 We are back with another English presentation. This one comes to us prerecorded from the CUT UA, the communication analysis team for Ukraine. And the title of this presentation, as you can see already, is war communication in Ukrainian social media space. And I hope that everything will work smoothly. I'm still downloading the file as it is being played, but I will start the video file now. Dear friends, I'm going to tell you about our volunteering initiative about social media analysis. We call our CUT UA communication analysis team Ukraine. We started as a volunteering group from February 24. It was the first day of the war. So we as media analysis guys tried to help our militaries and all who need this to feel better in communications and in information work. So we started this analysis and on a daily basis, we provided to them. And I will show a bit about the first results. You can see that these results were February 24, April 4. So these are data that it is safe to show to people. And I can share it to you because now more recent data is much more sensitive. It's impossible to share it, but this presentation will show how do we work and what we usually do and maybe how you can help us. So let's see. Let's start with the short explanation how both sides of the war participate in information war, how these both machines work. Let's start with Russian machine. It is very traditional. You maybe know that it was started with so-called historic, actually pseudo-historic speech of Putin. He told that Ukraine never existed and a lot of other conspiracy theories. It was published at night, his speech, and then he started, he launched the war and we Ukrainians hear the explosions and so on. Other people who helped Putin to create strategic narrative. Let's call this, so his court gestures, maybe you know some of these people, Medvedev, Dugin, Margarita Simonyan, a lot of people who can tell something much more radical than Putin can, but usually nobody considers them seriously. And at last all propaganda machine of Russian state is Russian TV, you know, Russia first channel and Russia today and other channels. The second is RIA Novosti and other web and news websites. And at last so-called troll factory in Olgyno. These guys who leave a lot of posts, comments and so on. And also promote state outlook. So they work in highly centralized machine and all this structure leads by one guy, you know it on the first number. And this is their power and their weakness and how Ukrainian machine works. And as you see from the headline, it is highly decentralized. And I have to make some explanation about what is so-called effective public. It is the definition of Zizipapachirisi. This scholar provided it when studied revolution, so-called Arab Spring, but applied it also to another events. This is the state of public that appears during large-scale social movements like Arab Spring, Revolution of Dignity, Occupal Wall Street or even Me Too. So it can be completely online movement, but people also can be engaged in something in this state called effective public. This state provides its own connective gatekeeping. What maybe you know, what does it mean gatekeeping? So it selects news that are valuable. Usually media selects news, but here a web community selects news when it is engaged in some process. So it provides connective gatekeeping. It provides connective framing. It's also very interesting thing because a lot of facts may be framed by some and other circumstances. So usually media provide framing for facts, but in the state of effective public, just people provide framing for the events. And also that's interesting that they provide connective storytelling. When same people became both actors and narrators of the story of movement, so they first act as an actors and then became tellers who provide the narrative about this story. So this is how social media environment now works in Ukraine. There is nothing similar in Russia because they are all environment highly centralized, but here people decide by themselves what they have to discuss, what they have to analyze, and so on. And as we show before, not before, but then I will show it, neither Ukrainian nor Russian authority had crucial impact on the Ukrainian strategic narrative. Well, let's move to our research. We took daily the set of web posts from commercial monitoring system Usekan. These are posts from Facebook, Instagram, Twitter, YouTube, Telegram, Vkontakte, and TikTok. They are searched for the words typical in describing the military actions or its automatic programs that selects such posts. These are posts that have geolocation Ukraine or not defined. So we don't want to lose these posts who didn't set their geolocation. And also we select about 3% of daily amount of posts. And that is the amount that we can process manually. It's 100 and excuse me, 1,050 posts from all data set. Coding, manual coding. I will tell a bit more about why it is manual in the end, but only manual coding that we apply. First of all, we select a post from Ukraine from the not defined geolocation posts. So our analysts look on not only the content of the posts, but also on the profiles of people and determine whether they are from Ukraine. Then side of the conflict, we determine whether this outlook is pro-Russian or pro-Ukrainian. Some scholars asked me, didn't you see posts that are not pro-Russian, not pro-Ukrainian, something in the middle? And I always answer that it's impossible because when you see shelling bombs on your house, it's impossible to remain neutral in this situation. Then for pro-Ukrainian posts, we also got mood of the post. It is the topic belongs to good news, biggest news or bad news. It is determined based on whether the facts reported in the letter are positive from the Ukrainian perspective of the war or not. Then whether the post contains link to the official source, we determine emotions also manually, and a message of communication that is the most interesting. It's my own approach developed for commercial media analysis for detection and evaluation of the success in, for example, information campaigns for different brands. But here it was applied to propaganda detection. Messages of communication are evaluated for both sides of the conflicts. So additional system automatically indicates gender and the region of residence. And we can use this also for our analysis. What do we have? From the beginning of the war, the share of pro-Russian attitude is very, very small. In all Ukrainian information space, we have only about 9% of posts with pro-Russian attitude. The most interesting that this 99% of pro-Ukrainian attitude are very different posts. These are posts from media, posts from authorities of different level, and most of all posts from ordinary people who are engaged in this situation, who want to participate in war as volunteers, as militaries, as other people who just support the economy of Ukraine, and so on. So this is really like people's movement. But this 9%, it was just opposite. Most of these posts are official posts, as we can expect from highly centralized machine. So usually these are people who maybe share posts from official sources, or just these posts from them. And also these are posts from known pro-Russian bloggers that live usually on occupied, many time ago occupied territories like Donetsk, Lugansk