Algorithmic Recommendation Empowers Core Values Education

Algorithmic recommendation is a technical form in which, supported by computer programming technology, computers analyze network users’ basic personal information, online behavior, and other data through computation and data processing to segment and precisely profile users’ interests and needs, thereby distributing and pushing information to targeted audiences with high frequency. Advancing the digitalization of education and making good use of algorithmic recommendation technology can not only precisely identify and promptly track newly emerging ideological and behavioral trends, providing support for ideological guidance and behavioral direction, but also promote the coordinated development of algorithmic recommendation and core socialist values education, offering new methods for the high-quality development of core socialist values education in terms of its models, scenarios, and driving forces.

算法推荐是在计算机编程技术支持下，计算机通过运算、数据处理等方式对网络用户的个人基本信息、网络行为等数据进行分众化和精准化剖析，推算出网民的兴趣点和需求，进而有针对性地高频分发推送信息的一种技术形态。推进教育数字化，用好算法推荐技术，不仅能够精准识别、及时追踪新发生的思想和行为动向，为思想引领和行为引导提供支撑，而且能够推动算法推荐和核心价值观教育协同发展，为核心价值观教育模式、场景、动能的高质量发展提供新的方法。

The Manifestations of Algorithmic Recommendation Empowering Core Values Education

算法推荐赋能核心价值观教育的表征

Personalized recommendation activates core values education resources. Personalized recommendation automatically predicts learners’ preferences based on their learning conditions (such as personal information, learning records, and geographic location) and pushes suitable core values education resources to them individually. The traditional one-to-many education model focuses on homogeneity in education, whereas in the intelligent era, core values education leverages the portability of mobile devices to return to a one-to-one model. Whether it is the authenticity of information content or the diversification of information channels, both provide strong support for intelligent recommendation technology to assist personalized recommendation in core values education. By utilizing the personalized recommendation function of algorithmic recommendation technology, educators can analyze learners’ individual characteristics, optimize the allocation of educational resources, and achieve cross-temporal and cross-spatial sharing of educational resources, causing a qualitative change in the subject of core values education and generating a new type of teacher-student relationship.

个性化推荐激活核心价值观教育资源。个性化推荐是根据受教育者的学习情况（如个人信息、学习记录、地理位置）自动预测其学习偏好，个性化推送其合适的核心价值观教育资源。传统一对多的教育模式关注教育的同质性，智能时代的核心价值观教育以移动设备的便携特点让教育回归一对一的模式。无论是信息内容的真实化，还是信息渠道的多样化，都能为智能推荐技术助力核心价值观教育个性化推荐提供有力支持。教育者借助算法推荐技术的个性化推荐功能，可以分析受教育者的个性、优化教育资源配置，并实现跨时空共享教育资源，使得核心价值观教育主体发生质的变化，生成新质的师生关系。

Data-driven recommendation empowers core values education decision-making. Data-driven recommendation is a method that, by establishing teaching models and learning modeling representations, creates model portraits of identity information, preference information, and learning information, and then pushes targeted educational resources. Traditional small-data education decision-making and management focus on data obtained in real life, whereas data-driven education decision-making in the intelligent era relies on big data technology to organize the correlations of massive amounts of data, thereby predicting the possible development of educational activities. Whether it is small-data decision-making or big-data decision-making, both are the foundation and prerequisite for intelligent technology to assist data-driven recommendation in core values education and are important means of improving educational quality. By utilizing the data-driven recommendation function of algorithmic recommendation technology, it is possible to comprehensively monitor learners’ learning behaviors, scientifically formulate education management policies, and deeply explore the characteristics of learner groups, causing significant changes in core values education decision-making and generating new types of learning materials.

数据化推荐赋能核心价值观教育决策。数据化推荐是通过建立教学模型与学习建模表征的方式，对身份信息、偏好信息和学习信息进行模型画像，再进行有针对性的教育资源推送的方法。传统小数据教育决策和管理关注现实生活中获得的数据，而智能时代数据驱动的教育决策依托大数据技术将海量数据的相关性进行整理，进而预测教育活动发展的可能。无论是小数据决策还是大数据决策，都是智能技术助力核心价值观教育数据化推荐的基础和前提，是提升教育质量的重要手段。借助算法推荐技术的数据化推荐功能，可以全面监测受教育者的学习行为，科学制定教育管理政策，深挖受教育者群体特性，使得核心价值观教育的决策发生显著变化，生成新质学习资料。

Dynamic recommendation optimizes core values education methods. Traditional passive learning path design focuses on the information provided by educators, whereas in the intelligent era, digital-native learners tend toward proactive optimization of learning paths, collecting, organizing, analyzing, and outputting learning materials through intelligent tools and online platforms to achieve self-growth and self-development. Whether it is traditional passive learning path design or current proactive learning path optimization, both are conditions for intelligent technology to assist dynamic recommendation in core values education and are important pathways for promoting the high-quality development of learners’ learning. By utilizing the dynamic recommendation function of algorithmic recommendation technology, it is possible to design learning paths tailored to learners’ needs, allowing learners to autonomously plan learning content and enabling educators to efficiently impart learning content. This can broaden the possible space for optimizing learners’ learning paths in core values education, make organization and content design more precise, and generate new types of learners.

