I am Head of Graph Machine Learning at Stealth Health. Formerly, I worked at S&P Global as Associate Director and Head of Graph ML, at Spotify as Lead ML Engineer of the Music Industry Knowledge Graph, and at Graphika as founding ML Research Engineer (see my work on productionizing models covered in Built In!). At Cornell University (from where I earned my PhD in 2021), I was a Sage Fellow, and from 2018–2019 I was a NSF-funded Co-PI with Dr. Michael Macy at Cornell's Social Dynamics Laboratory. In 2018-2019, I was a NIH-funded Social Networks & Health Fellow with the Duke Network Analysis Center and was mentored by Dr. Christopher Bail.
See my GitHub profile for my projects' scripts, notebooks, and data. Many of my repos are private, however, given the sensitive or proprietary nature of the data with which I often work. In such cases, I describe how to acquire and sample data appropriately.
You can find old tutorials and reviews of specific data applications in my (now defunct) blog.
Please see my Google Scholar profile for more information on my research and links to these and similar articles. Also, visit Cornell University's Social Dynamics Laboratory, The University of Iowa's Center for the Study of Group Processes, and the University at Buffalo's Food Systems Planning and Healthy Communities Lab to see where I have previously done research!
Click here to see my CV
See my GitHub profile for my projects' scripts, notebooks, and data. Many of my repos are private, however, given the sensitive or proprietary nature of the data with which I often work. In such cases, I describe how to acquire and sample data appropriately.
You can find old tutorials and reviews of specific data applications in my (now defunct) blog.
Please see my Google Scholar profile for more information on my research and links to these and similar articles. Also, visit Cornell University's Social Dynamics Laboratory, The University of Iowa's Center for the Study of Group Processes, and the University at Buffalo's Food Systems Planning and Healthy Communities Lab to see where I have previously done research!
Click here to see my CV
~ Publications ~
Alexander Ruch, Ari Decter-Frain, Raghav Batra. "Millions of Co-purchases and Reviews Reveal the Spread of Polarization and Lifestyle Politics across Online Markets." ArXiv. Featured in The New York Times and on National Public Radio.
Polarization in America has reached a high point as markets are also becoming polarized. Existing research, however, focuses on specific market segments and products and has not evaluated this trend’s full breadth. If such fault lines do spread into other segments that are not explicitly political, it would indicate the presence of lifestyle politics – when ideas and behaviors not inherently political become politically aligned through their connections with explicitly political things. We study the pervasiveness of polarization and lifestyle politics over different product segments in a diverse market and test the extent to which consumer- and platform-level network effects and morality may explain lifestyle politics. Specifically, using graph and language data from Amazon (82.5M reviews of 9.5M products and product and category metadata from 1996–2014), we sample 234.6 million relations among 21.8 million market entities to find product categories that are most politically relevant, aligned, and polarized. We then extract moral values present in reviews’ text and use these data and other reviewer-, product-, and category-level data to test whether individual- and platform-level network factors explain lifestyle politics better than products’ implicit morality. We find pervasive lifestyle politics. Cultural products are 4 times more polarized than any other segment, products’ political attributes have up to 3.7 times larger associations with lifestyle politics than author-level covariates, and morality has statistically significant but relatively small correlations with lifestyle politics. Examining lifestyle politics in these contexts helps us better understand the extent and root of partisan differences, why Americans may be so polarized, and how this polarization affects market systems.
Alexander Ruch, Yujia Zhang, Michael Macy. "Demographic Confounding Causes Extreme Instances of Lifestyle Politics." ArXiv.
Lifestyle politics emerge when activities that have no substantive relevance to ideology become politically aligned and polarized. Homophily and social influence are able generate these fault lines on their own; however, social identities from demographics may serve as coordinating mechanisms through which lifestyle politics are mobilized and spread. Using a dataset of 137,661,886 observations from 299,327 Facebook interests aggregated across users of different racial/ethnic, education, age, gender, and income demographics, we find that the most extreme instances of lifestyle politics are those which are highly confounded by demographics such as race/ethnicity (e.g., Black artists and performers) and age (e.g., immigration). Decomposing political alignment for these demographic effects led lifestyle politics to decrease 27.36% toward the political center. Moreover, this led demographically confounded interests to no longer appear among the most polarized interests. Instead, the most liberal interests included electric cars, Planned Parenthood, and liberal satire and the most conservative interests included the Republican Party and conservative commentators. We validate our political alignment and lifestyle politics measures using the General Social Survey and find similar demographic entanglements with lifestyle politics existed before social media such as Facebook were ubiquitous, giving us strong confidence that our results are not due to echo chambers or filter bubbles. Likewise, since demographic characteristics exist prior to ideological values, we argue that these demographic effects we observe are causally responsible for the extreme instances of lifestyle politics that we find among the aggregated interests. We conclude our paper by relating our results to Simpson’s paradox, cultural omnivorousness, evolving belief networks, and opinion cascades.
