DOCSLIB.ORG

Cross-Lingual Link Discovery Between Chinese and English Wiki Knowledge Bases

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Cross-Lingual Link Discovery Between Chinese and English Wiki Knowledge Bases PACLIC-27 Cross-lingual Link Discovery between Chinese and English Wiki Knowledge Bases Qingliang Miao, Huayu Lu, Shu Zhang, Yao Meng Fujitsu R&D Center Co., Ltd. No.56 Dong Si Huan Zhong Rd, Chaoyang District, Beijing P.R. China {qingliang.miao, zhangshu, mengyao}@cn.fujitsu.com [email protected] Abstract of monolingual and cross-lingual alignment in Chinese and English Wikipedia. As it can be Wikipedia is an online multilingual seen that there are 2.6 millions internal links encyclopedia that contains a very large within English Wikipedia and 0.32 millions number of articles covering most written internal links within Chinese Wikipedia, but only languages. However, one critical issue 0.18 millions links between Chinese Wikipedia for Wikipedia is that the pages in pages to English ones. For example, in Chinese different languages are rarely linked 武术 except for the cross-lingual link between Wikipedia page “ (Martial arts)”, anchors are pages about the same subject. This could only linked to related Chinese articles about pose serious difficulties to humans and different kinds of martial arts such as “ 拳击 machines who try to seek information (Boxing)”, “柔道(Judo)” and “击剑(Fencing)”. from different lingual sources. In order But, there is no anchors linked to other related to address above issue, we propose a English articles such as “Boxing”, “Judo and hybrid approach that exploits anchor “Fencing”. This makes information flow and strength, topic relevance and entity knowledge propagation could be easily blocked knowledge graph to automatically between articles of different languages. discovery cross-lingual links. In addition, we develop CELD, a system for automatically linking key terms in Chinese documents with English Concepts. As demonstrated in the experiment evaluation, the proposed model outperforms several baselines on the NTCIR data set, which has been designed especially for the cross-lingual link discovery evaluation. 1 Introduction Wikipedia is the largest multilingual encyclopedia online with over 19 million articles Figure 1. Statistics of English to English links in 218 written languages. However, the anchored (E2E), Chinese to Chinese links (C2C) and links in Wikipedia articles are mainly created Chinese to English links (C2E). within the same language. Consequently, Consequently, automatically creating cross- knowledge sharing and discovery could be lingual links between Chinese and English impeded by the absence of links between Wikipedia would be very useful in information different languages. Figure 1 shows the statistics flow and knowledge sharing. At present, there 374 Copyright 2013 by Qingliang Miao, Huayu Lu, Shu Zhang, and Yao Meng 27th Pacific Asia Conference on Language, Information, and Computation pages 374－381 PACLIC-27 are several monolingual link discovery tools for problem is non-trivial and poses a set of English Wikipedia, which assist topic curators in challenges. discovering prospective anchors and targets for a Linguistic complexity given Wikipedia pages. However, no such cross- lingual tools yet exist, that support the cross- Chinese Wikipedia is more complex, because lingual linking of documents from multiple contributors of Chinese Wikipedia are from languages (Tang et al., 2012). As a result, the different Chinese spoken geographic areas and work is mainly taken by manual, which is language variations. For example, Yue dialect1 is obviously tedious, time consuming, and error a primary branch of Chinese spoken in southern prone. China and Wu 2 is a Sino-Tibetan language One way to solve above issue is cross-lingual spoken in most of southeast. Moreover, these link discovery technology, which automatically contributors cite modern and ancient sources creates potential links between documents in combining simplified and traditional Chinese different languages. Cross-lingual link discovery text, as well as regional variants (Tang et al., not only accelerates the knowledge sharing in 2012). Consequently, it is necessary to normalize different languages on the Web, but also benefits words into simple Chinese before cross-lingual many practical applications such as information linking. retrieval and machine translation (Wang et al., Key Term Extraction 2012). In existing literature, a few approaches have been proposed for linking English There are different kinds of key term ranking Wikipedia to other languages (Kim and methods that could be used in key term Gurevych, 2011; Fahrni et al., 2011). Generally extraction, such as tf-idf, information gain, speaking, there are three steps for Cross-lingual anchor probability and anchor strength (Kim and link discovery: (1) Apply information extraction Gurevych, 2011). How to define a model to techniques to extract key terms from source incorporate both the global statistical language documents. (2) Utilize machine characteristics and topically related context translation systems to translate key terms and together? source documents into target language. (3) Apply Translation entity resolution methods to identify the corresponding concepts in target language. Key term translation could rely on bilingual However, in key term extraction step, most dictionary and machine translation. This kind of works rely on statistical characteristics of anchor methods could obtain high precision, while suffer text (Tang et al., 2012), but ignore the topic from low recall. When using larger dictionaries relevance. In this case, common concepts are or corpus for translation, it is prone to introduce selected as key terms, but these terms