Bestand wählen
Merken

Pointing Fingers at 'The Media'

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
to thank
thank you
I I
think the next speaker was born and raised in Germany and who lives and works as a PhD student in Canada is a member of a research group on extremist politics and democratic systems and he'll give us an insight into the public discourse in Germany focused on the so called identity for Deutschland please welcome Alexander by you it thank you very much I think you people for showing up In this bought think you the Internet for watching a very big thank you for the organizers for a given the opportunity to to give this will talk um de minimis little about I and everywhere went this winter I didnt after world a winter jacket because the temperatures are where a mild and I will tell you in a minute why that matters as I said I am not a member of a research group in Vancouver where we look at what happens whole whole whole extremist parties and politics fair democratic systems and we decided to focus of this research project on the fascinating for researchers fascinating case of Germany uh in uh asking the questions if we can point fingers and so is that of the available available on a valid judgment to say the media the media is to blame for the rise of the of the uh for anyone who would you decide at the end of 2017 that they would spend most of 20 16 in hibernation which seems like a pretty good idea at the time uh I will give a quick rundown what happened so we had elections in September uh and the the domino piece that was Germany fell domino piece in the sense that all around in Europe far-right parties had been considered a success in the past in the recent past uh in Germany was a sort of last stall words in in Central Europe were far apart he did not get the government this happened in september and then it uh did not only get into government enter parliament and also the way it looks like now it might become the official leader of the opposition so when the results came in uh pundits were really quick quick to call the shots a b the you thought of dominated sentiment was that it was the media's they took the positions of the area of the and gave a disproportionate amounts of coverage to the SFA i'd extremist party yeah the and this sentiment and a lot of truthiness to it so if you have a lot of yeah sure I can see whether the red everyone and opens newspaper or opened a news website the stories or if scenes or about anything about something that's related to the area of the seems to dominate coverage and this went along with a little bit of of of of of of of of Feltz truth uh uh a truth that was perceived by people about how all of the campaigning season was a lot of season and a lot of campaigning despite March issuances best efforts uh all other sunshine but no a conflict and this was something that uh then was perceived to be very so there will be more and more savory skillfully but somehow filled by the a of the uh and and the topics that that are kind of concern to this party yeah so what are we doing today here the 1st of I'm a public by trade uh and political scientists uh like theory and all that this is a an event for a theory Madonna at the forefront everyone's minds but do this for me because of talking in are game to political science about theory is like mud wrestling with a pig but you do that for 2 or 3 hours and you realize 0 this pig eggs in action does this uh so I'll be have 1 slide what we would would previous theories would suggest has had happened and how it could have happened uh that I'll show you what kind of data we have uh collected to systematically answer this question and talk about public discourse in Germany I commented to the the meat and potatoes of the talk about how the campaign unfolded in the media and I will uh then to end I will show some more data that is a bit different uh that paints a picture on but why this election was a special election and wide was sort of a perfect storm of an election for a foreign party that and why this actually makes us claim that the media could be said to have behaved pretty reasonable as a little teaser of M theory 1 slide by 2 possible mechanisms of media effects there's this normative very endearing and and and wonderful idea that if you read something but someone carefully crafts and uh P. here she constructs uh an argument that is well written well maintained you read this we take it in your persuaded by that regardless of of what this argument is at the 60 years of media research suggests that this doesn't
happen uh preexisting opinions are extremely difficult a change in the regions every everything 1 of us in the were likely to admit that no no sure of the arm rational thinker I take standpoints if the convincing to me and I I internalize them but it's works the 2nd possible effect and the 1 that will be of of concern to us today uh at the core of the presentation is something that's called priming so the media uh can't tell people what to think it can persuade people independently of the previous opinions that people have but it's really really successful in telling people what to think about that its super good the media super good reading something is very effective in bringing something to the front of your mind and here I am the uh yeah can tell you what I told you about my my choice of attire in winter and that's iterative of you probably thought when I set this on a node where would jacket uh well what's what what it was this guy uh but maybe a few of you thought the sure was pretty mild that's climate change their so without naming the shell I am uh better a chance found a few of you to consider climate change and pull that in your frontal lobe but the following in mind and and this is important as a uh that the central thing that we have to consider we ask them if the media wrote at a party like DFT for that's also important areas that priming is easier what is an indirect effect of finding as well where at a topic that is owned by a specific party uh that's the thing that then favors the party subsequently so if the media writes a lot about refugees as xenophobic forehead party that has this problem of refugees at the core of their gender will reaping benefits in our minds in that its agenda will be will will follow from diagram the so far the theory that's all that so what we do bases