Vladimir Barash, Polina Kolozaridi, Dmitry Muravyov, Konstantin Gabov, Margarita Kiryshnina, Alexandra Goncharova, Alexander Ruch. 2021. "Evolution of the Digitally Mediated Public Sphere in Russia, 2012-2018." SSRN. *Made Top Ten download list for InfoSciRN*
This paper uses a social media perspective to describe the evolution of the Russian social and political landscape in cyberspace over a period of seven years, which no other paper to date has mapped and explored. We use network analysis and language models to compare Russian users of Twitter and Facebook from 2012 to 2018. Over this period, politically active communities grew ten-fold in prevalence and presently dominate discussion in Russian cyberspace. Russian state-sponsored media and disinformation outlets monopolize the Russian landscapes of both Twitter and Facebook platforms. For example, content pushed by pro-Putin communities is highly focused on the Ukraine conflict and supported by official government accounts, whereas content from anti-Putin communities is on a wide range of topics and lacks official government support. These advantages enable pro-Putin communities to reach a bigger audience and influence the Russian population more effectively relative to anti-Putin communities. Language models corroborate these findings and show an increasing prevalence of political topics in communication among both groups over time. Together, these results expand our understanding of political communication in cyberspace via network and language dynamics and provide a richer context for understanding and explaining the results of recent work on the social effects of online disinformation campaigns and government efforts to shape social media and discourse.
Alexander Ruch. 2020. "Can x2vec Save Lives? Integrating Graph and Language Embeddings for Automatic Mental Health Classification." Journal of Physics: Complexity
Graph and language embedding models are becoming commonplace in large scale analyses given their ability to represent complex sparse data densely in low-dimensional space. Integrating these models' complementary relational and communicative data may be especially helpful if predicting rare events or classifying members of hidden populations – tasks requiring huge and sparse datasets for generalizable analyses. For example, due to social stigma and comorbidities, mental health support groups often form in amorphous online groups. Predicting suicidality among individuals in these settings using standard network analyses is prohibitive due to resource limits (e.g., memory), and adding auxiliary data like text to such models exacerbates complexity- and sparsity-related issues. Here, I show how merging graph and language embedding models (metapath2vec and doc2vec) avoids these limits and extracts unsupervised clustering data without domain expertise or feature engineering. Graph and language distances to a suicide support group have little correlation (ρ < 0.23), implying the two models are not embedding redundant information. When used separately to predict suicidality among individuals, graph and language data generate relatively accurate results (69% and 76%, respectively) but have moderately large false-positive (25% and 21%, respectively) and false-negative (38% and 27%, respectively) rates; however, when integrated, both data produce highly accurate predictions (90%, with 10% false-positives and 12% false-negatives). Visualizing graph embeddings annotated with predictions of potentially suicidal individuals shows the integrated model could classify such individuals even if they are positioned far from the support group. These results extend research on the importance of simultaneously analyzing behavior and language in massive networks and efforts to integrate embedding models for different kinds of data when predicting and classifying, particularly when they involve rare events.
Liz McQuillan, Erin McAweeney, Alicia Bargar, Alexander Ruch. 2020. "Cultural Convergence: Insights into the Behavior of Misinformation Networks on Twitter." SBP-BRiMS 2020
How can the birth and evolution of ideas and communities in a network be studied over time? We use a multimodal pipeline, consisting of network mapping, topic modeling, bridging centrality, and divergence to analyze Twitter data surrounding the COVID-19 pandemic. We use network mapping to detect accounts creating content surrounding COVID-19, then Latent Dirichlet Allocation to extract topics, and bridging centrality to identify topical and non-topical bridges, before examining the distribution of each topic and bridge over time and applying Jensen-Shannon divergence of topic distributions to show communities that are converging in their topical narratives.
Michael Macy, Sebastian Deri, Alexander Ruch, Natalie Tong. 2019. "Opinion Cascades and the Unpredictability of Partisan Polarization." Science Advances 5(8):1-7.