动态化推荐优化核心价值观教育方式。传统被动性的学习路径设计关注教育者提供的信息，而智能时代网络原住民的学习者更倾向于主动性学习路径优化，通过智能工具和网络平台进行学习素材收集、整理、分析、输出，实现自我成长成才。无论是传统被动的学习路径设计，还是现在主动性的学习路径优化，都是智能技术助力核心价值观教育动态化推荐的条件，是推动受教育者学习高质量发展的重要途径。借助算法推荐技术的动态化推荐功能，为受教育者设计其所需要的学习路径，让受教育者自主规划学习内容，让教育者高效传授学习内容，可以使得核心价值观教育的受教育者学习路径优化可能空间更广，组织和内容设计更加精确，生成新质的学习者。

The Contradictions between Algorithmic Recommendation and Core Values Education

算法推荐与核心价值观教育的矛盾

The contradiction between algorithmic customization and the comprehensive development of learners. Algorithmic customization is the process by which algorithms, through analyzing learners’ data and preferences, determine the scope and content of education and carry out targeted recommendations. From the perspective of values education, algorithmic customization emphasizes meeting students’ short-term developmental needs, while paying insufficient attention to national development, talent cultivation, and economic and social development needs, which can easily lead to distorted individual cognition and polarized behavior. Algorithmic customization has individual or group specificity, and within a certain time frame, it is difficult for it to bear the educational goal requirements of people’s free and comprehensive development, causing educational subjects to fall into circles of information closure and self-isolation. Algorithmic customization easily attracts learners’ attention, but there is an inherent tension between this short-term comfort and the long-term nature of values formation. It is evident that three aspects—subject cognitive limitations, subject information closure, and subject learning fatigue—constitute the root causes of the contradiction between algorithmic customization and the comprehensive development of learners.

算法定制与受教育者全面发展的矛盾。算法定制是算法通过对受教育者的数据和偏好分析，进行教育内容和范围的确定，并进行定向推荐。从价值教育角度看，算法定制侧重对学生短期发展需求的满足，对国家发展、人才培养、经济社会发展的需要关注不足，容易导致个体认知失真、行为极化。算法定制具有个体针对性或群体针对性，在一定时间范围内难以承担对人的自由全面发展的教育目标要求，使得教育对象陷入信息封闭、自我封闭的圈子。算法定制容易吸引受教育者的注意力，但这种短期的舒适感与价值观生成的长期性存在着内在张力。可见，主体认知局限、主体信息封闭和主体学习倦怠三方面构成了算法定制与受教育者全面发展矛盾产生的根源。

The contradiction between algorithmic substitution and the exertion of learners’ creativity. In the field of core values education, influenced by the complexity of algorithmic technology itself and the commercial policies of educational institutions, educators and learners are often unaware of the goals and intentions of algorithms, and lack understanding of the subject responsibility and evaluation supervision of algorithmically generated content, leading to the phenomenon of educational algorithmic substitution. Humans develop and design algorithms, but the existence of the algorithmic black box leads to a dramatic human-machine paradox: “I can give an explanation, but you humans cannot understand it,” which prevents human ideas from being implemented, replacing people-centered action with technology-driven action. Algorithmic dependence also easily leads educators to blindly trust its rational thinking, reducing the role of thought and emotion, and resulting in educational behaviors that follow computer programs absolutely without considering consequences or reasons. It is evident that three aspects—technological substitution, lack of dialogue, and subject weakening—constitute the root causes of the contradiction between algorithmic substitution and the exertion of learners’ creativity.