are not translation ambiguities. How to increase recall related to the topic of the Wikipedia page. For without introducing additional ambiguities? example, in Chinese Wikipedia page “ 武术 In order to solve the above challenges, we (Martial arts)”, some countries’ name such as investigate several important factors of cross- “ 中国(China)”, “ 日本(Japan)” and “ 韩国 lingual link discovery problem and propose a (Korea)” are also selected as key terms when hybrid approach to solve the above issues. Our using anchor statistics. For term translation, contributions include: existing methods usually depends on machine (1) We develop a normalization lexicon for translation, and suffers from translation errors, Chinese variant character. This lexicon could particularly those involving named entities, such be used for traditional and simplified Chinese as person names (Cassidy et al., 2012). Moreover, transformation and other variations machine translation systems are prone to normalization. We also discovery entity introduce translation ambiguities. In entity knowledge from Wikipedia, Chinese resolution step, some works use simple title encyclopedia, and then we build a knowledge matching to find concept in target languages, graph that includes mentions, concepts, which could not distinguish ambiguous entities translations and corresponding confidence effectively (Kim and Gurevych, 2011). scores. In this paper, we try to investigate the (2) We present an integrated model for key problem of cross-lingual link discovery from terms extraction, which leverages anchor Chinese Wikipedia pages to English ones. The 1 http://zh-yue.wikipedia.org 2 http://wuu.wikipedia.org 375 PACLIC-27 statistical probability information and topical the ID of the knowledge base entry to which the relevance. Efficient candidate selection name refers; or NIL if there is no such method and distinguishing algorithm enable knowledge base entry. Due to the intrinsic this model meet the real-time requirements. ambiguity of named entities, most works in (3) We implement a system and evaluate it entity linking task focus on named entity using NTCIR cross-lingual links discovery disambiguation. For example, Han and Sun dataset. Comparing with several baselines, (2012) propose a generative entity-topic model our system achieves high precision and recall. that effectively joins context compatibility and The remainder of the paper is organized as topic coherence. Their model can accurately follows. In the following section we review the disambiguate most mentions in a document using existing literature. Then, we formally introduce both the local information and the global the problem of cross-lingual link discovery and consistency. some related concepts in section 3. We introduce Following this research stream, researchers the proposed approach in section 4. We conduct have been paying more and more attention on comparative experiments and present the cross-lingual semantic annotation (CLSA). experiment results in section 5. At last, we Knowledge Base Population (KBP2011) conclude the paper with a summary of our work evaluations propose a cross-lingual entity link and give our future working directions. task, which aims to find link between Chinese queries and English concepts. NTCIR9 cross- 2 Related Works lingual link discovery task is another kind of cross-lingual semantic annotation. These two Generally speaking, link discovery is a kind of tasks are different in query selection criteria, semantic annotation (Kiryakov et al., 2004), leading to different technical difficulties and which is characterized as the dynamic creation of concerns. In KBP2011, key terms are manually interrelationships between concepts in selected to cover many ambiguous entities and knowledge base and mentions in unstructured or name variants. Consequently, disambiguation is semi-structured documents (Bontcheva and Rout, crucial in KBP2011. While in NTCIR9, 2012). participants have to extract key terms
Load more

Recommended publications
  • Cultural Anthropology Through the Lens of Wikipedia: Historical Leader Networks, Gender Bias, and News-Based Sentiment
    Cultural Anthropology through the Lens of Wikipedia: Historical Leader Networks, Gender Bias, and News-based Sentiment Peter A. Gloor, Joao Marcos, Patrick M. de Boer, Hauke Fuehres, Wei Lo, Keiichi Nemoto [email protected] MIT Center for Collective Intelligence Abstract In this paper we study the differences in historical World View between Western and Eastern cultures, represented through the English, the Chinese, Japanese, and German Wikipedia. In particular, we analyze the historical networks of the World’s leaders since the beginning of written history, comparing them in the different Wikipedias and assessing cultural chauvinism. We also identify the most influential female leaders of all times in the English, German, Spanish, and Portuguese Wikipedia. As an additional lens into the soul of a culture we compare top terms, sentiment, emotionality, and complexity of the English, Portuguese, Spanish, and German Wikinews. 1 Introduction Over the last ten years the Web has become a mirror of the real world (Gloor et al. 2009). More recently, the Web has also begun to influence the real world: Societal events such as the Arab spring and the Chilean student unrest have drawn a large part of their impetus from the Internet and online social networks. In the meantime, Wikipedia has become one of the top ten Web sites1, occasionally beating daily newspapers in the actuality of most recent news. Be it the resignation of German national soccer team captain Philipp Lahm, or the downing of Malaysian Airlines flight 17 in the Ukraine by a guided missile, the corresponding Wikipedia page is updated as soon as the actual event happened (Becker 2012.
