Division of Chinese opinion on Russia-Ukraine War: Analyzing Key Actors’ Strategic Framing on Weibo

Is strategic framing in authoritarian regimes fully under the control of the government? How could opinions on the Russia-Ukraine war of Chinese key actors differentiate from each other? Based on strategic framing theory, we use Structural Topic Models (STM) and a computational corpus-based approach to conducting a discourse analysis of 576 blogs produced by Chinese media outlets and opinion leaders on Weibo, the most popular social media platform in China. We argue that the strategic framing of private media is internally linked and influenced by state media in terms of topic distribution, topic trending, and position. However, the neutral position of the state is designedly to give space for private media to challenge the dominant frame of the state, which reflects an actively weakening control over the media ecosystem in China.


I. INTRODUCTION
Without any sign of ending, the Russian-Ukraine war has bombarded the lives of people all around the world since its beginning. The propaganda wing from all stakeholders is considered as another powerful weapon during wartime. China, as a permanent member of the UN Security Council and the comprehensive strategic partner of coordination of Russia, its position has been of great interest to the international community. Previous studies of the Chinese propaganda system show that the government dominates indoctrination and mass mobilization, restricting counter-narratives (Shambaugh, 2007;Gunitsky, 2015). However, other active participants are becoming more competent and more willing to advance their own framing in the era of digital transition (Shirk, 2011). It is important to study the division of opinions on the Russia-Ukraine war from different media outlets because their discrepancies in strategic framing reflect the underlying political ideology, national interest, diplomatic sensitivity, and cultural congruence within the media ecosystem of China. (J. Lin, 2022).
As earlier literature argued, the most important aim of strategic framing for the state during wartime is legitimizing the political demands. However, is the situation the same for commercial media and influencers in social media? Are state-owned media in China trying to promote strong framing among all media outlets, and what strategies have been adopted? Since the relatively insufficient number of studies in this area, this article aims to provide qualitative evidence for divergences in the authoritarian propaganda ecosystem.
We first collected coverages of the Russian-Ukraine war from 24 most influencing media outlets and opinions leaders on Weibo from Feb.24 to Apr.24. Then we used Structural Topic Modeling (STM) to compare similarities and differences in topic distribution, trending topics and correlation between different media types. At last, we tried to identify different framing of the Russian-Ukraine war which are promoted in Chinese social media through discourse analysis. By examining keywords, emotional leaning as well as the textual part of the news, the research investigated the changing power relationship between private media and state-run media in the time of war specifically. Meanwhile, we also find suggestive evidence that media information control in China is actively weakening, and in addition, political elites, scholars, and media are heterogeneous and fragmented actors that promote distinct strategic framing even under the dominant propaganda of the state.

A. Framing and Strategic Framing
According to the most widely accepted definition, framing essentially includes selection and salience. To frame is to select some aspects of a perceived reality and make them more salient in communication, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described (Entman, 1993). In other words, by making certain considerations salient, framing simplifies or shapes the way in which the audience understands a particular problem and its potential solutions as a crucial communicative process (Bolsen & Shapiro, 2017). This process is idealized by a conventional expectancy-value model which evaluates an individual's attitude by the weighted sum of evaluative beliefs about that object. (Chong & Druckman, 2007b). Although individuals may only have vague perceptions of political topics, they still can put different levels of attention on various aspects of a subject (Chong & Druckman, 2007a). For this reason, political actors tend to determine the parameters of a debate even before it begins by structuring the meaning and significance of a political message (Dan, 2018).
This core business refers to strategic framing, which occupies the heart of the political process and the aim of which is highlighting the causes of problems, encouraging moral judgments, and promoting favored policies (Entman, 2007). In general, social movements apply strategic framing to gain public recognition and approval, to define enemies, and to highlight the movement's own role in reaching its goals (Guenther et al., 2020). The idea underlying the theory is that how a social problem is diagnosed will affect the proposed solution and rationale for collective action. Some studies also point out that strategic framing in the political realm consists of framing coalition (Mayer et al., 2010) and framing contest from the perspective of dynamic progression (Dan et al., 2019). Framing is an indispensable tool through which leaders and their opponents influence or prime the agendas that people think. And they often succeed in imposing slant on mainstream media reports (Entman, 2007). Although strong frames have a good chance of winning the frame competition, (Chong & Druckman, 2007b) other factors occasionally reverse the situation. What's more, the active role that people at the bottom of the discourse power are playing in the strategic framing process as well since social media enables the voices of ordinary people to be heard (Jamal et al., 2015). However, the preferred news frame pushed down to the audience is controlled by the highest level of administration and other elites in most situations (Entman, 2003). Strategic framing effects discussed above lay the foundation for strategies that governments normally use to increase the power of the frame.