算法替代与受教育者创造力发挥的矛盾。在核心价值观教育领域，受算法技术本身的复杂性以及教育相关机构的商业政策影响，教育者与受教育者对算法目标和意图并不知晓，对算法生成内容的主体责任和评判监督也缺乏认识，出现了教育算法替代现象。人类开发设计了算法，但算法黑箱的存在导致戏剧性的人机悖论出现，即“我能给出解释，但你们人类听不懂”，使得人类的想法并不能践行，以技行事代替了以人为本行事。算法依赖也容易使教育者盲目相信其思维理性，降低思想感情的作用，出现按照计算机程序绝对执行的不计后果、不问缘由的教育行为。可见，技术替代、对话缺失和主体弱化三方面构成算法替代与受教育者创造力发挥矛盾产生的根源。

The contradiction between algorithmic bias and educational justice. Algorithmic bias exists in the balance between algorithmic technology and the goals of values education. Algorithmic recommendation systems emphasize the individuality and diversity of data but neglect the value-oriented nature of data, thereby triggering a contradiction between technological embedding and educational values, affecting educational equity. In the screening and labeling work of technical controllers who train data content for core values education, unconscious errors or subtle deviations can cause changes in the internal reasoning mechanisms of algorithms, leading to deviations in the deductive trajectory of algorithmic models, with the direct consequence being abnormal behavior in the application of artificial intelligence in education. In the process of collecting, storing, and processing algorithmic data, there are difficulties in attributing responsibility to multiple subjects, including collectors, designers, users, and platform operators, such as potential risks of illegal misuse and privacy breaches. It is evident that three aspects—deviation in educational subject needs, harm to educational content, and difficulty in attributing responsibility for educational risks—constitute the root causes of the contradiction between algorithmic bias and educational justice.

算法偏见与教育公正的矛盾。算法偏见存在于算法技术与价值教育目标的平衡之中。算法推荐系统注重数据的个体性与多样性，却忽视了数据的价值主导性，进而引发技术嵌入与教育价值的矛盾，对教育公平性产生影响。在训练核心价值观教育数据内容的技术掌握者的筛选和标识工作中，无意识的失误或细微偏差都有可能致使算法内部推理机制发生变化，进而引起算法模型推理演绎的轨道偏移，其直接后果是人工智能教育运用出现行为异常。在算法数据的收集、存储和处理过程中，会面临采集者、设计者、使用者、平台方等多重主体责任归属的难题，比如非法滥用、隐私泄露等潜在风险。可见，教育主体需求偏离、教育内容危害和教育风险归责难三方面构成算法偏见与教育公正矛盾产生的根源。

Measures for Algorithmic Recommendation to Promote Core Values Education

算法推荐助推核心价值观教育的举措

Guide algorithms with value orientation to promote the connotative development of core values education. First, optimize the content of the algorithmic recommendation system for core values education. On the basis of existing algorithmic recommendation education services, continuously improve and optimize pushed content according to the physical and mental health needs of educational subjects. Thoroughly address the problem of induced addiction in the field of core values education and establish a mechanism to prevent “information cocoons.” [1]1 “Information cocoons” (信息茧房) refers to a phenomenon where algorithmic recommendation systems confine users to a narrow range of information that reinforces their existing views, limiting exposure to diverse perspectives. Integrate high-quality content such as Marxist theory to achieve resonance at the same frequency among value orientation, value direction, and value tendency. Combine the latest research findings and practical experiences in core values education to provide learners with richer and higher-quality core values education content, thereby ensuring the effectiveness of the algorithmic recommendation system. Second, embed values of upward and good conduct into algorithmic education services. In accordance with the requirements of the “Guiding Opinions on Strengthening the Comprehensive Governance of Internet Information Service Algorithms,” [2]2 The “Guiding Opinions on Strengthening the Comprehensive Governance of Internet Information Service Algorithms” (《关于加强互联网信息服务算法综合治理的指导意见》) is a Chinese government document issued to regulate the use of algorithms in internet information services, emphasizing security, fairness, and transparency. continuously optimize the algorithmic mechanism review for core values education and the management systems and technical measures for data security; the sources of algorithmic training data for core values education must be lawful, and by promptly detecting and repairing code security vulnerabilities and algorithmic logic defects, conduct regular security assessments of the usability, controllability, and explainability of core values education algorithmic models, as well as the links of data processing, model training, and deployment and operation.