    [Show full text]
  • Modeling Popularity and Reliability of Sources in Multilingual Wikipedia
    information Article Modeling Popularity and Reliability of Sources in Multilingual Wikipedia Włodzimierz Lewoniewski * , Krzysztof W˛ecel and Witold Abramowicz Department of Information Systems, Pozna´nUniversity of Economics and Business, 61-875 Pozna´n,Poland; [email protected] (K.W.); [email protected] (W.A.) * Correspondence: [email protected] Received: 31 March 2020; Accepted: 7 May 2020; Published: 13 May 2020 Abstract: One of the most important factors impacting quality of content in Wikipedia is presence of reliable sources. By following references, readers can verify facts or find more details about described topic. A Wikipedia article can be edited independently in any of over 300 languages, even by anonymous users, therefore information about the same topic may be inconsistent. This also applies to use of references in different language versions of a particular article, so the same statement can have different sources. In this paper we analyzed over 40 million articles from the 55 most developed language versions of Wikipedia to extract information about over 200 million references and find the most popular and reliable sources. We presented 10 models for the assessment of the popularity and reliability of the sources based on analysis of meta information about the references in Wikipedia articles, page views and authors of the articles. Using DBpedia and Wikidata we automatically identified the alignment of the sources to a specific domain. Additionally, we analyzed the changes of popularity and reliability in time and identified growth leaders in each of the considered months. The results can be used for quality improvements of the content in different languages versions of Wikipedia.
    [Show full text]
  • International Journal of Computational Linguistics
    International Journal of Computational Linguistics & Chinese Language Processing Aims and Scope International Journal of Computational Linguistics and Chinese Language Processing (IJCLCLP) is an international journal published by the Association for Computational Linguistics and Chinese Language Processing (ACLCLP). This journal was founded in August 1996 and is published four issues per year since 2005. This journal covers all aspects related to computational linguistics and speech/text processing of all natural languages. Possible topics for manuscript submitted to the journal include, but are not limited to: • Computational Linguistics • Natural Language Processing • Machine Translation • Language Generation • Language Learning • Speech Analysis/Synthesis • Speech Recognition/Understanding • Spoken Dialog Systems • Information Retrieval and Extraction • Web Information Extraction/Mining • Corpus Linguistics • Multilingual/Cross-lingual Language Processing Membership & Subscriptions If you are interested in joining ACLCLP, please see appendix for further information. Copyright © The Association for Computational Linguistics and Chinese Language Processing International Journal of Computational Linguistics and Chinese Language Processing is published four issues per volume by the Association for Computational Linguistics and Chinese Language Processing. Responsibility for the contents rests upon the authors and not upon ACLCLP, or its members. Copyright by the Association for Computational Linguistics and Chinese Language Processing. All rights reserved. No part of this journal may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical photocopying, recording or otherwise, without prior permission in writing form from the Editor-in Chief. Cover Calligraphy by Professor Ching-Chun Hsieh, founding president of ACLCLP Text excerpted and compiled from ancient Chinese classics, dating back to 700 B.C.
    [Show full text]
  • China Date: 8 January 2007
    Refugee Review Tribunal AUSTRALIA RRT RESEARCH RESPONSE Research Response Number: CHN31098 Country: China Date: 8 January 2007 Keywords: China – Taiwan Strait – 2006 Military exercises – Typhoons This response was prepared by the Country Research Section of the Refugee Review Tribunal (RRT) after researching publicly accessible information currently available to the RRT within time constraints. This response is not, and does not purport to be, conclusive as to the merit of any particular claim to refugee status or asylum. Questions 1. Is there corroborating information about military manoeuvres and exercises in Pingtan? 2. Is there any information specifically about the military exercise there in July 2006? 3. Is there any information about “Army day” on 1 August 2006? 4. What are the aquatic farming/fishing activities carried out in that area? 5. Has there been pollution following military exercises along the Taiwan Strait? 6. The delegate makes reference to independent information that indicates that from May until August 2006 China particularly the eastern coast was hit by a succession of storms and typhoons. The last one being the hardest to hit China in 50 years. Could I have information about this please? The delegate refers to typhoon Prapiroon. What information is available about that typhoon? 7. The delegate was of the view that military exercises would not be organised in typhoon season, particularly such a bad one. Is there any information to assist? RESPONSE 1. Is there corroborating information about military manoeuvres and exercises in Pingtan? 2. Is there any information specifically about the military exercise there in July 2006? There is a minor naval base in Pingtan and military manoeuvres are regularly held in the Taiwan Strait where Pingtan in located, especially in the June to August period.