“Culture wars” involve the puzzling alignment of partisan identity with disparate policy positions as well as lifestyle choices and personal morality. Explanations point to deep-rooted ideological divisions, core values, moral emotions, and cognitive hardwiring. We used the “multiple worlds” experimental paradigm to test an alternative explanation based on the sensitivity of opinion cascades to the initial conditions. Two online experiments (N=4581) generated cascades by exposing participants to social influence on 20 novel political and cultural issues. Consistent with recent studies, partisan divisions in the influence condition were much larger than in the control group (without influence). The surprise is that bigger divisions do not mean greater predictability. An issue backed by Republicans and opposed by Democrats in one experimental “world” had the opposite outcome in another parallel world. The results suggest that ideology may be used to rationalize positions arrived at through a tipping process that might just as easily have tipped the other way. Public awareness of this counter-intuitive possibility has the potential to encourage greater tolerance for alternative opinions.
Alexander Ruch. 2016. "Perceived Organizational Risks and Reputations Are Related to Individuals’ Decisions to Eat Genetically Modified Foods." MA Thesis published through the University of Iowa.
Sociologists have studied how organizations respond to perceived risks, but overlooked how individuals react to perceptions of organizational risks. We may expect individuals to avoid the goods and services of supposedly risky organizations, but how do other social judgments of organizations, such as those concerning reputation, relate to individuals’ risk aversion independently from their perceptions of risk? Social psychological theories on legitimacy and status and psychological theories on risk perception can bridge these gaps. Using data from the 2006 General Social Survey, this paper tests how individuals’ aversion to genetically modified foods (GMOs) relates to their perceptions of organizational risks and other qualities of business leaders, medical researchers, and political officials who are involved with producing, evaluating, and regulating GMOs. Logistic regression models find that individuals’ perceptions of medical researchers’ ignorance and disagreement about GMOs’ possible risks synergistically interact to increase the probability of rejecting GMOs. Individuals’ deferral of political influence to medical researchers attenuated the increased odds of rejecting GMOs among individuals who believe that industry scientists are disreputable. Surprisingly, perceived risks among business and political leaders were unrelated to GMO aversion. These results extend sociological risk research by demonstrating how individuals’ responses to perceived organizational risks are shaped by social characteristics such as reputations. Finally, links are drawn to inform social movement literatures and debates on GMOs, as reputational correlates exist independently from individuals’ knowledge of science, environmentalism, and generalized trust.
Alexander Ruch and Marina Zaloznaya. 2016. "Corruption Discourse in Current Debates on Genetically Modified Foods: Moral and Ethical Challenges to Law and Science." Pp. 233-265 in Corruption: Political, Economic, and Social Issues. New York, NY: Nova Science Publishers.
Social movements often use corruption accusations to discredit their opponents. In this chapter, we argue that corruption scandals offer an excellent opportunity to study what happens when moral understandings conflict with positivistic maxims based on science and law. We analyze three cases of alleged corruption related to genetically modified foods where corruption claims based on ethical-critical logics were confronted with objective-formal counter-arguments. In the first case, several scientists accused Food and Chemical Toxicology journal editors of corruptly retracting their 2012 study on GMO toxicity. This instance of corruption contestation ended in circumvention of ethical-critical logics by objective-formal framings when the journal editors and GMO proponents successfully argued that moral arguments were irrelevant to the debate since the standards of scientific evidence, allegedly, justified the retraction. In the second case, activists accused a private pro-GMO industry group of corruption after it spent $10.6 million to campaign against an initiative to label GMOs in Washington State. This confrontation resulted in the partial accommodation of ethical-critical arguments: the industry group immediately disclosed the sources and sizes of their contributions, but proceeded to launch a counter-suit and defeat the initiative. The last case of alleged corruption involved the controversial passing of public legislation that limited the role of courts in regulating GMOs. The confrontation between activists’ ethical-critical logics and pro-GMO politicians’ defenses based on objective-formal arguments resulted in the temporary silencing of the former. Although GMO proponents initially overcame the opposition, the law failed when it came up for renewal six months later. We explain this variation in the outcomes of corruption-related discursive contestations with the different ratio of public and private actors across the three cases. Our analysis suggests that the larger the presence of public actors in a given debate, the more bargaining power ethical-critical logics have relative to objective-formal claims.
Alison Bianchi, Alexander Ruch, Michael Ritter, and Ji Hye Kim. 2016. "Emotion Management: Unexpected Research Opportunities." Sociology Compass (10)2:172-183.
During the process of emotion management, individuals perceive that they are feeling emotions that differ from what is expected within the situation. Consequently, they use cognitive, physical, and/or other means either to display more appropriate emotions or to change their emotions on a deeper level to be consistent with what is customarily expressed. Beginning with the first examinations of emotion management in 1979 by the pioneer Arlie Hochschild, emotion scholars have produced over 6,000 studies of this phenomenon. We join this vibrant research program by proposing new avenues of research using an interdisciplinary strategy. First, we explore possibilities for emotion management research within its “home base” of sociology; then, we branch out to the areas of morality and political science. In so doing, we craft new and unexpected pathways for advancements in theory, theory adjudication, and methodology, for the future of emotion management research.