fear we we collected data the at lots of data we have but we understand this data to be at wins dentist text so we collected to be data and we use natural language processing to analyze a natural forcing basically means that uh we're giving language to a computer that wasn't written specifically to be understood by a computer and then try to extract meaningful analysis based on what the computer's doing with this so we use some uh and some sifting methods to collect uh about 8 thousand of but articles from 4 central German news subsides for whispered beds in speed parents we have that results in a unique dataset that draw knowledge no 1 else uh has if so please reach out to us uh and there's a there's a 3 need that it deserves at least 6 so far modems um is also pretty exciting because that was pretty cheap uh word to people remain concerned with with the glycan they don't the underwater calculated my hourly wage um but uh it was almost and with no financial expense and this is cool because were sort scientists were were faced with this problem with this very interesting case uh of Germany and sort of falling in line very delayed with with lots of our countries are are ordered in terms of of the forehead party uh getting the since Parliament and we can use methods that are available to us for is sort of like saying that a reader stack overflow in teaching those methods to us to women to systematically tried answers questions let's start writing the share of party mentions an online news so we did for each day the we calculated what the total number of mentioned political actors as we did that based on word lists that we that we carefully crafted that included that can that's names and party abbreviations and party names and things like cancer in and cancer that for the CDC is you and the SPD respectively and we let that thing ripped through are little or a script that we have so the average of of mentions of each party over the course of the campaign look something like this uh between July 1st of September 24th that's the uh that's the time from the because their on but we see a clear uh income the bonus comes level of comfort and bonus uh the for the CDU CSU Social Democrats high twenties India the at 10 . 7 % here we might say that that's all parties but there a little note to the green and to the left so with this dictionary method is a cat tricky because we can't say where uh what we discover count every occurrence of occurrence of cooling and link it uh for at the Green Party the Left Party because then we get stuff like the green banana and the left hand discounted for them uh so that's why you were only using cameras and that's why they probably this sort of underperformed but for our purpose of talking about why the if the AFD get uh get favored by the media research letting that drop on the table so the story here is over the course of the campaign 10 . 7 % of mentions were there were happening but that mentioned the basically disclosed right if the of 12 . 7 % of election that does really solid so it was favored by the media and a few of you might done uh might know this this analysis from a blog posts that me and consent to courts road for netlytic summer like 45 seconds after the election when worked on truncated data uh and we also focused on print uh on on print media uh and this is sort of what this graph look like that we based our uh our conclusion on that have the generally get any disproportion lot of coverage it actually is and that in the last week of the campaign less we saw the campaign action is outperformed by the FTB uh science is the the current state of airing for the uh so uh now that we have better data in terms of online news data this whole story looks a bit different if we take the average over the whole course of the campaign and actually haven't shown to us stay by the day and this is 1 of what I wanna focus on them the so just looking at sort of tail and of this all the way to the right when we get close to the election date the order of this is surprisingly close to the actual election results the parties actually do get in in the order so that they came out of the election but uh we do see a little curve that gets closer to a curve that should be bigger but and the fear of where the where the ball from boss magic this is where the the interesting stuff like so it's look at the curves is 1 of the other that the year the CDU you as you would expect as the incumbent anything that is remotely political in a domestic international politics will score mentions for the transfer and the cities is that's why this curve is going so we have any others but we do see a downward tendency the closer we get to the campaign when campaign coverage shifted from the incumbent to the competitors especially the underdog competitors that which is kind of that's a bad transferred the SPD now uh but if we look at the curve of the of of
Democratic Party and there a slight bump around August modern shows really tried to drive home the system this issue of of justice as the central campaign promises uh and isn't another little slide Humber on September 1st being of september when the of the televised debate happened uh but the overall trend it's pretty linear doesn't seem to be from would just smooth this plot out to be as straight line it probably would be pretty much horizontal the a not so for the FDA so remember for over the course of the campaign they had a 10 . 7 % on average of mentions I and that's true if we calculate an average of that course uh this looks like it's cause for concern lower the 2 major parties but something happens in late August uh and all of a sudden this party gets actually closer to the to the Social Democrats it like sorry late August the tendency becomes 1 that is pretty considerably upwards and if we take the average of only the 2 last weeks before the election uh due to a number of 90 . 