    [Show full text]
  • THE CASE of WIKIPEDIA Shane Greenstein Feng Zhu
    NBER WORKING PAPER SERIES COLLECTIVE INTELLIGENCE AND NEUTRAL POINT OF VIEW: THE CASE OF WIKIPEDIA Shane Greenstein Feng Zhu Working Paper 18167 http://www.nber.org/papers/w18167 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge, MA 02138 June 2012 The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. NBER working papers are circulated for discussion and comment purposes. They have not been peer- reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications. © 2012 by Shane Greenstein and Feng Zhu. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. Collective Intelligence and Neutral Point of View: The Case of Wikipedia Shane Greenstein and Feng Zhu NBER Working Paper No. 18167 June 2012 JEL No. L17,L3,L86 ABSTRACT We examine whether collective intelligence helps achieve a neutral point of view using data from a decade of Wikipedia’s articles on US politics. Our null hypothesis builds on Linus’ Law, often expressed as “Given enough eyeballs, all bugs are shallow.” Our findings are consistent with a narrow interpretation of Linus’ Law, namely, a greater number of contributors to an article makes an article more neutral. No evidence supports a broad interpretation of Linus’ Law. Moreover, several empirical facts suggest the law does not shape many articles. The majority of articles receive little attention, and most articles change only mildly from their initial slant.
    [Show full text]
  • A Natural Experiment at Chinese Wikipedia
    American Economic Review 101 (June 2011): 1601–1615 http://www.aeaweb.org/articles.php?doi 10.1257/aer.101.4.1601 = Group Size and Incentives to Contribute: A Natural Experiment at Chinese Wikipedia By Xiaoquan Michael Zhang and Feng Zhu* ( ) Many public goods on the Internet today rely entirely on free user contributions. Popular examples include open source software development communities e.g., ( Linux, Apache , open content production e.g., Wikipedia, OpenCourseWare , and ) ( ) content sharing networks e.g., Flickr, YouTube . Several studies have examined the ( ) incentives that motivate these free contributors e.g., Josh Lerner and Jean Tirole ( 2002; Karim R. Lakhani and Eric von Hippel 2003 . ) In this paper, we examine the causal relationship between group size and incen- tives to contribute in the setting of Chinese Wikipedia, the Chinese language ver- sion of an online encyclopedia that relies entirely on voluntary contributions. The group at Chinese Wikipedia is composed of Chinese-speaking people in mainland China, Taiwan, Hong Kong, Singapore, and other regions in the world, who are aware of Chinese Wikipedia and have access to it. Our identification hinges on the exogenous reduction in group size at Chinese Wikipedia as a result of the block of Chinese Wikipedia in mainland China in October 2005. During the block, mainland Chinese could not use or contribute to Chinese Wikipedia, although contributors outside mainland China could continue to do so. We find that contribution levels of these nonblocked contributors decrease by 42.8 percent on average as a result of the block. We attribute the cause to social effects: contributors receive social benefits from their contributions, and the shrinking group size reduces these social benefits.
