Computational social science is a product of the big bet365 horse racing odds era，Currently computational social science is more understood as an emerging field that uses computational methods to conduct research on social science issues。February 2009，15 scholars headed by David Lazer published a perspective article titled "Computational Social Science" in Science，marks the birth of this field。10 years later，These scholars once again published articles in the Science Policy Forum，Reflecting on the shortcomings and problems in the field of computational social science research，At the same time make some suggestions。As a new research field，Self-reflection through “10 Years Looking Back”，Several problems facing the field of cleanup research，It is necessary for the future development of the subject。This article attempts to review the relevant empirical analysis of computational social science in the past ten years，Analysis of several controversies existing in the field of computational social science research methods，To present some of the current challenges faced by this field。

　　Very different from traditional small bet365 horse racing odds

Whether it is social science in a general sense，Or computational social science，The quantitative research is based on bet365 horse racing odds。Judging from the name，“Big bet365 horse racing odds” seems to emphasize greater quantity/scale than traditional quantitative bet365 horse racing odds，But the nature of the bet365 horse racing odds between the two、There are qualitative differences in the relationship with research。On the one hand，The properties of the two bet365 horse racing odds are completely different。In general social sciences，The bet365 horse racing odds is exponential bet365 horse racing odds；In the context of big bet365 horse racing odds，The complexity of bet365 horse racing odds types and structures is much higher，includes text、image、Video、Unstructured information such as audio，Also includes spatial location、Highly complex information such as complex sequences。On the other hand，From the relationship with research，Traditional quantitative bet365 horse racing odds is the use of experiments for specific research purposes、Results of planned observations using questionnaire surveys and other methods。In other words，The production of bet365 horse racing odds is itself part of the research。However，Due to the big bet365 horse racing odds era, bet365 horse racing odds production and social science research bet365 horse racing odds mining have become two relatively independent processes，The production relationship between research and bet365 horse racing odds has been changed or even inverted—from producing bet365 horse racing odds to mining bet365 horse racing odds。

Therefore，Compared to traditional small bet365 horse racing odds，Big bet365 horse racing odds doesn’t just mean more bet365 horse racing odds，Also means in the way of production、Form、Dimension、structure、Analysis method、The relationship with scientific research and other aspects are completely different。From a methodology perspective，Social science research based on big bet365 horse racing odds，Actually represents a new social science research paradigm。

　　Dialectical view of objectivity

Some scholars believe that，“Naturally occurring” is a basic attribute of social science big bet365 horse racing odds。Whatever digital traces on the Internet、Information bet365 horse racing odds in the Internet of Things，An email from an activist、Chat records, etc. are considered to occur naturally without intervention，Reflects the true state of the actor。Compare with that，bet365 horse racing odds collected by traditional research methods will be significantly affected by investigator intervention effects or self-report/social desirability bias effects。Except for IoT bet365 horse racing odds mining，The above view is also accepted in some empirical studies on social media bet365 horse racing odds。This group of scholars will start from Facebook、The bet365 horse racing odds collected by social networking sites such as Twitter is regarded as “traces” or “symptoms” of people’s actual behaviors and emotions，The website itself is considered a neutral service provider。This position presupposes that the relationship between bet365 horse racing odds and users is self-evident，Believe that individual (and group) behavior can be predicted by studying bet365 horse racing odds。In the existing big bet365 horse racing odds research field，The prediction function is given absolute priority，To solve some social problems by predicting political and social phenomena、Political and Business Issues。Relatively speaking，The function of explanation is placed in a relatively secondary position，Even ignored。

However, some scholars have criticized the "spontaneous generation theory"，RuJ. Dijck (2014) points out that this position implies such a belief，That is, quantification is objective，It is possible to track everyone’s behavior through online bet365 horse racing odds，(Meta)bet365 horse racing odds is the “raw material” that can be analyzed and processed into predictive algorithms about future human behavior。However，In the author’s opinion，This concept cannot be established at both the ontological and methodological levels。At the ontological level，The bet365 horse racing odds objectivity claimed by this concept does not exist。For example，Facebook and Twitter keep debugging，Friendship、Convert popularity, etc. into some algorithm，Also calls this algorithm “social” values。Site buttons like “Like” and “Top Topics” may be considered natural online social activities，But the algorithms that make up these buttons are carefully crafted to guide people to click in response。At the methodological level，Analysis and interpretation of bet365 horse racing odds “raw materials”，Will definitely focus on a certain issue，And use some kind of analysis tool。Therefore，To make bet365 horse racing odds patterns meaningful，We need to critically ask: whose interests are we serving、Why to find massive metadata。

