QUANTIFYING POLITICAL LEANING FROM TWEETS, RE TWEETS, AND RE TWEETERS
Abstract—The general public’s, news media’s, and political actors’ widespread use of online social networks (OSNs) to disseminate information and exchange opinions has opened up new avenues of research in computational political science. In this paper, we investigate the problem of quantifying and inferring Twitter users’ political leanings.
We define political leaning inference as a convex optimization problem that incorporates two ideas: (a) users are consistent in their tweeting and retweeting about political issues, and (b) similar users are retweeted by similar audiences.
Then, for evaluation and quantitative analysis, we apply our inference technique to 119 million election-related tweets collected over the course of seven months during the 2012 U.S. presidential election campaign. When compared to manually created labels, our technique achieves 94 percent accuracy and high rank correlation.
Our numerical study sheds light on the political demographics of the Twitter population, as well as the temporal dynamics of political polarization as events unfold, by studying the political leanings of 1,000 frequently retweeted sources, 230,000 ordinary users who retweeted them, and the hashtags used by these sources.
1 INTRODUCTION
In recent years, large amounts of online social media data have found numerous applications in the fields of politics and computer science. Answering questions in political and social science (for example, proving or disproving the existence of media bias [3, 24] and the “echo chamber” effect [1, 5]), using online social media to predict election outcomes [37, 25], and personalizing social media feeds to provide a fair and balanced view of people’s opinions on controversial issues [30] are examples.
The ability to accurately estimate the political leanings of the population involved is required for answering the above research questions. If it is not met, the conclusion in the preceding examples will be invalid, the prediction will perform poorly [29, 31] due to a skew towards highly vocal individuals [27], or the user experience will suffer.
Accurate political leaning estimation on Twitter presents two major challenges: (a) Quantification: Is it possible to assign meaningful numerical scores to tweeters based on their political leanings? (b) Scalability: Given Twitter’s large scale and server constraints, how can we devise an efficient and scalable method? To infer the political leanings of a target set of Twitter users, we propose a new approach that incorporates the two sets of information listed below.
Tweets and retweets: the temporal patterns of target users being retweeted, as well as the tweets published by their retweeters. The insight is that a user’s tweet contents should be consistent with who they retweet; for example, if a user tweets frequently during a political event, she is expected to retweet frequently as well. This is the data’s “time series” feature.
The identities of the users who retweeted the target users are known as retweets. Because of the homophily principle, similar users are followed and retweeted by similar audiences. This is the data’s “network” aspect.
Our technical contribution is to frame political leaning inference as a convex optimization problem that maximizes tweet-retweet agreement with an error term and user similarity agreement with a regularization term that is designed to account for data heterogeneity.
The result is an inference technique that is: • Scalable: it does not require explicit knowledge of the network topology and works within Twitter API rate limits;
• Efficient: computationally efficient because it is formulated as a convex optimization problem, and data efficient because the time required to collect sufficient data to obtain good results is short; and
• Intuitive: the computed scores have a simple interpretation of “averaging,” i.e. Figure 1 shows an example. We extensively evaluate our method using a set of 119 million tweets on the 2012 U.S. presidential election collected over seven months to show that it outperforms several standard algorithms and is robust to algorithm variations.
QUANTIFYING POLITICAL LEANING FROM TWEETS, RE TWEETS, AND RE TWEETERS
Do You Have New or Fresh Topic? Send Us Your Topic
INSTRUCTIONS AFTER PAYMENT
- 1.Your Full name
- 2. Your Active Email Address
- 3. Your Phone Number
- 4. Amount Paid
- 5. Project Topic
- 6. Location you made payment from