    [Show full text]
  • 100125-Wikimedia Feb Board Background Final2 Objectives for This Document
    Wikimedia Foundation Board Meeting Pre-read January 2010 Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Contents • Strategic planning process update • Goal setting • Priority 1: Building the platform • Priority 2: Strengthening the editing community • Priority 3: Accelerating impact through innovation and experimentation • Implications for the Foundation TBG 100125-Wikimedia_Feb Board Background_Final2 Objectives for this document • Provide reminder on design of strategic planning process and current stage of work • Highlight research and analysis that informed the development of priorities for the Wikimedia Foundation • Introduce action items, including priorities for 2010-11 annual plan to be discussed at Board meeting in February TBG 100125-Wikimedia_Feb Board Background_Final3 A reminder: Two interdependent objectives for the planning work • Define a strategic direction that advances the Wikimedia vision over the next five years • Broader reach and participation • Improved quality and scope of content • Defined community roles and partnerships • Develop a business plan to guide Wikimedia Foundation in executing this direction • Organization, capabilities, and governance • Technology strategy and infrastructure • Economics, cost structure, and funding models TBG 100125-Wikimedia_Feb Board Background_Final4 Our frame for developing the strategic direction and Wikimedia Foundation’s business plan • Wikimedia’s vision is “a world in which every single human being can
    [Show full text]
  • Does Wikipedia Matter? the Effect of Wikipedia on Tourist Choices Marit Hinnosaar, Toomas Hinnosaar, Michael Kummer, and Olga Slivko Discus­­ Si­­ On­­ Paper No
    Dis cus si on Paper No. 15-089 Does Wikipedia Matter? The Effect of Wikipedia on Tourist Choices Marit Hinnosaar, Toomas Hinnosaar, Michael Kummer, and Olga Slivko Dis cus si on Paper No. 15-089 Does Wikipedia Matter? The Effect of Wikipedia on Tourist Choices Marit Hinnosaar, Toomas Hinnosaar, Michael Kummer, and Olga Slivko First version: December 2015 This version: September 2017 Download this ZEW Discussion Paper from our ftp server: http://ftp.zew.de/pub/zew-docs/dp/dp15089.pdf Die Dis cus si on Pape rs die nen einer mög lichst schnel len Ver brei tung von neue ren For schungs arbei ten des ZEW. Die Bei trä ge lie gen in allei ni ger Ver ant wor tung der Auto ren und stel len nicht not wen di ger wei se die Mei nung des ZEW dar. Dis cus si on Papers are inten ded to make results of ZEW research prompt ly avai la ble to other eco no mists in order to encou ra ge dis cus si on and sug gesti ons for revi si ons. The aut hors are sole ly respon si ble for the con tents which do not neces sa ri ly repre sent the opi ni on of the ZEW. Does Wikipedia Matter? The Effect of Wikipedia on Tourist Choices ∗ Marit Hinnosaar† Toomas Hinnosaar‡ Michael Kummer§ Olga Slivko¶ First version: December 2015 This version: September 2017 September 25, 2017 Abstract We document a causal influence of online user-generated information on real- world economic outcomes. In particular, we conduct a randomized field experiment to test whether additional information on Wikipedia about cities affects tourists’ choices of overnight visits.
    [Show full text]
  • Word Segmentation for Chinese Wikipedia Using N-Gram Mutual Information
    Word Segmentation for Chinese Wikipedia Using N-Gram Mutual Information Ling-Xiang Tang1, Shlomo Geva1, Yue Xu1 and Andrew Trotman2 1School of Information Technology Faculty of Science and Technology Queensland University of Technology Queensland 4000 Australia {l4.tang, s.geva, yue.xu}@qut.edu.au 2Department of Computer Science Universityof Otago Dunedin 9054 New Zealand [email protected] Abstract In this paper, we propose an unsupervised seg- is called 激光打印机 in mainland China, but 鐳射打印機 in mentation approach, named "n-gram mutual information", or Hongkong, and 雷射印表機 in Taiwan. NGMI, which is used to segment Chinese documents into n- In digital representations of Chinese text different encoding character words or phrases, using language statistics drawn schemes have been adopted to represent the characters. How- from the Chinese Wikipedia corpus. The approach allevi- ever, most encoding schemes are incompatible with each other. ates the tremendous effort that is required in preparing and To avoid the conflict of different encoding standards and to maintaining the manually segmented Chinese text for train- cater for people's linguistic preferences, Unicode is often used ing purposes, and manually maintaining ever expanding lex- in collaborative work, for example in Wikipedia articles. With icons. Previously, mutual information was used to achieve Unicode, Chinese articles can be composed by people from all automated segmentation into 2-character words. The NGMI the above Chinese-speaking areas in a collaborative way with- approach extends the approach to handle longer n-character out encoding difficulties. As a result, these different forms of words. Experiments with heterogeneous documents from the Chinese writings and variants may coexist within same pages.