B. Situation in Democratic Countries
As Clausewitz's famous argument, "war is the continuation of politics by other means." Not only campaigns on the front matter but also the war of discourse influences the progression of wars. That's why in times of war, states have to strategically control information and shape public opinion so that they could get popular support, which is crucially important for the legitimacy of their political demands (Baumgartner et al., 2009;Vladisavljević, 2015). Some studies focused on what kind of strategic framing tactics were employed by democratic regimes after the outbreak of wars. The first strategy is storytelling. As George Lakoff (1999, p. 24) proposed that a common fairy tale structure is an effective way to justify a war, in which actors, in reality, are corresponded to the casts of hero, villain, or victim. Metaphor is another tactic used to legitimate the existence of western powers in conflicts. For example, the United States along with France and UK portrayed the Iraq war (1990)(1991) as an exciting adventure while associating the war in Bosnia (1995) with horrible nightmares and violent natural catastrophes. Metaphors related to people's daily lives are dissolving the cruelty of military intervention, thus it is logical to confirm that metaphorical frames are selective and restrictive, trying to weaken features incompatible with a particular structure (Kuusisto, 1998).
Another important element that decides the degree to which a single news construction favors one side is the use of bias. Content bias, mobilize bias and media decision bias are three common forms that serve to promote or enhance the influence of one side in conflicts through news slanting (Entman, 2003). What's more, strategic framing modifies the first core concept, namely, history and identity arguments which leads to the reconstruction of the whole story. Researchers from seven western countries have found that the core concept in the American framing of the ongoing Russian-Ukrainian conflict is that Russia takes full responsibility for its aggression, and deterrence must be used to resolve it. By contrast, the core framing in France is legitimizing Russia's actions as a response to the perceived threat from the West. It is mainly due to France's increasing political strength (Koval et al., 2022). We could conclude that the predominant framing reflects political preferences in a large part and that's one of the main reasons why they are mutually exclusive but internally correlated.

C. Situation in Authoritarian Countries
In contrast to the abundance of literature on democratic countries, there is much less information about strategic framing applied by authoritarian countries to engage in wars of discourse during wartime. To date, we found that studies about framing in authoritarian countries focus on who is the dominant actor in social media. Mainstream articles argue that authoritarian governments hold the power to shape strategic framing. controlling the content of most information available to the public. In the Chinese case, it remains an important instrument in the Party-state's toolbox of control (Shambaugh, 2007). Authoritarian governments are also exploiting social media to further social control and to foster their goals during conflicts by adapting their activities to the logic of media operation (Vladisavljević, 2015). Hashtag campaigns that are subject to state-led manipulation and engaging genuine users in building popular engagement are widely used strategies (Leber & Abrahams, 2019). What's more, the Chinese government is considered to deploy full censorship on social media, which serves as a tool of autocratic stability (Gunitsky, 2015). Strategic framing seems to be increasingly weaponized by Chinese media particularly as the conflict intensifies, whilst for American journalists, it seems to be a dominant routine. Chinese press keeps adhering to political authorities and serving the national interests as a whole.
However, some scholars hold a different opinion as studies suggest that other actors in the political realm consciously withstand the absolute power of governments and play an active role in shaping the process of strategic framing in authoritarian systems. The first reason is that rapidly changing power relationships selfimposed constraints on the media so that they can introduce underreported interpretations of wars (Bennett et al., 2007). Based on previous research, the social media era in China shows a liberal leaning with the emergence of opinion leaders and a close involvement of traditional media (Tong & Lei, 2013). And by no means always a unified actor, the administration includes a variety of players, and disunity has significant implications for the dominant strategic framing of media coverage (Entman, 2003). In line with theoretical arguments, several studies suggest that the Chinese government is facing a crisis of hegemony over framing truth and disseminating information in the war of position (Shambaugh, 2007).
The second reason is that the subservient or interventionist role of the press is not mutually exclusive, on the contrary, they may coexist and complement each other. What's more, news media, as indexing theorists suggest, functions actively to boost or hinder the governmental preferred frame in international conflicts (Liu et al., 2022). In fact, the exertion of state powers trying to control events or influence political outcomes may generate opposition and hostility that reverberates through social media, which counterbalances the hard power (Jamal et al., 2015). Despite the advocacy frame stemming from the government, this dominant position is not that guaranteed.
Existing discrepancy over the strategic framing capacity and intention of authoritarian states on social media leaves space for further investigation. This thesis aims to explore the following questions.
RQ1：How does key strategic actors in Chinese social media frame the Russia-Ukraine war? RQ2：Are there any differences among strategic framing of key actors on Weibo?