~ Work in Progress ~
Ian Paul, Alexander Ruch. "In the Middle of the Ride: Mapping Mental Health Arcs Before and After Crises." Revising.
Advances in natural language processing and rising engagement with social media make it possible to study individuals’ social and behavioral relationships through the digital footprints they leave behind in online interactions. Support groups for those struggling with mental health have formed on platforms such as Reddit. We investigate whether the language expressed by users who post in a suicide support group can be classified into distinct trajectories using unsupervised learning and find four core arcs. These results are robust to whether clustering is based on centroid or connectivity. We also analyze these trends by quantifying users’ language similarities with two variants of the Paragraph Vector model. Lastly, we examine the relative changes in language that occur in each cluster arc and relate these shapes to models of crises that have been studied in psychiatric literature.
Alexander Ruch, Seunghyun Kim, Jennifer Ruch. "Not All Support Is the Same: Multilabel Psychiatric Classification and Prediction of Social Support among Members of an Online Suicide Support Group." Presented at the 2018 Cornell University Sociology Research Symposium. Revising.
Suicide is a leading cause of death in the US. SuicideWatch, a social support group on Reddit, provides an online source of help for people struggling with suicide and mental health crises. Using SuicideWatch submissions, we estimate the prevalence of suicidality and different psychiatric conditions associated with it and then test if users who struggle with different psychiatric conditions receive different amounts of social support from community members. We estimate that 90% of submissions to the group may be classified as suicidal and over 95% of submissions present with anxiety and/or depression. A negative binomial regression model finds significant differences in how many comments are posted to submissions labeled with different conditions. Dissociative labels receive the fewest comments, and psychotic labels receive the most. We conclude by discussing our results in light of suicide support groups’ mission and by suggesting how such inequalities may be circumvented.
Alexander Ruch, Juliana Hong, Hannah Lee. "Lifestyle Politics on Facebook: Political and Demographic Polarization and the Interests that Bring Us together." Presented at Cornell University's 2019 BOOM Conference in Ithaca, NY.
We constructed a unique dataset of more than 223,000 interest topics from Facebook.com and collected aggregated data on the number of followers each interest topic has by a large number of political and demographic categories. Our goals are to examine which Facebook interests are the most/least politically polarized, to decompose these estimates of observed political polarization by demographic polarization, and to assess which of the politically polarized interest topics are most confounded by demographic polarization. We will use graph-based methods of unsupervised learning to classify interest topics into larger interest categories, after which we will identify which interests within polarized categories are the least polarized across parties and thus serve as cultural bridges between partisan groups.
~ Side Projects/Presentations ~
Alexander Ruch and Ion Vasi. "The Empire Strikes Back: Activism, Industry Mobilization, and the Adoption of Pro-GMO Policies." Paper presented at the 2017 Annual Meeting of the American Sociological Association Annual Meeting, Palais des Congres de Montreal, Montreal, Canada.
Research on social movement outcomes has typically focused on the intended consequences of social movements. In contrast, we contribute to theories of social movements, political sociology, and organizations by examining how movements’ successes result in unintended consequences and trigger countermovement mobilizations and the passage of opposing policies. We do this by analyzing the biotechnology industry’s responses to activism against genetically modified organisms (GMOs). Using an original dataset assembled from various sources, we conduct quantitative analyses of both the introduction and adoption of pro-GMO policies between 2001 and 2015. Preliminary results suggest that these policies were introduced and adopted in states where anti-GMO activists had achieved some successes, not in states where the movement had a weak infrastructure or where the political opportunity structures were more favorable to the industry. These findings suggest that the biotechnology industry’s strategy has been to go head-to-head against the anti-GMO activists in states where it was under attack.
Alexander Ruch and Jennifer Ruch. "Food Environments and Diet: A Multilevel Longitudinal Analysis of Farmers' Markets and Community Health."In this paper, I use multilevel longitudinal models with time-varying covariates to test how changes in farmers’ markets and neighborhood structure affected the diets of 95,000 individuals from 180 communities over time.
|
Alexander Ruch. "Better than Bad: Halo and Devil Effects in Moral Judgments of Organizations."Using surveys, experiments, text analysis, and observations of a protest, I am studying how morality, identity, beliefs about science, and perceived risks shape individuals’ moral judgments of corporations and their products.
|