6 October 2018 in Journal of Open Source Software
Quanteda is an R package providing a comprehensive workflow and toolkit for natural language processing tasks such as corpus management, tokenization, analysis, and visualization. It has extensive functions for applying dictionary analysis, exploring texts using keywords-in-context, computing document and feature similarities, and discovering multi-word expressions through collocation scoring. Based entirely on sparse operations, it provides highly efficient methods for compiling document-feature matrices and for manipulating these or using them in further quantitative analysis. Using C++ and multithreading extensively, quanteda is also considerably faster and more efficient than other R and Python packages in processing large textual data.
1 November 2020 in Communication Methods and Measures
Many social scientists recognize that quantitative text analysis is a useful research methodology, but its application is still concentrated in documents written in European languages, especially English, and few sub-fields of political science, such as comparative politics and legislative studies. This seems to be due to the absence of flexible and cost-efficient methods that can be used to analyze documents in different domains and languages. Aiming to solve this problem, this paper proposes a semisupervised document scaling technique, called Latent Semantic Scaling (LSS), which can locate documents on various pre-defined dimensions. LSS achieves this by combining user-provided seed words and latent semantic analysis (word embedding). The article demonstrates its flexibility and efficiency in large-scale sentiment analysis of New York Times articles on the economy and Asahi Shimbun articles on politics. These examples show that LSS can produce results comparable to that of the Lexicoder Sentiment Dictionary (LSD) in both English and Japanese with only small sets of sentiment seed words. A new heuristic method that assists LSS users to choose a near-optimal number of singular values to obtain word vectors that best capture differences between documents on target dimensions is also presented.
2 January 2017 in The Journal of International Communication
The description of the Ukraine crisis as an ‘information war’ in recently published studies seems to suggest a belief that the Russian government’s propaganda in the crisis contributed to Russia’s swift annexation of Crimea. However, studies focusing on Russia’s state-controlled media fail to explain how Russian’s narrative spread beyond the ‘Slavic world’. This study, based on quantitative and qualitative analyses of news coverage by ITAR-TASS, Reuters, the Associated Press, and Agence France-Presse over two years, reveals that Russian’s narratives were internationally circulated in news stories published by a western news agency. Although this by no means suggests that the western news agency was complicit in Russia’s propaganda effort, these news stories were published on the most popular online news sites, such as Yahoo News and Huffington Post. These findings highlight the vulnerability of today’s global news gathering and distribution systems, and the rapid changes in relationships between states and corporations in the media and communications industry.
29 May 2023 in Social Science Computer Review
Topic models have been widely used by researchers across disciplines to automatically analyze large textual data. However, they often fail to automate content analysis, because the algorithms cannot accurately classify individual sentences into pre-defined topics. Aiming to make topic classification more theoretically grounded and content analysis in general more topic-specific, we have developed Seeded Sequential Latent Dirichlet allocation (LDA), extending the existing LDA algorithm, and implementing it in a widely accessible open-source package. Taking a large corpus of speeches delivered by delegates at the United Nations General Assembly as an example, we explain how our algorithm differs from the original algorithm; why it can classify sentences more accurately; how it accepts pre-defined topics in deductive or semi-deductive analysis; how such ex-ante topic mapping differs from ex-post topic mapping; how it enables topic-specific framing analysis in applied research. We also offer practical guidance on how to determine the optimal number of topics and select seed words for the algorithm.
20 March 2017 in European Journal of Communication
Objectivity in news reporting is one of the most widely discussed topics in journalism, and a number of studies on bias in news have been conducted, but there is little agreement on how to define or measure news bias. Aiming to settle the theoretical and methodological disagreement, the author redefined news bias and applied a new methodology to detect the Russian government’s influence on ITAR-TASS during the Ukraine crisis. A longitudinal content analysis of over 35,000 English-language newswires on the Ukraine crisis published by ITAR-TASS and Interfax clearly showed that ITAR-TASS’ framing of Ukraine was reflecting desirability of pivotal events in the crisis to the Russian government. This result reveals Russia’s strategic use of the state-owned news agency for international propaganda in its ‘hybrid war’, demonstrating the effectiveness of the new approach to news bias.
2 March 2017 in Digital Journalism
This paper presents the results of an evaluation of three different types of geographical news classification methods: (1) simple keyword matching, a popular method in media and communications research; (2) geographical information extraction systems equipped with named-entity recognition and place name disambiguation mechanisms (Open Calais and Geoparser.io); and (3) a semi-supervised machine learning classifier developed by the author (Newsmap). Newsmap substitutes manual coding of news stories with dictionary-based labelling in the creation of large training sets to extract large numbers of geographical words without human involvement and it also identifies multi-word names to reduce the ambiguity of the geographical traits fully automatically. The evaluation of classification accuracy of the three types of methods against 5000 human-coded news summaries reveals that Newsmap outperforms the geographical information extraction systems in overall accuracy, while the simple keyword matching suffers from ambiguity of place names in countries with ambiguous place names.