6 of all mentions of are so uh are talking about the ST there now it's something to think about the mechanism of priming those are the those short-term so short and facial looking for things that happen over short and have an even effect In a pretty short term so this is something that is extremely extremely important that at the beginning of this time frame where in the plot becomes some something that has a trend that shows words around like August 28th with that 1st Crystal Mountain crystal summit occurs at 2 things happened 1 a refugee boats capsized and mentoring and an event that we uh sadly and have to see terrifyingly often and what people died there and the 2nd thing that happened was that Alexander garlands but in an interview claims that the German politician should be dumped in Anatolia that it's interesting if you if you uh if you talk about if you if you extract the topics that are covered in relation to the area of the before and after this moment before this august 20th the it's a lot about uh and this vital writing e-mails were turns out she's not the public persona that she claims she is and it's a lot about internal rifts of this foreign party the year the internal tensions between the super forehead ring who wing and the far right or right wing wing uh and afterwards there there is a surprising amount of of citations this forward we should dump but this person in our in our country that's a unique as the strategy of of sort of provoking a scandal that paid off it the but if before we get into that let's look into the topics that we'll cover over the course of the combined with the same thing we developed a topic dictionaries with with uh with keywords for each category and we let our script read through all the data and count occurrences so looking at this we see a sort of a band there and 10 % rich words all the colorful rainbow with the topics don't really and only diverge from each other uh except for uh the topic of domestic security which is there at the low end of the range uh but we do have 1 topic that stands out quite considerably in the early months of the year of items some of which is the European Union General European Union topic and this is because on July 1st hand would cause the internal Chancellor uh get the 1st European active states and a lot of things were written about his legacy in terms of the European Union and lots of people showed up from Strasbourg and Brussels and pay their respects this is why this topic and seems like where does this topic comes in as strong as it does here the the topic as a sort of unusual uh curve here on our graph is the topic of the environment our dictionaries evolved were topical uh and so what causes this deep sleep summoned there in early August is that the deal give very uh the diesel summits where German car manufacturers tried to sort of get out of the fact that they basically uh ripped off customers with selling cars that emit emit toxic amounts of of poisonous gas and dust this is why this is extremely important in the hive and then low 40 % range early August but afterwards but the trend line points steeply down the topic that was pretty consistent over the course of the campaign uh in in its overall dynamic or at the sort of that although all Valdemar dynamic but the role that it played is the topic of immigration immigration means migration and refugees in in our case here at of thinking about what that means in in relation to all the priming we would think that sure that's a topic that is owned by the end of the it's it's like it's super tightly connected to to the party's rise and so this is something but does favor the forehead party like those those are like it is that we can do as sort of more systematic investigation into this so this graph shows you will the polls each dot represents uh polling results 48 of the bad and the line is the average of of those poles again over the course of the of the time frame that we surveyed pretty much constant until mid august and we have increasing variance and we have a tendency a trend line that points upwards and now this is where the heart of the story lies is this is this dependent on the mentions that the FT got in the media this is the orange line now we have a sort of with a different a different scale of our graph that's why it looks way more nervous than in the B 1 that we had at noon difficult to but if you have there like this time series data you actually get rid of trends in terms of what the analysis should be like so 1 way to do this in a graphic representation is them by not showing the absolute values and how the developed but only showing the change from day to day and plotting that this is what this graph does so here uh these 2 lines dance around the 0 mark because the fish the blue 1 uh where it's the polling results there wasn't a lot of variation from day to day it's in incremental steps that the curve points up and down it gets a bit more uh but higher variance around so that the the uh after the mid of August and whereas the a of the mentions in the media the stain they say you rich in variance hard to tell if anything systematic is there you would think that after uh the so the 1st 3rd of August those those lines are connected we ran an analysis of a vector or regression model time series statistics we couldn't find any systematic relation in in a time frame that made sense for theory and priming which is a few days you will force of the talk about times user talk about lag and lead and so you tried to connect the data point is further down the line with a data point that is but that is not as far down the line and nothing of statistical significance of uh show up here that this gas numbers we found regulators there was something that we serve took a step back and we considered another possibility to as wide to as to why but the media reported as they
did that the media just give the people what the people want and here is the why I wanna talk to you about why this was a special election this graph I that graph from the BN long pulse and they base it on the uh on surveys connect by and for the steam up uh onto the of the better than have any any access to but this impressively shows why there was a special election that in 5 out of the 6 preceding elections employment was the topic that was on top of people's minds when they mate the decision in terms of which party to vote for and employment means unemployment at in 2017 with unemployment being at record lows and after 2015 having uh or having a serious so war still going on we're having uh having having refugees come into into western uh Europe that immigration Johnson out as the as a topic that that was the most important for people that and here every if we look at all so that the topis are further down the importance scale uh for water borders all topics where 1 could