随着信息化、数字化程度的不断提高以及互联网、物联网和云计算技术的迅猛发展，人们生产数据、获取数据的能力越来越强，数据的量不仅以指数的形式递增，而且这种增长趋势已经突破了摩尔定律。不但如此，数据在类型、维度、速度、完备性等方面都有了突出的变化，而技术的发展又使得对数据的获取、存储、传输、处理的速度和自动化程度也大大增加，这就赋予“数据”以更加深层的内涵——即“大数据”。辩证唯物主义认为：“量变会引发质变”，这一质变所引发的效应在众多领域都已经显现，比如社会风险、疾病预防、经济预测等复杂领域，在大数据的框架下能得到有效的处理。正因如此，大数据已经在企业、政府和学术等各个领域，都被予以足够的重视，大数据业已成为这个时代重要的资源。 从大量成功的大数据研究案例来看，大数据不光改变了科学研究的方式、方法，而且还改变了科学研究的对象。最为关键的是，大数据也有能力教会人们如何理解自然世界、人类社会和生活，即“它为人类提供了一种全新的认识论和方法论，以便人类更为有效地理解世界与自身。”当大数据在各领域攻城略地、开疆拓土的时候，人们很容易陷入对大数据的无限崇拜。这种崇拜是在工具主义与实用主义思想的裹挟下的必然产物。对大数据过度崇拜的最直接后果是对长期以来人们所追求的事物发生、发展的因果、机制机理等人类理性的怀疑，同时也伴随着对理论驱动、问题驱动的研究方式以及对理论的价值的否定。 在此背景下，科学哲学作为审视和反思科学与技术的利剑，理清大数据的认识论和方法论等哲学问题，将对于人们理解大数据、使这项技术更好地、更全面地发展具有积极意义。本论文将以中立的立场与视角进入大数据这一研究对象，结合具体案例，对大数据认识论问题展开研究，并且试图为大数据建立一种可以接受的认识论方案。通过研究，我们希望达成两方面的目的：第一，明晰大数据认识论中的基本问题，如大数据是如何认识世界的？它与经典的方法，尤其是小数据的方法、经验科学的方法有什么异同？大数据是否构成了库恩意义上的范式革命？大数据背景下如何重新理解因果与相关的关系以及理论的价值等。第二，由于大数据涉及对经验的数据化，因此，经验、经验主义是大数据认识论绕不开的话题。尤其是近几年，由于大数据与深度学习结合而引发的人工智能领域的巨大进展，深刻地揭示了机器、机器的“经验”在大数据认识论中的地位。因此，本文以“大数据的认识论研究”为题，并辅以“从人的经验到机器‘经验’”的副标题，目的就是要为大数据建立一种可靠的认识论方案，由于这种认识论是以机器为认识主体的，并且强机器“经验”的基础地位，所以其本质上是一种非人类中心的认识论。

With the continuous improvement of informatization and digitalization, as well as the rapid development of Internet, Internet of things（IoE）and Cloud Computing technology, People's ability to produce and acquire data is growing, and the Volume of data is not only exponentially increasing, but this increase has already broken Moore's Law. Furthermore, the data has outstanding changes in terms of Dimension, Velocity, and Completeness. In addition, along with advances in technology, the speed and automation of acquiring, storing, transmitting, and processing data has also increased greatly, which has endowed "data" with a deeper meaning which has given “data” a deeper meaning, namely “big data”. For quantitative accumulation leads to qualitative transformation, now, complex areas such as social risk prevention and control, Flu prediction, and economic forecasting etc., can be effectively processed under the framework of big data. For this reason, big data has gained enough attention in various fields such as business, government, and academia. Obviously, big data has become an important resource in this era. Combined with the big data research cases, we can see that big data not only changed the way and method, but also changed the object of science. Most importantly, big data also can tell people how to understand the natural world, human society and life. In other words, “Big Data analytics enables an entirely new epistemological approach for making sense of the world; rather than testing a theory by analyzing relevant data, new data analytics seek to gain insights ‘born from the data’”. It can be expected that with the widespread penetration of big data and the great success achieved in many areas，it is easy to breed blind worship of big data. This kind of worship is an inevitable outcome of the ideology of instrumentalism and pragmatism. The direct consequence is the suspicion of human rationality, such as the causality and mechanism of the occurrence and development of things that people have been pursuing for a long time. It is also accompanied by the doubt of the research methods of theory-driven and question-driven, even the denial of the value of the theory. Under this background, clarifying philosophical issues such as epistemology and methodology of big data, will be important for people to understand big data and make it more comprehensive to human. In this paper, with a neutral position and perspective, and combined with some specific cases of big data, some epistemological issues (as described above) of big data will be studied and an acceptable epistemological approach will be established for big data. And we hope to solve some problems as following: (1) Clarify the basic questions epistemology in big data, such as, how big data understands the world? How does it differ from classic methods, especially small data methods? Does big data constitute a paradigm revolution? How to re-understand causality and correlation as well as the value of theory? (2) Big data involves translating the form of experience into data, and mainly used for machine-learning, therefore, based on experience and machine-learning, constructing a reliable solution of epistemology which is essentially a non-anthropocentric for big data is the focus of this paper.