A. Data Collection
To learn about the Chinese government's strategic framing, we examined posts about Russian-Ukraine War on Weibo (the largest social media platform in China) and collected two-month data from the beginning of the war, which was February 24 th , 2022. The data-gathering process was composed of three steps. The first step is selecting representative Weibo accounts. We browsed the Most Influential Users Rank (a rank automatically produced by Weibo according to each account's real-time attention) in the discussion of the Russo-Ukrainian war and initially identified 42 accounts. To ensure diversity within each group, we eliminated accounts having a similar number of followers and those merely retweeted short messages, remaining a total of 7 state-owned media outlets, 5 commercial media accounts, 4 scholars as well as 8 key opinion leaders at last.
Second, we used a Weibo Crawler programming(https://github.com/dataabc/weibo-crawler) written in Python gathering all microblogs posted by these chosen accounts from Feb 24 to Apr 24. Keywords included "俄" (Russia), "乌" (Ukraine), "普京" (Poutine), "北约" (NATO). Notice that the starting date was the official beginning point of the Ukraine war and the closing date was set as the day before our formal analysis. User profile (including Username, user id, gender, birthday, and location), number of followers, posting data, content, link, hashtag, and also the number of likes, replies, and retweets is collected. Taking into account that the frame theory itself is applicable to long texts and that the structural topic model performs best with long texts, this study chose posts that exceeded 250 words. In total, we identified N=596 articles.

B. Methods
Structural topic modeling (STM), a type of modeling method which estimates changes in the proportion or content of estimated topics according to the meta-information of the documents is being used to explore the main themes covered by Chinese stakeholders during the Russian-Ukraine wars. Topic modeling considers each document as being composed of terms, each topic as a distribution over terms, and each document as a combination of topics (Chen et al., 2020). This method enables researchers to incorporate metadata such as the date, the author, and the source of the document into a topic model so that text could be mapped into meaningful categories, which might not be apparent to a reader (Xia et al., 2022). In this study, we have used an R package named STM (Roberts et al., 2014) to compare the topic prevalence between different accounts. It turned out that the 6-topic model has the greatest semantic consistency as well as exclusivity between topics. Thus, we finalized the input by selecting the 6-topic model. What's more, the texts analyzed in this study are all in Chinese. We used Jieba to tokenize sentences and then remove emotions, punctuation as well as stop-words to make results analyzable.

A. General Description of Russian-Ukraine War in Chinese Media
Concerning the first research question, we need to find out topics that are embedded in all texts through STM. Fig. 1 shows the topic distribution and suggested label of the corpus. Table II shows keywords and the most frequent words of each topic. Based on these words and find Thoughts function, the author finally determined the implication of each topic. First of all, we find that major stakeholders of the Russian-Ukraine war in the world have all been included in the coverage of Chinese media. Direct messages related to military actions are shown in Topic 1, and political reactions from other countries as well as the gain and loss of each country in this war are also reported (Topic 6). What's more, controversial Biolabs of America and NATO-allied countries in Ukraine (Topic 4) gains great attention from Chinese media. Some topics show apparent sentiment orientations. For example, the keywords of Topic 3(Involvement of United States) and Topic 4(NATO's Biolabs) are mostly negative and critical, conveying condemnation towards the United States. The estimated topic distribution is also displayed in Fig. 1. Among all, a deep analysis of how Russia-Ukraine war could influence the supply of living resources such as natural gas represents the dominant position while the battlefield situation in Ukraine attracts the least attention. Similar to our finding, Hanley (Hanley et al., 2022) also proved that Chinese news media have concentrated on the conflict's political and economic consequences, on the contrary, the western press conveys more information on the progress of this war. Another feature is the emphasis on the diplomatic reaction from friendly states such as Serbia.
More importantly, China's neutral position on the settlement of the Russian-Ukrainian issue is highlighted. High tendentious distribution of different topics somehow demonstrates that it is national interests keep leading the press instead of the direct needs of information from normal citizens, especially in times of war (Liu et al., 2022).