conceivably think that those can be spun in a way that they are connected to this refugee situations that social justice economic injustice that something the party like DFT can very effectively turn into an ideal and group based conflict past verses them that same with pensions tho those people come here to take our jobs and our money and especially from the old people from our elderly that so 20 17 people methodologically 17 uh is a special case but if we consider it could be compared to other parts so having the situation where uh we find that it's is something that this you never happened in recent history and in Germany before in terms of what what made people decided the poles we wanted OK well is there a way to more Acula accurately measure this demand side of things this this need for information for of orders and what better way there's 2 to measure something them to measure the salience in the population uh than to look at Google queries so so we we collected google trends more specifically the cool searches on refugees fledgling in general terms um and again he this this this way to even out a trend line this is the daily change in the whole uh in how this topic developed every put our daily change of a of dimensions over that we do seem that there's something there and this some sort of systematic related uh relationship to and then uh crunching these numbers and putting them again through the electrode regression model that we come to the conclusion that with a lag of only 1 day of Google searches for refugees actually leads AFD mentions in the media so if on Tuesday a higher number of people in germany googled refugees on Wednesday the a 58 was mentioned more often the day before the the end effect wasn't big but it was there was significant we also of course but consider the alternatives and the magic word is here it's it's Granger causality so you can actually calculate uh and reliably calculate the that the temporal success success and that means that 1 follows the other that and so all of a sudden it becomes a bit difficult to point the finger is at the media uh because if the media just reaction interests it operates like a business if like out there's the normative idea of the media especially in a country that is rich in high-quality publications as Germany with the media is a public good that educates people and so and brings out the best in them in challenged them and and persuading them of the best of the argument that any other day in the online worlds is a business with a measurable outcome if clicks you've trackers you have and durations and you can measure and so you can see which articles our favorite in which articles people uh people last on that they were not saying was important bosses thinking to make your way of saying that there's Jarrett call the link between people googling refugees and the media directly reacts reacts that prompts because there's some uh some search engine optimization of guy girl every uh every every media publishing house that's met that monitors what people are interested in we're saying that there is an intermediate step there that it's not a diet cause it's just a sort of delay that is in there that allows for other mechanisms to to get in so wondering what about the consumer focusing on the demand side ends in 2017 uh there's a few things that he could actually look at to gage what uh the demand side demands and and we decided to focus on Twitter because without actually knowing as we 1st start out with collecting all this data uh we decided to set up the out uh to set up the the transcriber uh and that way between September 1st and September 24th we collected 4 . 5 billion tweets that contained that keyword to contain any 1 of a list of keywords that had that that that political notation and so looking at this body of data uh we can extract things like the top 2 200 most used hashtags uh if we do that and we we counted the tweets that contains 1 of the top 2 and hashtags uh and we pay special attention to which 1 of these hashtags are decidedly prole a of the that we get a number that 30 . 9 % of the tweets that could that contained any of those stopped hashtags actually contain 1 that is in favor of the FT whereas if we count the decidedly Knoll A of T the and TFT no way of the in all ways of spelling and capitalization and so forth there's only 1 . 2 % at and becomes uh um you becomes a bit ticklish so there are 2 sort of different better idea of of the of what vol Twitter might have played in our little uh in a little uh relationship here between the demand-side and supply-side supplies of supplying the news that we have the beautiful network graphs so this is a retweeting network this is system that we extract all the mentions of of of of of of an actor each dot is a Twitter user each line is 5 or more retweets retweets were aware of that we treat start automatically mean endorsement you might we did something as outlandish and crazy at but for the sake of visualizing what the weights are on Twitter the Treaty on is the same as anyone who ever has worked with with epigraphs of that size that take a long time to generate and it's catch to to label them so I'm very proud that he was able to do so but if you look at this island down there and uh that blob that Blueboar blob accounts that cluster around safety accounts the uh the coloring here was done by uh by Walktrap algorithms I just adjusted the
colors that that algorithm used to action match the colors and the uh in the German party landscape it so we do have the the a hefty continents at the bottom right there that connects all kinds of people to yeah to the area of the therefore look at the little at the appendix below here that is colored in brown uh that is and may be organized around a movement called reconquista Europe which is an even further right wing right wing uh movement uh that is sort of like diet attacked to this island of the FTB uh and the connecting is Jenica which is the uh quite interesting so if you have the on there is the other parties of their logo the rainbow uh that is the political landscape uh we have those those 2 extreme points there at the at the super top right and there at the bottom left that is um and so the very extremely extreme Twitter user parties it's due uh UDP and define their so they don't didn't seem to engage with uh with the nodes that are in this