张 旗.演绎、归纳、证伪和大数据:科学研究的方法论[J].甘肃地质,2021,(4):1-15
演绎、归纳、证伪和大数据:科学研究的方法论
Deduction,Induction, Falsification and Big Data Method: Methodology of Scientific Research
  
DOI:
中文关键词:  演绎  归纳  证伪  大数据  科学研究  方法论
英文关键词:deduction method  induction method  falsification method  big data method  scientific research  methodology
基金项目:中国科学院地质与地球物理研究所岩石圈演化国家重点实验室项目《镁铁—超镁铁岩大数据研究》(81300001)资助的研究
作者单位
张 旗 中国科学院地质与地球物理研究所北京 100029 
摘要点击次数: 401
全文下载次数: 289
中文摘要:
      通常认为,人们认识世界,进行科学研究有两种方法,即归纳法和演绎法。实际上应当有4种方法,另外两种是最近才崭露头角的,一种是证伪法(不到100年的历史),另一种是大数据法(几十年的历史)。归纳是从特殊到一般,演绎是从一般到特殊,证伪是批判,大数据是第四科学范式。休谟和波普尔都对归纳法提出质疑,指出归纳法不可能从特殊到一般。为此,波普尔提出证伪的方法和知识增长的四段图式,波普尔认为,证伪是知识增长的核心,没有经过证伪的理论只是猜测,只有经过证伪的检验,猜测才可能具有理论的属性。归纳法有问题,但是,人们又须臾离不开它。波普尔否定了归纳法,但是,没有给出解决的方案。本文经过思考认为,采用“归纳 + 大数据”的方法可以解决这个问题。既然特殊不可能达到一般,那就想办法使特殊变为一般。方法是加上大数据:“归纳 + 大数据”,可能就解决了归纳法的天然不足,挽救了归纳法,使归纳法换发出青春。   科学研究遇到的瓶颈告诉我们,大数据不是可有可无的。采用“归纳 + 大数据”方法,即可使归纳法的“特殊性”加上大数据方法的“全局性”从而达到一般。用这种互证互补的效应既避免了归纳法的局限性,又可以加速科学研究进程。本文指出,我们现在的科学研究方法基本上采用的是实证主义的方法,这种状况急需改变。在当今的量子时代、现代科学阶段,大数据方法已经应用于很多领域,本文提出在科学研究中强调采用证伪的方法,增加大数据(尤其全数据)方法,是新的科学研究思路,也是科学探索道路上新的尝试。
英文摘要:
      It is generally believed that there are two methods for people to understand the world and conduct scientific research, namely, induction and deduction. In fact, there should be four methods, the other two are only recently emerging, one is the falsification method (less than 100 years of history), and the other is the big data method (decades of history). Induction is from special to general, and deduction is from general to special. Falsification is criticism, and big data is the fourth scientific paradigm. Both Hume and Popper questioned the inductive method, pointing out that the inductive method cannot be from special to general. For this reason, Popper proposed the method of falsification and the four-stage schema of knowledge growth. Popper believes that falsification is the core of knowledge growth. Theories that have not been falsified are only guesses. Only after the test of falsification can have the attribute of theory. There is a problem with induction, but people cannot live without it. Popper rejected the inductive method, but did not give a solution. This article argues that this problem can be solved by adopting the method of “induction + big data”. It is impossible for the special to reach the general, so try to make the special become the general. The method is to add big data. “Induction + big data” may solve the natural shortcomings of induction, save induction, and make induction change its youth.   The bottleneck encountered in scientific research tells us that big data is not dispensable. In quantum age, in the modern science stage, researchers must master preliminary big data research methods in order to meet the needs of scientific research. Big data is not mysterious, nor so inscrutable. Adopting the “induction + big data” method can not only avoid errors in scientific research conclusions, but also speed up the process of scientific research, which is a fast, better, and more economical research method. This article emphasizes the application of the full data model, but in theory, full data is difficult to achieve, because the world is uncertain, and full data is all under certain conditions, not all in an absolute sense. This article points out that what we are currently adopting is basically a positivist research method, and this situation needs to be changed urgently. The use of falsification methods in scientific research and the addition of big data (especially full data) methods are the current important tasks of the academic community, otherwise it will be difficult to undertake the tasks given to Chinese scientists by history.
查看全文  查看/发表评论  下载PDF阅读器
关闭