region, or Crimea. So it's almost nothing from other parts of Ukraine, only on the territories who were occupied 8 years ago. There are some posts with pro-Russian attitude. Let's focus on messages. Motivating messages of pro-Ukrainian users usually focus on motivating messages about military achievements. That was the most interesting circumstance that allowed Ukrainians to keep themselves motivated for the war, because they saw that it's possible to repulse Russian troops from Ukraine. So that was the main message from the first months. Also, other messages like Russia is attacking civilians at the end of this period when Northern Ukraine was liberated. A lot of eyewitnesses about war crimes appeared in Buca. Maybe you know this word, this name of the city, Buca, Ustomil, different villages of Northern Ukraine. All these stories appeared in the information space. And also it was very interesting that these stories didn't provide sadness. We will see it on the plot about emotions. These stories provided only anger. So people became angry, more and more angry. The same plot we have about pro-Russian attitude and propaganda focused on the Donetsk and Luhansk. It was not about all Ukraine. They usually spoke about these two regions, who were occupied eight years ago. And they still told that Ukrainian soldiers, by their version, are shelling the Donetsk and Luhansk still, that for eight years I will not just focus on this issue, because it's not true. But they didn't find nothing more actual to communicate with Ukrainians. They just repeat and repeat the same message. The second place is that Russia takes control of cities and strategic objects. And so on, the humanizing of Ukrainian people, that Ukrainians are Nazis and extremists and so on and so on, they are trying to persuade, but they had no success in it. Let's look on mood and sources of pro-Ukrainian citizens. We saw that despite a lot of tragedies, despite of all loser territories, the largest share of pro-Ukrainian posts was about good news. A lot of people focused not on losses, but on maybe small victories, sometimes big victories, but on things that show that we can really win this war. Also, just look on the share of official sources. It is about 18% for the first months. So you can see that most of the discussion is kept on the side of that is not controlled by the state. So as I told, it's really people's war. Let's look on the dynamics. We saw very interesting dynamics that, yeah, till the end, it's not full plot, but it was almost the same till the April 4th. So we can see that there are a lot of spikes of optimism or pessimism, but till the beginning of the April, it was slow pessimism strengthening. After that time, the dynamics changed, but maybe you see that sometimes there were huge spikes of optimism. It was different events like destruction of Russian aircrafts or ships, and some peaks of bad news were caused by Russian crimes against a civilian population and so on. Let's look to the next slide, and here we see that news from official sources have a very different share of good, bad news and ambiguous news comparing to informal sources. Official sources during the first period of the war were much more optimistic. 55% of official communication was dedicated to the good news, but people didn't rely on this. People, as I said, created their own storytelling. They didn't just tell or retell the news. They changed this news, changed this story. They participated in it, so they have just a very different share. We see that a lot of ambiguous news, for example, these are posts about volunteering activity. It's not good news. It's not bad news. It's just calls to participate in the volunteering movement, to join different activities that support army, or at least pay money for army, and we can see that it is a very prominent part of the civil communication of informal sources. Interesting thing that, as in most of the online discussions, males are much more optimistic than female accounts. We use just definition from social media as people report them. If they report themselves as males, we consider them males and the same about females. As it usually, males are usually more interesting, maybe these are gender stereotypes, but they usually discuss more techniques, our interesting new technique that military received. Also, they are much more about to discuss geopolitics, geo-strategies and so on. And the international support of Ukrainian army is very important for them. And females are usually much more focused on personal stories and particularly on stories about sad stories about death, about war crimes and so on. We also see how it was distributed on different regions. Of course, east is much more focused on bad news because it was shallowed much more. The west is much more optimistic because it is far from the front line, so the largest share of good news is on the west and so on. And at last about emotions, let's focus on it. And we see that the most prominent was the emotion of pride. The second is anger. These first two emotions are highly mobilizing. They allow people to keep themselves moving forward and help the army. That is very helpful for our motivation to struggle. And also a lot of love. We can see almost seven percent of love. Ukrainians are very humoristic nation. They like to laugh on their enemies. We told a lot of stories of poor, prepared Russian army of stupid Russian authorities and so on. It is very impressive that seven percent of all posts, the most prominent emotion is love. So at the end of the first month, we made the conclusions and showed it to our receivers that Ukrainian society in general remained highly motivated and mobilized during the war. Even bad news are more likely to lead to anger than sadness. But Russian propaganda has no success in Ukraine. Its own messages are unpopular. Ukrainian government mostly controls this situation in the information space, but not the agenda. Ukrainian government sometimes successfully communicates, but most of all people usually provide, determine which news they have to focus. Ukrainian society is fast in overcoming traumas. So when something bad happened, people usually try to overcome it, work internally in their society, discuss that. But usually in one, two, three days, also good news became more prominent again and again. So it's a classical overcoming of traumas, as it usually happens in Ukraine. In personal psychological state, the same thing is usually in the Ukrainian society. Yeah, and the overall emotional trend is still deteriorating over time. But this conclusion was just about those periods. Let's stop a bit more about how we process this data. It is only manually, because we use very, very complicated analysis categories. Some people tried to help us, for example, the team of scholars from the New York University. And they had no success. They tried to feed machine learning process, just training on our data sets, and tried to receive some results. But the best result for one of the categories was 60% of confidence. So it was very bad. If somebody can have any ideas how to make this complicated process that works with Ukrainian and Russian language became more automatized, how to make it faster, how to make it more convenient and less manual work, you may just tell me and I would be happy to discuss it with you. Thank you very much.