area that but what is also available to nodes is that for the other part is for the sellers is starting from the left orange the pod party and than red sauce Democrats purple stealing give the green and good and a yellow FTP in black uh the conservative party CDC assume the all of these parties have a central nodes a central this into account around which a lot of other users are out so there's for each party there's a smaller number of relatively small number of accounts that are highly favored in often they are retreated they've d doesn't have that even so this is of course a a projection of something that three-dimensional in a two-dimensional place so there might be some schooling going on here in terms of fraudulent on-screen but even turning it and and trying to identify which party is at the center that wasn't wasn't really possible so the internal rifts in the internal power struggles they do show in how the whole members of the party or are we treated that also interesting to note is which nodes which users are connecting these tools continent so to speak 1 is that blue dot is validates contain appalling aggregator course everyone is interested in getting the polling numbers out there and there's it's tough to see here but there is a beach uh Bayes user in the middle there uh which is Beatty so 1 of the media uh uh 1 of the media publications that we we actually collected dead answered and I think that is an adjustment mention you is that uh that light pink color at uh but light pink colored inserts between the greens and the central the gray-beige thought there those are young on your to show an example I to um yes it as a that the the dynamics are are clear that we have this this party that is pretty well organized in social media the end and thus is able to to dominate the media agenda is based on algorithms basically if you think about how other logic them of of information as a nation works and where the training aspects that if you have a party this is as as the order was organized as tightly clustered around it so within itself is the if the shows up here uh there's a good good chance that that that will influence would all of us get to see when we check out the the truth uh of the trail page and I know that it's probably a good chunk of you have a burning questions in their minds uh than going to want to know so how many of these of these bright blue bottle of blobs are box a Twitterbots that we try to find out using a tool called so of the bottle meter which is something that has an API available online where you can submit to serve project from from a research in Indiana a we can submits the name of a to the user and that gives you that read it friends at lots of lots of um Analyses and analysis lots of things about this user that done frequency of tweets the time at which a tweets who is who was it following was followed by was talking to the cost of uh but when tried to submit that I I broke the API uh and death uh so if if there had been to watch apologize it was me and so it it wasn't able that to do so in time but said there's a bunch of tasks Molotov but the exact about things I'm happy to have this sort of as a lead-in for for the day tomorrow so what can we go what can we take from this the bonus oxide 2017 was a perfect storm for a fart party like DFT yet high issue salience of the topic is at the center of its agenda the and you have a sort of unregulated wild west of uh um of of social media uh we'll see how that changes uh with with recent lodging changes come into effect where all of a sudden the the platform itself has some liability to which kind of messes spread but if that's effective for Twitter is all other bag worms so in that sense that's why I was where was soaring hinting at uh in in this issue environment we have people be interested in the topic the central of for the party like DFT is that the media behave like a a pretty surprisingly uh surprisingly predictable uh and did not of cities for the for the topic of of of of full publication that we covered and it did so that in it uh and then for the context-aware writing here and uh that's DFT only gets like 20 % of the shared towards the end of the campaign is something that is a little bit about and also leads into um into a different question of and what does this so it's the journalists faults actually mean what is it really mean at the the service is based on the normative expectation of the media being an impartial uh and partial the delivery of information and if you think about what else is going on and on the internet with with alternative media and an alternative news fear that establishing itself uh with the news blogs like uh why don't add or quality names because as a there's there's a there's a there's sort of scene of far-right fringe blogs in in Germany that we also collected uh and so we're a further aligned with a look at what the topics were that were covered in that and how that connected to to influencing public opinion in injury but having set this with these alternative ways of getting you news information that being available if you have the prosody of the mainstream does not covering a party like the a of the to a certain extent you only give the father to those cries of the compressor mandate mendacious press in and I know that members of the of the of the population are sort of uh at the risk of being lost as audience members so this is kind difficult to call that to call the shots here I a point the thing is that the media because they they delivered on informing on interests that existed in the part in the population before they reported on something like the FT that and with this only that that that I think the retention and a and highly