B. Strategic Framing of State Media and Private Media
In the next section, we introduced the covariate "media type" to compare internal distinctions between the strategic framing of state-owned media and private media. Fig. 2 displays the statistical estimation effect of media sources on the topic relationship in a visualized way. The themes located to the right of the dashed line indicate that the state media has more analytical coverage on the topic, while themes distributed to the left of the dashed line indicate that the private media presents more on the topic. In the image below, the scale represents the normalized percentage distribution of topics, where 0.1 indicates that state media covered the topic 10% more than private media did.
The private media specifically analyzed how could this war impede the trade of natural resources from both Ukraine and Russia, the economic situation of Russia Fig. 3 (i.e., Topic 2). Meanwhile, these news producers and opinion leaders also provided deep insights into United States military assistance to Ukraine and the huge fortune that US military enterprises gain through this war. Fig. 3 (i.e., Topic 3) On the contrary, state media focuses on specific details about the military progress including weapons used, dynamic messages from major cities as well as causalities. What's more, the existence and of NATO's Biolabs, diplomatic responses from China and other stakeholders are also major concerns. Fig. 3 (i.e., Topic 1, 4, 6) But deep reports about the humanitarian needs of Ukraine don't show a significant difference between state media and private media considering the overall distribution. In contrast to the prevailing opinions that authoritarian governments are exploiting social media as weapons of propaganda by which to divide and discourage their opposition (Leber & Abrahams, 2019), our findings show that state-owned media take serious attitudes towards controversial topics. It still bears the responsibility of providing timely and accurate information to the audience while intellectuals and market-based media had heated discussions about topics that are directly related to national interests and the position of China towards this war in an explicit way. They develop their own position to lead public opinion with instantaneous communication and width of reach (Tong & Lei, 2013). STM also allows us to investigate the interaction between covariates for the reason that one variable may have a moderate effect on another (Roberts, Stewart, Tingley, Lucas, Leder-Luis, et al., 2014). Therefore, in order to get a clear picture of how the strategic framing of Chinese media changed as time went by, we further displayed the dynamic trends of topic proportion in state media and private participators' coverages. In Fig. 3, the X-axis represents the number of days, and the Y-axis shows the proportion of topics. The first finding is that the distribution of topics discussed in state and private media displayed an overall consistent trend and correlation. This suggests that the choice of topics by private media and opinion leaders affects the presentation of the topic in the state media in China. The second finding focuses on the relationship between time sequence and topic proportions. Fig. 3, Topic 2 witnessed an increasing trend as the Ukraine crisis intensified, however, a dramatic downward trend is shown in Topic 5 and Topic 6. Take topic 2 as an example, the fact that natural resources supply became a focus of both state and private media 50 days after the war has broken out indicates that record oil prices, food security, and natural gas shortage caused a negative spillover effect. Chinese opinion leaders and private media show complex attitudes because, on one hand, they criticized the humanitarian crisis caused by military conflicts, however, on the other hand, they highlighted imposed sanctions on Russian people are hegemonic. More importantly, the majority The decline in humanitarian aid towards Ukraine Fig. 3 (Topic 5) and the diplomatic reactions of China, as well as its allies (Topic 6), are intrinsically linked. During the initial period of war, Chinese media reported the government's humanitarian assistance to Ukraine which served as a signal to demonstrate neutrality and consistency with international order. As time went by, the main focus shifted to condemning the US smear campaign against China's position on the Russia-Ukraine conflict and further, attributing this conflict to NATO. By selectively emphasizing issues favorable to political parties and de-emphasizing issues that are unfavorable to them, the political space is defined (Elias et al., 2015). The downward tendency of how China and its friendly nations view this conflict is consistent with the salience theory in political competition. Since the Chinese public and leaders have a sympathetic understanding of Russia's actions in Ukraine which do not meet the expectations of western countries, tremendous propaganda may damage China's national image of being "peace-loving and respectful of territorial sovereignty." Paradoxically, both state and private media put lots of effort into reporting the fortunes that US military enterprises gain, NATO's expansion, and controversial Biolabs in Ukraine Fig. 3 (Topic 3, Topic 4) since the proportion of these two topics increased slightly. Thereby, state officials in China are not supporting the war, but they explicitly agree that many conflicts should be attributed to international orders based on the benefits of the US and its allies. In other words, the dominant framing of state media is moral/ condemnation, with the aim of revealing the double standards of the US in several military actions and protecting the strategic relationship with Russia. Private media and intellectuals have much more freedom to predict the situation and analyze the underlying motions of all stakeholders.
At last, through keywords of each topic, we selected the most representative coverages to analyze different positions in state media and private media. Content analysis results show that there are three major features when covering each topic.
First of all, the inclination toward entertainment is evident in intellectuals and opinion leaders. There is a highly popular category on Weibo called "Watchers of Russia-Ukraine situation" where military bloggers, journalists, scholars in international relations, and other relevant influencers express their opinions. In general, coverage from the private sector shows colloquial features with high emotional words and emotional tendencies. When talking about war progress and humanitarian aid from Ukraine, private media show highly positive attitudes while making fun with the possible gap between the two countries in terms of military power. Another feature is personalization. Stakeholders in this war are usually referred to by their leaders. In this way, the political and military game is simplified as decisions are made on a personal level. What's more, in light of the Western suppression of China and Russia, the Chinese public naturally understands and supports Russia which intensifies highly positive attitudes towards Russia while sympathy for Russia has been dissolved. And there exist some more objective and neutral voices reminding previous Russian Chinese conflicts.
State media coverage is dominated by a moral evaluation frame with consistent attention to the United States. In most coverages, America is framed as the destroyer of the current international order even itself created it. The double standard serves as the most important narrative when talking about sanctions against Russia, military progress as well as the humanitarian crisis in Ukraine. The United States should likewise be sanctioned for the wars he has committed is what Chinese state media trying to convey. Secondly, the United States is framed as the biggest winner of this great conflict. Military enterprises, energy-supply alternatives, reunition with European allies as well as weakening of the international image of Russia are major benefits being mentioned. In contrast to private media, state media keeps cautious when talking about Ukraine and European countries. Generally, they are framed as victims in need of help and friends who still keep reliable cooperation and friendship with China.