以价值导向引领算法，助推核心价值观教育内涵式发展。一是优化核心价值观教育算法推荐系统的内容。在已有算法推荐的教育服务基础上，根据教育对象的身心健康需要持续完善和优化推送内容。深入整治核心价值观教育领域的诱导沉迷问题，构建“信息茧房”防范机制。融入马克思主义理论等优质内容，做到价值导向、价值取向、价值倾向的同频共振。结合最新的核心价值观教育研究成果和实践经验，为学习者提供更加丰富、优质的核心价值观教育内容，从而保证算法推荐系统有效。二是在算法教育服务中嵌入向上向善价值观。根据《关于加强互联网信息服务算法综合治理的指导意见》的要求，对核心价值观教育的算法机制审核、数据安全的管理制度和技术措施持续优化；核心价值观教育的算法训练数据来源要合法，通过及时检测修复代码安全漏洞和算法逻辑缺陷，对核心价值观教育算法模型的可用可控可释以及数据处理、模型训练、部署运行等环节开展定期安全评估。

Regulating algorithms with ethics and law promotes the human-centered development of core socialist values education. First, design and apply algorithms in accordance with ethical and legal requirements. Construct an algorithmic system that conforms to ethics and law, pushing values that align with public order and good customs, transforming the situation where people are trapped by algorithms into one where algorithms serve people. In the process of collecting, storing, and processing data from both sides of education, adhere to the principles of data security and privacy. Continuously enhance the detection and identification capabilities for synthetic educational information in core socialist values education, promptly detecting and removing harmful synthetic information generated in violation of laws and regulations. Second, design and apply algorithms following the rules of openness and transparency. The standardization of algorithmic procedures requires that algorithm model design focus on the comprehensiveness and completeness of the educational system, fully considering the diversity of educational types, orientations, and levels. For algorithmic recommendation services related to core socialist values education, appropriately disclose their purpose, intent, operational mechanisms, and basic principles, and ensure that educational algorithm recommendations are simple, clear, and explicit.

以伦理法律规约算法，助推核心价值观教育人本化发展。一是遵循伦理法律要求设计运用算法。构建起符合伦理法律的算法体系，推送符合公序良俗的价值观，化人被算法所困为让算法为人所用。在教育双方数据采集、存储和处理过程中，坚持数据的安全性和隐私性原则。持续提升核心价值观教育方面合成教育信息检测识别能力，对违法违规生成的有害合成信息要及时发现并清除。二是遵循公开透明规则设计运用算法。算法程序的规范需要算法模型设计聚焦教育系统的全面性、完整性，充分考虑教育类型、向度和层面的多样性。对于核心价值观教育相关算法推荐服务的目的意图、运行机制和基本原理要进行适当公开，并确保教育算法推荐的简单、清晰、明确。

Developing algorithms in accordance with educational laws promotes the high-quality development of core socialist values education. First, optimize the educational process to precisely recommend feasible and optional educational plans. Core socialist values education through algorithmic recommendation is a process of using algorithms to process adaptive content—such as knowledge, learning materials, texts, videos, and images—in the education strategy model and learner model, providing organized input for the learner’s processes of distinguishing, screening, learning, memorizing, and deciding. Platforms such as online open courses can recommend personalized content and resources based on learners’ study habits and thinking, promoting deep learning and autonomous learning, thereby shifting content supply from silent to vocal. Second, smooth algorithmic feedback to advance efficient educational evaluation. Core socialist values education must establish efficient algorithmic feedback channels and response mechanisms, identify weak links and learning bottlenecks in education, and accordingly adjust the allocation of educational resources and the priority of educational decisions. By formulating reliable and effective feedback indicators, track the development trends of learners’ skills such as creativity and curiosity, explore the most effective educational methods and contexts suited to learners. Utilize tools such as smartphones to allow learners to follow, click on, and listen to content pushed by smartphones in real time according to their own needs, thereby continuously optimizing the educational feedback system, and adjust educational strategies and content in a timely manner based on precise feedback data.

以教育规律发展算法，助推核心价值观教育高质量发展。一是优化教育过程，精准推荐可选可行的教育方案。算法推荐的核心价值观教育是运用算法处理设计的教育策略模型和学习者模型中的知识、学习素材、文本、视频、图片等自适应内容对受教育者的辨、筛、学、记、决进行有组织输入的过程。线上开放课程等平台可以根据学习者的学习习惯与思维，推荐个性化内容和资源，促进深度学习和自主学习，使内容供给由无声走向有声。二是畅通算法反馈，推进高效教育评价。核心价值观教育要建立高效的算法反馈渠道和响应机制，识别教育的薄弱环节和学习瓶颈，并据此调整教育资源配置和教育决策优先级。通过制定可靠有效的反馈指标，追踪受教育者创造力和好奇心等技能发展态势，探索最有效的教育方法以及契合受教育者的语境。利用智能手机等工具，让受教育者依据自身需求实时关注、点击、收听智能手机推送的内容，据此持续优化教育反馈系统，凭借精确反馈数据适时调整教育策略和教育内容。

(The author is a professor at the School of Marxism, South China University of Technology)

（作者系华南理工大学马克思主义学院教授）

Online editor: Huihui

网络编辑：慧慧