为顺应国家政策和服务社会需要，教师的角色发生了较大的变化。教师不仅仅作为知识的传授者，还需要承担培养学生高阶思维能力、创新能力、鼓励个性化发展等多元任务，不仅要精于学科知识技能和教学技能，还必须进行多种资源整合、保持终身学习以有能力应对教学情境中的各种复杂问题。在这样的时代背景下，教育供给更加全面均衡，教师角色趋于多样化。为了提升教师培养和持续发展水平、加速教育评价改革，需要对新时代教师素养展开新的认识和思考，对其评价方式展开新的探索和研究。本文通过深入剖析和探讨教师职业认同与教师胜任力两个关键方面，系统性地研究新时代教师素养的评价模型与方法。具体研究内容包括：

首先，以教师职业认同为例对新时代教师素养评价研究进行了深入剖析和探讨。对于教师职业认同评价，通过对现有文献的总结和归纳，提出了包含4个典型维度的教师职业认同模型及规范化测评工具质量验证流程，针对教师素养评价过程中社会称许性影响的问题，从真实情境问题、开放式作答、人工智能技术应用等视角提出未来展望。

其次，针对开放式情境测验这一测评新时代教师素养的重要工具，以教师胜任力测评为例，探索了开放式情境测验自动化评分的实现与应用。针对教学中典型问题场景开发了开放式情境判断测验，收集中小学教师作答文本，采用有监督学习策略分别从文档层面和句子层面应用深度神经网络识别作答类别，各题评分准确率理想，与人类评分一致性高，证明了机器自动化评分在开放式情境测验的可用性。

然后，在以ChatGPT为代表的生成式大语言模型普及的背景下，针对如何构建开放式情境测验自动化评分方法的研究问题，探索了应用大语言模型技术实现开放式情境测验自动化评分的可行性。结果表明，以ChatGPT为代表的大语言模型，通过恰当的提示信息，能够进行准确的自动化评分，并取得良好的信效度，证明了大语言模型评分在开放式情境测验的可用性。

最后，面向教师素养测评任务，基于教师胜任力开放式情境测验题目，研发教学场景支撑的交互式情境测验，能够有效完成测评场景中多模态反应数据收集。针对未来研究中存在的挑战，提出了跨情境多模态数据表征方法以及多模态数据驱动的教师胜任力智能评分方法构建思路，为实现多模态数据驱动的自动化评分模型，获取多维度、可解释的胜任力评价结果提供了系统性的解决方案。

In response to national policies and societal needs, the role of educators has undergone significant changes. Teachers are now not only knowledge transmitters but also bear the responsibility of fostering students' higher-order thinking skills, innovation, and encouraging personalized development. They are required to excel not only in subject matter and teaching skills but also in integrating diverse resources and engaging in lifelong learning to effectively address the complexities of teaching situations. In this era of comprehensive and balanced educational offerings, the role of teachers has become more diversified. To enhance teacher education and continuous development and expedite educational assessment reform, there is a need for a fresh understanding and exploration of the evaluation methods for the literacyof teachers in the new era. This paper systematically investigates the assessment models and methods for the literacy of teachers in the new era by delving into and exploring two crucial aspects: teacher professional identity and teacher competence.

Firstly, taking teacher professional identity as an example, this research provides an in-depth analysis and discussion of the evaluation of teachers' competence in the new era. For the assessment of teacher professional identity, a model incorporating four typical dimensions and a standardized evaluation tool quality verification process are proposed based on a summary and induction of existing literature. To address the issue of social desirability bias in the evaluation of teacher competence, future prospects are suggested from perspectives such as real-life scenarios, open-ended responses, and the application of artificial intelligence technologies.

Secondly, focusing on the importance of open-ended scenario-based assessments for evaluating the competence of teachers in the new era, the paper explores the implementation and application of automated scoring in open-ended scenario tests using teacher competence assessment as an example. Developed open-ended scenario judgment tests for typical teaching scenarios at the elementary and middle school levels were used to collect teacher response texts. The application of supervised learning strategies using deep neural networks for identifying response categories at both the document and sentence levels demonstrated ideal scoring accuracy, high consistency with human scoring, proving the utility of machine automated scoring in open-ended scenario tests.

Subsequently, in the context of the widespread adoption of generative large language models like ChatGPT, the study investigates the feasibility of applying these models to automate the scoring of open-ended scenario tests. The results indicate that large language models, represented by ChatGPT, can accurately score open-ended scenario tests with proper prompt information, achieving good reliability and validity, establishing the applicability of large language model scoring in open-ended scenario tests.

Lastly, for teacher competence assessment tasks, an interactive scenario test based on open-ended questions about teacher competence was developed, supported by teaching scenarios. This test effectively collects multi-modal reaction data in assessment scenarios. Addressing challenges in future research, the paper proposes a systematic solution involving cross-scenario multi-modal data representation methods and a construction approach for a multi-modal data-driven intelligent scoring model for teacher competence. This offers a systematic solution for achieving multi-dimensional, interpretable competence assessment results.