The author thinks，The objectivity of big bet365 horse racing odds can be viewed in two ways。On the one hand，From the perspective of the production process，relative to small bet365 horse racing odds，Big bet365 horse racing odds is the huge amount of information generated in the digitalization process of social life，The researcher did not intervene，Thus objectivity。On the other hand，Due to different types of big bet365 horse racing odds generation channels、The production logic is different，Researchers can take bet365 horse racing odds production logic into consideration during the research process，And analyze the possible impact of bet365 horse racing odds production form on research conclusions，You can also try to clean the noise in the bet365 horse racing odds through bet365 horse racing odds cleaning methods and algorithms。For example，In terms of social network big bet365 horse racing odds，There may be a "maritime army"、Factors such as chatbots play an important role in the bet365 horse racing odds production process，In turn, the bet365 horse racing odds is filled with false information, etc.，At this time, researchers need to use algorithms to eliminate information that cannot reflect real public opinion or user characteristics。

　　Representativeness awaits continuous accumulation of bet365 horse racing odds

　Not just objectivity，Representativeness (or bias) is also an important issue in the debate on the nature of big bet365 horse racing odds。In the book "Big bet365 horse racing odds Era"，Viktor Mayer-Sch?Junberger) et al. (2012) will "more are not random samples，But overall bet365 horse racing odds” is listed as the primary content of big bet365 horse racing odds thinking。They think，The arrival of the big bet365 horse racing odds era not only means an increase in bet365 horse racing odds size，Under certain circumstances, it is even possible to collect all the information about a certain phenomenon。Then，Is the “overall” in the big bet365 horse racing odds paradigm the true “all”?？If not，What about the representativeness of this "overall"？

Regarding the above-mentioned "total" theory，E.Hargittai (2015) raised strong doubts。In his opinion，If the "overall" hypothesis is true，That means that a wide range of insights can be derived through big bet365 horse racing odds analysis、General conclusion，Even if the bet365 horse racing odds used for analysis is limited to specific social networking sites or services。However，No matter how large the user scale of these websites or projects is，Research on these users will naturally miss or exclude non-user groups；And，User preferences of different websites、Characteristics and experience with Internet technology (such as frequency of use and proficiency) further weaken the representativeness of the bet365 horse racing odds。

Some empirical studies have also shown the doubtfulness of the “overall” view，For example, Brenner et al. published a research report in 2013 entitled "72% of adult Internet users are social networking site users"。You can tell just from the title，The number of Internet users is not equal to the total population，The number of users of social networking sites is not even equal to the total number of Internet users。So，Analysis that relies solely on social network big bet365 horse racing odds is obviously not an “overall” analysis。It can be seen from this，In the sense of "science"，Big bet365 horse racing odds may not be able to avoid more research biases than small bet365 horse racing odds；In specific research，Regardless of the size of the bet365 horse racing odds, conclusions must be carefully derived based on its sampling frame。

Whether the analysis results based on big bet365 horse racing odds are representative？Or，Is the conclusion based on big bet365 horse racing odds analysis unbiased？This problem can be viewed from two aspects。On the one hand，The bet365 horse racing odds generated by different channels and platforms may be different in how close they are to the overall population。For example，In terms of social media big bet365 horse racing odds，The number of users on some platforms may be closer to the overall number，The number of users of some platforms is only a part of the social members。One possible research path is，Compare the bet365 horse racing odds analysis results from different channels，Or gradually approach the overall reality in the form of a puzzle，Then test the robustness and reliability of the analysis conclusions。On the other hand，Looking at the long-term trend，With the digital transformation of social life，The user base of various platforms is also growing，Users gradually approach all users，At the same time, these platforms are playing an increasing role in social life，Therefore，The laws and characteristics of the social system described based on these big bet365 horse racing odds are even more important。