highly year-to-year questions and problems and ideas so we could pursue this further pH and you have the feel free to attend microphones if the microphone don't see and the cameras less of a number to the to make sense of thank you thank you very much for amazing work um after only 1 question do you plan on releasing those those collected data on what license and that's the question that we that we are so selves to um uh we would laugh to collect the data and ultimately it will happen we have to make sure that we actually have the right to do so with the way we collected it but where deathly looking into that going on with life that you hello is 1 thing that I and did something I'm from the Netherlands to compare these experiences with a the which says the experience that you know we feel that we have so long as we had time and now we have what would that and it seems that there is a major difference between the DAB day because I pressed the have frankly I don't know the name of the leader of the is produced the front fog baby you now I don't know but that them In the mammals the leaders often uh those of populist right-wing parties they were there are any good in manipulating the media they were sending out messages to stemming good in Germany was works like um that prof of propagating seminar provocations and that detected attention of the media are so there are people saying that you should react all provocations but anyway irradiated to draw the attention and I wonder whether if you have been to the same extent that active in the field of drawing attention purposely using even um the agents seized that there are specialized in advertising earning regression uh there's this idea of a DFT was very skillful at it's sort of insinuating scandal and it's purposely doing things on a public stage that would draw attention to them so for example this so this by I say again this this the expression values in column to dispose of the German politician and were uh the other leading candidate and this vital leaving a talk show what was being broadcast and so this the devil is this this this element of the of the of actually taking a scandal and using it for your own property gender uh whereas the use of ad agencies for the media campaign at the data comparing was highly professionalized um in terms of what their their posters were and how the campaign ads were worked and do work with a company that also was involved with Donald Trump's campaign um but the interval of sort of new new media or like online media so that you know more than just it online media uh I mean I just only have anecdotal total sense if the use something like uh which is also a way of of of of of buying buying intentions and I can I can sort of tell you about 1 specific case we investigated which Twitter users were the most active and tweeting on the DFT on German prayers stores a talk about where you recall by which is the which a name that has been there which is great education that that is the fully bot that has been planted and has been controlled by someone else what by a bunch of by any group of actors that is not actually a ballerina um what we found was a Twitter user called the these yourself on that we in those 3 weeks reserve at 6 thousand 500 times and mostly just retweeted um retweeted calls to go and cast your ballots that all put out by the central idea of the accounts and do not allow for a success with 500 so would just give retweeting over and over and over and over um and where we actually want to check out the other page of that part was deleted the user was deleted that so there's controversy and this is how of degree of personalization that the the the uh of the title fight has the Netherlands is not as extreme for the air in Germany because there's more leading candidates This interval rifts like cantilevers basically his own party uh in that's so that the same but the strategy to use scandal and to use something that is outrageous push the boundary is a little bit more than just like can say on no we do not mean that all of this way that is the exact same spot a strategy lies is set maps issued should active FIL which made it would like to read many people queuing OK them out out of all the things you with questions from the entombment yes MQP Keiji is asking a why you come to the conclusion that this was a special election while the last election Austria has exactly the same issues that you see this site as some sort of and global effect that you have a serious civil war the pushes people to flee from from from warrants says the likelihood of something that is not only filled in Germany but for the context of Germany special this this sort of situation is never has the records in this way before but absolutely at each election in Europe basically since 2004 15 and there was a special election in that sense uh but not in terms of the outcomes in a way that because far apart in other European countries already have um at the foot in the door and especially in Austria were with the as a work pretty well established with privacy having been part of part of a government and of being part of a government again before Germany um in what the issues were that worked of people's minds that's the special case that the microphone these thank you have 1st uh I really appreciate the sincerity and transparency of should talk thank you very much we need more of this in such circumstances and maybe less common comics sometimes and then there's trust through little trifle in your method of where I was wondering how did you filter the link and clean new stuff did your mom did you will yeah how exactly did you do it did you may be column of all dimensions of in with a capital and non-capital and link it with the capital and non-capital and then filtered out for a of what did you do it the other way around I know that you focus specifically on the of the stuff and maybe you were focused on representing all the parties that might be relevant but I would still be interested in that part thanks that's a great question and the thing is that we use that we have when when we actually put all that we have collected text before we put it through the downloading methods we put all into lower case just so we can still have a consistent way of analyzing and and with capitalization can sometimes it just strips of medium be way to treat this and therefore you have issues with with link to included where we have to resort to all means of taking this is the candidate's names and then also link time including a time and a few conjugations so uh uh dealing with time being right like for medically the cases is only like the conduit