    [Show full text]
  • The Copycat of Wikipedia in China
    The Copycat of Wikipedia in China Gehao Zhang COPYCATS OFWIKIPEDIA Wikipedia, an online project operated by ordinary people rather than professionals, is often considered a perfect example of human collaboration. Some researchers ap- plaud Wikipedia as one of the few examples of nonmarket peer production in an overwhelmingly corporate ecosystem.1 Some praise it as a kind of democratization of information2 and even call it a type of revolution.3 On the other hand, some researchers worry about the dynamics and consequences of conflicts4 in Wikipedia. However, all of these researchers have only analyzed the original and most well- known Wikipedia—the English Wikipedia. Instead, this chapter will provide a story from another perspective: the copycat of Wikipedia in China. These copycats imitate almost every feature of Wikipedia from the website layout to the core codes. Even their names are the Chinese equivalent of Pedia. With the understanding of Wiki- pedia’s counterpart we can enhance our sociotechnical understanding of Wikipedia from a different angle. Wikipedia is considered an ideal example of digital commons: volunteers gener- ate content in a repository of knowledge. Nevertheless, this mode of knowledge production could also be used as social factory,5, 6 and user-generated content may be taken by commercial companies, like other social media. The digital labor and overture work of the volunteers might be exploited without payment. Subsequently, the covert censorship and surveillance system behind the curtain can also mislead the public’s understanding of the content. In this sense, the Chinese copycats of Wiki- pedia provide an extreme example. This chapter focuses on how Chinese copycats of Wikipedia worked as a so-called social factory to pursue commercial profits from volunteers’ labor.
    [Show full text]
  • Wikipedia Matters∗
    Wikipedia Matters∗ Marit Hinnosaar† Toomas Hinnosaar‡ Michael Kummer§ Olga Slivko¶ September 29, 2017 Abstract We document a causal impact of online user-generated information on real-world economic outcomes. In particular, we conduct a randomized field experiment to test whether additional content on Wikipedia pages about cities affects tourists’ choices of overnight visits. Our treatment of adding information to Wikipedia increases overnight visits by 9% during the tourist season. The impact comes mostly from improving the shorter and incomplete pages on Wikipedia. These findings highlight the value of content in digital public goods for informing individual choices. JEL: C93, H41, L17, L82, L83, L86 Keywords: field experiment, user-generated content, Wikipedia, tourism industry 1 Introduction Asymmetric information can hinder efficient economic activity. In recent decades, the Internet and new media have enabled greater access to information than ever before. However, the digital divide, language barriers, Internet censorship, and technological con- straints still create inequalities in the amount of accessible information. How much does it matter for economic outcomes? In this paper, we analyze the causal impact of online information on real-world eco- nomic outcomes. In particular, we measure the impact of information on one of the primary economic decisions—consumption. As the source of information, we focus on Wikipedia. It is one of the most important online sources of reference. It is the fifth most ∗We are grateful to Irene Bertschek, Avi Goldfarb, Shane Greenstein, Tobias Kretschmer, Thomas Niebel, Marianne Saam, Greg Veramendi, Joel Waldfogel, and Michael Zhang as well as seminar audiences at the Economics of Network Industries conference in Paris, ZEW Conference on the Economics of ICT, and Advances with Field Experiments 2017 Conference at the University of Chicago for valuable comments.
    [Show full text]
  • Visual Narratives and Collective Memory Across Peer-Produced Accounts of Contested Sociopolitical Events
    XX Visual Narratives and Collective Memory across Peer-Produced Accounts of Contested Sociopolitical Events EMILY PORTER, University of Washington, USA P.M. KRAFFT∗, University of Oxford, UK BRIAN KEEGAN, University of Colorado, USA Studying cultural variation in recollections of sociopolitical events is crucial for achieving diverse understand- ings of such events. To date, most studies in this area have focused on analyzing variation in texts describing events. Here we analyze variation in image usage across Wikipedia language editions to understand if like text, visual narratives reflect distinct perspectives in articles about culturally-tethered events. We focus on articles about coup d’états as an example of highly contextual sociopolitical events likely to display such variation. The key challenge to examining variation in images is that there is no existing framework to use as a basis for comparison. To address this challenge, we use an iterative inductive coding process to arrive at a 46-item typology for categorizing the content of images relating to contested sociopolitical events, and a typology of network motifs that characterizes structural patterns of image use. We apply these typologies in a large-scale quantitative analysis that establishes clusters of image themes, two detailed qualitative case studies comparing Wikipedia articles on coup d’états in Soviet Russia and Egypt, and four quantitative analyses clustering image themes by language usage at the article level. These analyses document variation in imagery around particular events and variation in tendencies across cultures. We find substantial cultural variation in both content and network structure. This study presents a novel methodological framework for uncovering culturally divergent perspective of political crises through imagery on Wikipedia.
    [Show full text]