　　Computation and theory promote each other

According to Anderson (2008)，Because big bet365 horse racing odds can “speak for itself”，Social science is about to enter an "era of the end of theory"。This view has been questioned by many scholars as soon as it was proposed。For example，Although the book "Big bet365 horse racing odds Era" claims to end causal analysis，In order to "let the bet365 horse racing odds speak for itself"，But also admit that the end of causality does not mean the end of the theory，“The era of big bet365 horse racing odds is definitely not an era in which theory dies；On the contrary，Theory runs through all aspects of big bet365 horse racing odds analysis”。Theory in bet365 horse racing odds mining、bet365 horse racing odds Analysis、These three aspects of bet365 horse racing odds understanding all play a key role。bet365 horse racing odds cannot “speak” by itself，Only analysis guided by theory can “speak”。Even in the bet365 horse racing odds mining stage，Also depends on the use of statistical tools and bet365 horse racing odds models，And the establishment of the model、The selection of parameters is inseparable from the guidance of theory；The excavated bet365 horse racing odds must rely more on theoretical speculation。However，Current big bet365 horse racing odds research，Especially domestic big bet365 horse racing odds research，There is a rather superstitious tendency to "let the bet365 horse racing odds speak for itself" and ignore or even despise theory。

Computer scientist Jim Gray once proposed the view that big bet365 horse racing odds is the "fourth research paradigm" in 2007。According to this view，There have been experiments in human history、Theoretical deduction、Three scientific discovery paradigms of computer simulation，In the era of big bet365 horse racing odds, a fourth paradigm called "bet365 horse racing odds-intensive scientific discovery" has arrived。Although this view highlights the driving role of big bet365 horse racing odds in the scientific exploration process，But we do not deny the guiding significance of the theory。The essence of the fourth paradigm is not to use big bet365 horse racing odds to completely replace the experiments in the first three paradigms、Theory and Simulation，But unify them with bet365 horse racing odds on a new basis。The "Gray's Law" in the fourth paradigm is the embodiment of the leading role of theory。

Actually，The relationship between calculation and theory is not just as simple as guiding and being guided，Computation, in turn, can promote the verification and development of theories。Specifically，Computational social science provides a "third way" different from experimental methods and sample survey methods for testing previous research conclusions and social theories。Not only that，Over time，Social science big bet365 horse racing odds with strong timeliness will gradually become multi-time series、Ultra-high-dimensional "longitudinal bet365 horse racing odds"。From the perspective of scientific discovery，These "longitudinal bet365 horse racing odds" contain huge theoretical value。

　　Promote the integrated development of multiple research methods

With the advent of the big bet365 horse racing odds era，Are the traditional experimental methods and sampling survey methods obsolete？Whether the original cognitive path of "theoretical hypothesis-model-test" should be abandoned？Are the traditional analysis paradigms and the new paradigm of big bet365 horse racing odds substitutes or complementary relationships？Many scholars tend to use traditional experimental methods、Sampling survey method、Methods of content analysis through manual coding and computing paradigms under the big bet365 horse racing odds framework (such as machine learning、Neural Network、Natural language processing) combined，And emphasize the complementarity between these different research methods。But many empirical studies under the big bet365 horse racing odds paradigm tend to believe，Traditional research methods、Research logic and cognitive paths are basically outdated。With the help of massive bet365 horse racing odds and powerful complex algorithms，Researchers can make accurate predictions in many research fields without theory。Some researchers believe that，The theory-driven and bet365 horse racing odds-driven aspects of traditional analysis methods should be combined，Through two-way interaction，Organic integration promotes computational social science research。However, some scholars believe that，With the further development of artificial intelligence，The computing paradigm is likely to produce a revolution in research methods in the social sciences，The era when the analysis paradigm based on small bet365 horse racing odds dominated the world may be gone forever。

In summary，As an emerging comprehensive research field，The development of computational social science is not only facing the issue of whether big bet365 horse racing odds can effectively describe the operation of social systems and effectively analyze social facts，At the same time in research methods，There are also issues of competition or complementarity with traditional research methods。For these controversial issues in research methods，Researchers can only answer in the practice of social science。In big bet365 horse racing odds、The application of technologies such as cloud computing and artificial intelligence in the research process of social sciences，The methodology of computational social science will gradually mature。

　　(Author’s unit: Institute of Advanced Finance and Economics, Hubei University of Economics；School of Social Sciences, Wuhan University)