ASArthur Spirling

Arthur Spirling

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Arthur Spirling has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Measuring Distances in High Dimensional Spaces

    by Breanna Green, William Hobbs, Sofia Avila, Pedro L. Rodriguez, Arthur Spirling, Brandon M. Stewart

    Analysts often seek to compare representations in high-dimensional space, e.g., embedding vectors of the same word across groups. We show that the distance measures calculated in such cases can exhibit considerable statistical bias, that stems from uncertainty in the estimation of the elements of those vectors. This problem applies to Euclidean distance, cosine similarity, and other similar measures. After illustrating the severity of this problem for text-as-data applications, we provide and validate a bias correction for the squared Euclidean distance. This same correction also substantially reduces bias in ordinary Euclidean distance and cosine similarity estimates, but corrections for these measures are not quite unbiased and are (non-intuitively) bimodal when distances are close to zero. The estimators require obtaining the variance of the latent positions. We (will) implement the estimator in free software, and we offer recommendations for related work.


  • Open Access

    Multilanguage Word Embeddings for Social Scientists: Estimation, Inference, and Validation Resources for 157 Languages

    by Elisa M. Wirsching, Pedro L. Rodriguez, Arthur Spirling, Brandon M. Stewart

    Word embeddings are now a vital resource for social science research. However, obtaining high-quality training data for non-English languages can be difficult, and fitting embeddings therein may be computationally expensive. In addition, social scientists typically want to make statistical comparisons and do hypothesis tests on embeddings, yet this is nontrivial with current approaches. We provide three new data resources designed to ameliorate the union of these issues: (1) a new version of fastText model embeddings, (2) a multilanguage “a la carte” (ALC) embedding version of the fastText model, and (3) a multilanguage ALC embedding version of the well-known GloVe model. All three are fit to Wikipedia corpora. These materials are aimed at “low-resource” settings where the analysts lack access to large corpora in their language of interest or to the computational resources required to produce high-quality vector representations. We make these resources available for 40 languages, along with a code pipeline for another 117 languages available from Wikipedia corpora. We extensively validate the materials via reconstruction tests and other proofs-of-concept. We also conduct human crowdworker tests for our embeddings for Arabic, French, (traditional Mandarin) Chinese, Japanese, Korean, Russian, and Spanish. Finally, we offer some advice to practitioners using our resources.


  • Top Cited Papers
    Open Access

    Using proprietary language models in academic research requires explicit justification

    by Alexis Palmer, Noah A. Smith, Arthur Spirling


  • Top Cited Papers
    Open Access

    Embedding Regression: Models for Context-Specific Description and Inference

    by Pedro L. Rodriguez, Arthur Spirling, Brandon M. Stewart

    Social scientists commonly seek to make statements about how word use varies over circumstances—including time, partisan identity, or some other document-level covariate. For example, researchers might wish to know how Republicans and Democrats diverge in their understanding of the term “immigration.” Building on the success of pretrained language models, we introduce the à la carte on text (conText) embedding regression model for this purpose. This fast and simple method produces valid vector representations of how words are used—and thus what words “mean”—in different contexts. We show that it outperforms slower, more complicated alternatives and works well even with very few documents. The model also allows for hypothesis testing and statements about statistical significance. We demonstrate that it can be used for a broad range of important tasks, including understanding US polarization, historical legislative development, and sentiment detection. We provide open-source software for fitting the model.


  • Text Analysis: Text as Data with R

    by Arthur Spirling

    This seminar introduces “text as data” statistical methods using R. The course is very applied, with the primary aim of helping social science researchers understand the types of questions we can ask with text, and how to answer them. The seminar covers how texts may be modeled and compared as quantitative entities, and then moves to supervised and unsupervised methods—including topic models and embeddings. At the seminar's conclusion, participants will know how conduct their own text as data research projects. An official Instats certificate of completion is provided at the conclusion of the seminar.

  • Text Analysis: Text as Data with R

    by Arthur Spirling

    This seminar introduces “text as data” statistical methods using R. The course is very applied, with the primary aim of helping social science researchers understand the types of questions we can ask with text, and how to answer them. The seminar covers how texts may be modeled and compared as quantitative entities, and then moves to supervised and unsupervised methods—including topic models and embeddings. At the seminar's conclusion, participants will know how conduct their own text as data research projects. An official Instats certificate of completion is provided at the conclusion of the seminar. For European PhD students, the seminar offers 2 ECTS Equivalent point.

  • conText: 'a la Carte' on Text (ConText) Embedding Regression

    by Pedro L. Rodriguez, Arthur Spirling, Brandon Stewart, Sofia Avila

  • Turning History into Data: Data Collection, Measurement, and Inference in HPE

    by Alexandra Cirone, Arthur Spirling

    There are a number of challenges that arise when working with historical data. On one hand, scholars often find themselves with too much archival data to read, code, or compile into large-N datasets; on the other hand, scholars often find themselves dealing with too little information and problems of missing data. Selection bias, time decay, confirmation bias, and lack of contextual knowledge can also be potential obstacles. This paper serves to identify common threats to inference when performing historical data collection, and provide a number of best practices that can guide potential scholars of historical political economy. We also discuss new advances in data digitization, text-as-data, and text analysis that allow for the quantitative exploration of historical material.


  • New evidence and new methods for analyzing the Iranian revolution as an intelligence failure

    by Matthew Connelly, Raymond Hicks, Robert Jervis, Arthur Spirling

    The U.S. government failure to predict the Shah’s 1979 fall is a prototypical example of intelligence failure. We complement the close reading of memoirs and State Department documents with computational analysis of the documents in the aggregate, including cables to and from embassies and consulates in Iran. Using several techniques including traffic and sentiment analysis, we find that officials in Iran reported on the protests but did not stress the situation’s severity until too late. D.C. officials were distracted by other events. Text analysis can complement qualitative approaches to more clearly indicate what good intelligence reporting can and cannot achieve.


  • Open Access

    Diplomatic documents data for international relations: the Freedom of Information Archive Database

    by Matthew J Connelly, Raymond Hicks, Robert Jervis, Arthur Spirling, Clara H Suong

    We introduce the Freedom of Information Archive (FOIArchive) Database, a collection of over 3 million documents about state diplomacy. Substantively, our database focusses on the USA and provides opportunities to analyze previously classified (or publicly unavailable) corpora of internal government documents which include the raw—often full—text of those documents. We also provide within-country diplomatic records for the USA, UK, and Brazil. The full span of the data is 1620–2013, but it is mainly from the twentieth century. Our database allows scholars to view text and associated statistics online and to download and view customized datasets via an application programming interface. We provide extensive metadata about the documents, including the countries and persons they mention, and their topics and classification levels. The metadata includes information we extracted with domain-specific, customized natural language processing tools. To demonstrate the potential of this data, we use it to design and validate a new index for “country importance” in the context of US foreign policy priorities.