V. CONCLUSION AND DISCUSSION
Theoretically, previous studies have given proof that private actors including commercial media, audiences and the internet are dynamically influencing the media ecosystem in authoritarian countries (Tang & Huhe, 2014), however, there is limited research directly comparing the different ways in which distinct actors in the public media sphere have used strategic framing on social media to launch the propaganda and to influence the flow of information. Practically, it is reasonable and urgent to study China's response to the conflict at home given the "no limits partnership" between China and Russia. To fill this gap, this study applied STM and systematic content analysis to analyze differences and similarities between coverages of Chinese state media and private media from the perspective of strategic framing. Through descriptive analysis of articles posted by the most influential media outlets on Weibo, we suggest that private actors are challenging the strategic framing set by the state-owned media, seeking to engage in this conflict in a tough, clear, and high-profile way.
First, we find that the discussion about the Russia-Ukraine war on Chinese social media has become an ideological war defending and polishing rising of China under the context of Sino-US strategic rivalry rather than drawing public attention to the war itself. In line with prior research, Chinese news media concentrate more on the conflict's political and economic consequences on the international order. Second, private media and scholars are increasingly raising their voices with the catalytic effects of the Internet to cast their influence. Private media distributed more resources to eye-catching topics, such as predicting warfare and making fun of power games among the US, Russia, and China. Moreover, private media no longer takes the same neutral stance as official media. On the contrary, it has become more explicit, and aggressive when expressing its emotional leanings, especially the blame, and condemnation towards the US. Such a finding further proves that by owning much more political resources, market-run media become more audacious in challenging the state (Lin et al., 2015). Despite differences, the dominant strategic framing is consistent with the leading geopolitical-game framing promoted by the national media, which provides a monodirectional and simple narrative for the Chinese public to understand the war in the context of US-China strategic competition.
However, in this case, state media are weakening its influence deliberatively. Such a finding is consistent with King's conclusion that strict censorship in China is oriented toward radical collective action rather than different, negative, or even criticism of the state (King et al., 2013). On one hand, Chinese state media shifted the focus to the origins of conflicts and adopted a selective framing approach to frame United States as the destroyer of the current international order and the last winner of this war. The private media's criticism of China and its accusations of Russian aggression are equally important and not suppressed or drowned out, as many scholars contend (Bradshaw & Howard, 2019). The "neutral" position of the state can be interpreted as resonating with western hegemonic discourses, seeking not to damage the important strategic partnership with Russia and to ease the democratization pressure from western countries at the same time (Jourde, 2007). In general, by emphasizing political and economic gain and loss, key actors on Chinese social media contribute to the justification of military actions launched by Russia, the aim of which is revealing the responsibility of the United States and gaining support for the Russian government. As time went by, both private media and government-run media were gradually dissolving the evil nature of the Russian-Ukraine war so as to frame this war from the point of geopolitical competition.
To summarize, our analysis demonstrates that key actors in the Chinese media system are legitimizing and making use of the Russian-Ukraine war for political and commercial aims, but such discrepancies have in fact contributed to breaking the hegemonic censorship, and regulation power owned by the state in the media ecosystem. Influenced by the web, global crisis, commercial factors, and public sophistication, the state's control over the information system can be weakened (Shambaugh, 2017;Shambaugh, 2007). Future studies shall be expected to deeply explore dynamic changes in the Chinese media ecosystem from the angle of power, information control, and hegemony.