through that you need so therefore we will hear the word the word especially so with that which is unfortunate but uh but for the purpose of this talk we decide use this work around OK thanks create migrants explains no thinks they're interesting that presentation I'm wondering if you and your team Ossietzky looked at mentions of the different parties and wondering if you looked at the content of the articles and how they talked about it here talked about positively or negatively and think about it the recursively actually consider um Al atoms question will cover questions social sciences like to do that anyone in this room using Amazon Mechanical Turk and works on hits to earn a few cents him there no OK so I can speak freely uh there's there's a method that uses a cheaper labor on Amazon Mechanical Turk and presented each worker with 2 sentences out of which they have to change the 1 that is more positive and and so we want to use this to train a machine learning algorithm to actually get away to gage the sentiments of positive and negative in the text we have collected and we saw that in early December and we we had like uh a workbook with 4 thousand so-called hits 4 thousand little jobs 4 thousand comparisons and when this job was done the 5 or 6 days later uh we we sort of put that was put the through the through test and compared it with our own hand coding that we had done and it turned out that 1 worker on Amazon Mechanical Turk spends uh over 7 hours and worked out of those 4 thousand little so we had he worked 3 thousand 980 uh and over 1 thousand 400 of which he did in less than 2 seconds that which is unfortunate because a this person did so from the bed this person that person question mark probably used a script by use the body or just randomly collect uh the coding that match up at all with what we did hand wise ourselves and and and that really screwed up our uh our approach there that if any of you plan on the on doing some hits in the new year for Amazon Mechanical Turk and you're asked to compare the 2 sentences that mention a political actor in Germany uh you can send me an e-mail and maybe a screenshot tell me how much you appreciate that were paying 6 cents for each for each comparison that but that's the that's the story we haven't we don't have any sentiments this analysis here do you know what is the very much it goes of Christmas future in its with that that's where you take sweets this world do you determine what are quotes and what are the rake retweets because in my experience and I work with this and then margin and in the UK a lot of people like to distance themselves from what you have to use similar saying by quoting everything this saying and giving them the press that's a very good point to make a and we did not make any any distinction between retweets or we did filter uh based on 5 retweets by thinking OK if you occasionally feel like you have to point something out that is outrageous and ridiculous that a person a member of parties doesn't we're and you would be inclined to do so less than a certain amount of time lost and with other cut off the graph basically always look the same them but if we think about what this means for for how the demand side is influenced it doesn't matter basically if if you retweeting out of endorsements or out of the uh out of out of spite that's right uh that's the that's the logic or we decided to to use mentions and retweets and you another question from the Internet yes 23 sigh asking do you think that the window of commonly acceptable ideas the so called all the time window was shifted to the right hand by the ideas of the 8th echoed in the media the and that's a good question that's a good question and so they is that means is that the phenomenal uh you're sort of likely but the question is like do you think that it the does something happening you because you certainly outlet or that you certainly other because something happening you know from the sense that I had uh I would say that the the degree to what is what is acceptable deftly was shifted over the course of this campaign that all of sudden were questioning of remembering the whole cost should be it should be something that is at the heart of oracles to dramatically that's something that uh um a political scientist would have never expected that this cleavage can be opened up again in a in a in a way that is so uh so potent as it now so definitely did something to you and to the overall discourse uh in Germany that words that is an effect of uh of media reporting on the FT that would require us to use something like this the sentiment analysis to actually turning to determine how the media talked about which aspects of the of the FT agendas I can see some movement behind microphone number 8 I'm sorry thank you very much and thank you for you were and still do have a Christian open questions in basically the things you showed is something like we all know that we could see this happening last year and so on I mean this year in the last elections so I'm wondering now whether the method used which was basically focusing on quantity and it's in this sort of mirroring what was happening and I'm wondering if you would to work to keep working moment of like you used by words and use the media instead of like narrowing it down on while using lost the 2nd specific I questions and I was wondering if you have this these results now and you have proved more than what are you next questions and how can you continue to use these these this is that the data you have to make it more specific so we can really have some outcome and some conclusions coming from this and actually 1 of person um of course we we thought about using this data for the Non-Aligned uh we are initial plan was to connect this uh not just with saline state that we that we derived from Google searches we also have uh a faithful there to collect that's that we want to look into but there it's a bit challenging to them judging uh analyze the comments in-depth onto language because language sees uh tends to be way more fluid and you have a certain problems with uh with selection and self selection and so you really really have to be careful to cross-connect which person that comment on Facebook that is the same person and thus if you only do qualitative stuff um would would appear disproportionately and as I mentioned we have also collected data from from far-right blocks from news blogs and that from that very actively endorsed the DFT and the topics uh and so were playing to pull this into the analysis along with the data from the German longitude elections study where in this time frame that we surveyed in the data each day uh 100 people in Germany were called up and asked about their feelings toward specific parties image and actor so we actually have day by day that data once it comes out on the on how people what people thought about all of those actors would replying to pull that in as a more reliable measure for salience thank you removed from the resort times up so there will be no more questions right no from of your audience Alexa mobile thank you very much wonderful fuite was to was you
