpretation type", "sentence interpretation type", "praise word type", "modification organization type", "support summary type", and "single evaluation type". Through lag sequence analysis, it was found that there were significant differences in the evaluation behavior of seven types of evaluative texts, reflecting the diverse evaluation behavior patterns in evaluative texts.

    (3)Explore the development trend of evaluative behavior types at the semester level. Latent profiles analysis(LPA) was used to divide the peer review behavior types at the semester level into five categories, named "single praise word type", "short evaluation type", "concise modification type", "quality explanation type of word sentence", and "quality praise type of word sentence". And through lag sequence analysis, it was found that there were more cases of primary school students maintaining consistent types of evaluative behavior at the semester level, with only some words and sentences of praise type being able to improve to words and sentences of explanation type in the next semester.

    Finally, suggestions are made based on the research results. By promoting the deepening promotion and application of new evaluation concepts, deepening guidance on evaluation standards and evaluation processes, strengthening the development and use of evaluation modules in online learning spaces, and thereby enhancing the core literacy of primary school students.