and me and
the and and the it it a child is too but
but that back
Resultante
Bit
Ortsoperator
Selbst organisierendes System
Gruppenoperation
Soundverarbeitung
t-Test
Gruppenkeim
Versionsverwaltung
Physikalische Theorie
Internetworking
Demoszene <Programmierung>
Hypermedia
Spieltheorie
Nichtunterscheidbarkeit
M-Theorie
Soundverarbeitung
Parametersystem
Kraftfahrzeugmechatroniker
Oval
MUD
Physikalisches System
Mechanismus-Design-Theorie
Ereignishorizont
Quick-Sort
Rechenschieber
Flächeninhalt
Hypermedia
Lesen <Datenverarbeitung>
Wort <Informatik>
Nabel <Mathematik>
Momentenproblem
Iteration
Schreiben <Datenverarbeitung>
Wärmeübergang
Diskrete Fourier-Transformation
Zählen
Gruppe <Mathematik>
Computersicherheit
Skript <Programm>
E-Mail
Gerade
Auswahlaxiom
Regulator <Mathematik>
Regressionsanalyse
Kraftfahrzeugmechatroniker
Kategorie <Mathematik>
Computersicherheit
Güte der Anpassung
Natürliche Sprache
Mechanismus-Design-Theorie
Ereignishorizont
Polstelle
Rechter Winkel
Benutzerschnittstellenverwaltungssystem
Ordnung <Mathematik>
Programmierumgebung
Lesen <Datenverarbeitung>
Tabelle <Informatik>
Mathematisierung
Soundverarbeitung
Green-Funktion
Überlagerung <Mathematik>
Knotenmenge
Spannweite <Stochastik>
Zeitreihenanalyse
Migration <Informatik>
Vererbungshierarchie
Varianz
Analysis
Soundverarbeitung
Binder <Informatik>
Data Dictionary
MIDI <Musikelektronik>
Wort <Informatik>
Steuerwerk
Mittelwert
Zentralisator
Resultante
TVD-Verfahren
Bit
Prozess <Physik>
Punkt
Web log
Gemeinsamer Speicher
Hochdruck
Formale Sprache
Selbstrepräsentation
Natürliche Zahl
Computerunterstütztes Verfahren
Computer
Übergang
Stützpunkt <Mathematik>
Kurvenanpassung
Tropfen
Zentrische Streckung
Statistik
Prozess <Informatik>
Physikalischer Effekt
Plot <Graphische Darstellung>
Programmierumgebung
Dialekt
Modem
Rechenschieber
Eingebettetes System
Twitter <Softwareplattform>
Betrag <Mathematik>
Verschlingung
Geschlecht <Mathematik>
Strategisches Spiel
Aggregatzustand
Rahmenproblem
Gruppenoperation
Keller <Informatik>
Zahlenbereich
Kombinatorische Gruppentheorie
Term
Physikalische Theorie
Hypermedia
Unterring
Mittelwert
Fokalpunkt
Graph
Diskretes System
Zwei
Relativitätstheorie
Mathematisierung
Mailing-Liste
Primideal
Physikalisches System
Vektorraum
Hochdruck
Fokalpunkt
Quick-Sort
Diagramm
Flächeninhalt
Formale Sprache
Pufferüberlauf
Rationale Zahl
Hypermedia
Speicherabzug
Bit
Einfügungsdämpfung
Punkt
Web log
Minimierung
Versionsverwaltung
Gruppenkeim
Schreiben <Datenverarbeitung>
Diskrete Fourier-Transformation
Ungerichteter Graph
Twitter <Softwareplattform>
Sondierung
Internetworking
Homepage
Zahlensystem
Algorithmus
Puls <Technik>
Suchmaschine
Prozess <Informatik>
Minimum
Meter
Einflussgröße
Gerade
Regressionsanalyse
Kraftfahrzeugmechatroniker
Parametersystem
Zentrische Streckung
Datennetz
Physikalischer Effekt
Wurm <Informatik>
Temporale Logik
Frequenz
Entscheidungstheorie
Polstelle
Dienst <Informatik>
Twitter <Softwareplattform>
Rechter Winkel
Benutzerschnittstellenverwaltungssystem
Information
Extreme programming
Ordnung <Mathematik>
Programmierumgebung
Gewicht <Mathematik>
Wellenpaket
Quader
Wasserdampftafel
Gruppenoperation
Mathematisierung
Zahlenbereich
Systemplattform
Term
Mathematische Logik
Demoszene <Programmierung>
Task
Hypermedia
Erwartungswert
Weg <Topologie>
Knotenmenge
Datensatz
Lesezeichen <Internet>
Äußere Algebra eines Moduls
Maßerweiterung
Analysis
Leistung <Physik>
Soundverarbeitung
Matching <Graphentheorie>
Graph
Green-Funktion
Diskretes System
Validität
Mailing-Liste
Physikalisches System
Binder <Informatik>
Quick-Sort
Flächeninhalt
Hypermedia
Mereologie
Wort <Informatik>
Energieerhaltung
Kantenfärbung
Resultante
Bit
Punkt
Web log
Formale Sprache
Annulator
Gruppenkeim
Diskrete Fourier-Transformation
Element <Mathematik>
Internetworking
Homepage
Chatbot
Arbeit <Physik>
Arithmetischer Ausdruck
Negative Zahl
Algorithmus
Prozess <Informatik>
Trennschärfe <Statistik>
Lineare Regression
Bildschirmfenster
Skript <Programm>
E-Mail
Einflussgröße
Umwandlungsenthalpie
Softwaretest
Kategorie <Mathematik>
Systemaufruf
p-Block
Kontextbezogenes System
Neue Medien
Arithmetisches Mittel
Randwert
Datenfeld
Twitter <Softwareplattform>
Geschlecht <Mathematik>
Rechter Winkel
Strategisches Spiel
Message-Passing
Aggregatzustand
Facebook
Web Site
Subtraktion
Rahmenproblem
Zahlenbereich
Ordinalzahl
Kombinatorische Gruppentheorie
Mathematische Logik
Term
Überlagerung <Mathematik>
Virtuelle Maschine
Datensatz
Inhalt <Mathematik>
Speicher <Informatik>
Bildgebendes Verfahren
Widerspruchsfreiheit
Analysis
Beobachtungsstudie
Soundverarbeitung
Videospiel
Datenmissbrauch
Graph
Likelihood-Funktion
Zwei
Paarvergleich
Automatische Differentiation
Binder <Informatik>
Quick-Sort
Mapping <Computergraphik>
Minimalgrad
Surjektivität
Mereologie
Hypermedia
Codierung
Wort <Informatik>
Innerer Automorphismus
Orakel <Informatik>
Hypermedia
Medianwert
Systemprogrammierung

Metadaten

Formale Metadaten

Titel Pointing Fingers at 'The Media'
Untertitel The Bundestagswahl 2017 and Rise of the AfD
Serientitel 34th Chaos Communication Congress
Autor alebey
Lizenz CC-Namensnennung 4.0 International:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/34868
Herausgeber Chaos Computer Club e.V.
Erscheinungsjahr 2017
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract The German election in September 2017 brought a tectonic shift to the layout of German politics. With the AfD in parliament far-right illiberalism has reached the mainstream. We investigate the communicative developments underlying this rise. Using web-scraping and automated content analysis, we collected over 10.000 articles from mainstream-news and far-right blogs, along with over 90GBs of Tweets and thousands of Facebook-Posts. This allows us a deep insight into how public discourse works in 2017 Germany.
Schlagwörter Ethics, Society & Politics

Zugehöriges Material

Folgende Ressource ist Begleitmaterial zum Video
Video wird in der folgenden Ressource zitiert

Ähnliche Filme

Loading...
Feedback