Citations are important if people are to trust the information, says John Chodacki, director of the Berkeley-based University of California Conservation Center. “This applies to journal articles and also to Wikipedia pages,” he says. In the past, however, analyzing and comparing citation data between scientific papers was only possible with paywall services. “One of the most interesting things is that this information is available.” A total of five articles in the top ten deal with DNA catalogs, including a study describing a method for creating such collections. A 2005 map of nearly 3,000 human protein interactions is also on the list, ranked seventh. (The original Wikimedia post states: “Not surprisingly, Wikipedians love reference books. A separate analysis of the Wikimedia data dump by Ross Mounce, who heads open access programs at London`s Arcadia Fund, shows the ten most cited DOI articles in all language editions of the encyclopedia (see “All language versions of Wikipedia”). Six of the articles are identical, but the first entry is markedly different. The most referenced DOI article is a 2007 article that updates a secular classification of global climate that has 2.8 million citations – but only 169 on the English Wikipedia (the second most cited source of all issues combined has just over 21,000 references). Our results should be viewed with caution given the limitations of model accuracy, the currently limited coverage of articles analyzed by scite and the fact that articles that could not be linked to a DOI in the dataset were excluded6. For a more detailed list of explanations, see this article.
Primary sources cannot (usually) be used for new Wikipedia articles. According to Wikipedia, a primary source is original material close to an event and is often a report written by people directly involved in the subject. A self-published website about you or your brand is not acceptable as a reference. To see how scientific articles referenced in Wikipedia compare to the scientific literature as a whole, we examined the citation distribution of 51,804,643 articles (429,780,086 smart citations in total) from journals indexed on the Web of Science. Of these citations, 18,940,149 (4.41%) reported providing supporting evidence, 2,710,605 (0.63%) reported providing contradictory evidence, and 408,129,332 (94.96%) reported the study citing without indicating that they provided supporting or contradictory evidence. As with Wikipedia, most articles have not been tested by other subsequent citation articles (mentions only citations; 17,441,574 [33.67%]) or have no citations at all (26,396,010 [50.96%]), while 6,038,194 (11.66%) were supported without contradictions, 1,407,829 (2.72%) were challenged with supporting and contradictory evidence, and 521,024 (1.01%) were contradicted without supporting citations. The mean number of citation articles received from these journals was 16.91 (SD = 58.20), the mean number of supporting citations was 0.75 (SD = 2.22), the mean number of conflicting citations was 0.11 (SD = 0.46), and the mean number of citations was 16.06 (SD = 56.71) (Table 1). Again, we looked at 50 random scientific papers indexed on the Web of Science with 0 citations and compared the scite numbers to the dimensions. We found that 26 (52%) had 0 citations in both indexes, 15 (30%) citations in dimensions but not in scite, 1 (2%) were unresolved (due to a defunct DOI or transcription error), and 8 (16%) were indexed in scite but not in dimensions. 4The scite database contains only DOI articles; As a result, other works may have received more citations, but were excluded from this analysis. In most cases, the existence of a Wikipedia page for a publication means that it can be considered trustworthy. Finally, with citations as an indicator of factual actuality, knowing which works support our collective knowledge gives us insight into popular understanding – both how we know what we know and what we know most.
However, this literature does not address an important question: how reliable are the sources cited by Wikipedia articles, especially with regard to scientific topics? To answer this question, we performed an analysis of citations of scientific articles referenced in Wikipedia, using scite`s “Smart Citation” data. Smart quotes provide the context of each quote and a classification that indicates whether it provides supporting or contradictory evidence for the statement cited. Classifications are performed by a deep learning model formed on 43,665 citation statements labeled by experts with accuracy scores of 0.800, 0.8519 and 0.9615 to support, contradict and mention classifications (internal scientific benchmarking data). To date, scite has analyzed more than 16 million full-text scientific articles and extracted more than 500 million citations citing more than 34 million articles. These scientific articles have been obtained in a variety of ways, including retrieval of open access articles, preprints, PubMed Central and through partnerships with various publishers such as Wiley, the British Medical Journal and Rockefeller University Press. The APA recommends linking to a specific archived version of the Wikipedia article so that the reader can be sure that they are accessing the same version. This can be achieved by clicking on the “View History” tab at the top of the article and selecting the latest version: On Wikipedia, the researchers` map serves as an important reference for almost everything from plant and animal species to specific geographical regions and countries. Unlike some of the other most commonly used sources on Wikipedia, such as a fish catalogue and a book on the history of Romania, the map of Australians is almost universal.
Articles that cite it are also published in many languages, which increases the number of references. The resource is also very popular elsewhere on the internet: Lonely Planet, a major travel guide publisher, uses it to provide general weather information for different parts of the world. 2This is based on the number of Wikipedia articles with references to scientific articles included in the dataset used in this article [3], divided by the number of Wikipedia articles existing at the time of writing this article [1]. It takes more than one reference to create a Wikipedia page. Much more. For example, a biography of a living person (we assume you are alive when you read this) requires web references for most, if not all, of the facts presented. Wikipedia editors are very picky about this. If we look at the percentage of total citations added over time, we see that some languages such as Arabic and Spanish are on a steady growth trajectory at the beginning of 2018, while the overall trend (black line) is flattening. As the number of articles in all languages continues to grow, this suggests that the citation rate is slowing down in some languages. Because Wikipedia offers such a high level of visibility, we reviewed 50 random scientific articles with zero citations referenced in Wikipedia to better understand why these articles were not cited or if this was due to scite`s limited coverage.
We compared the number of citations between scite and Dimensions, a freely available traditional citation index. In the subset of Wikipedia, we found that 8 (16%) scientific articles had 0 citations in both indexes, 21 (42%) citations in dimensions but not in scite, 12 (24%) DOI were not resolved, and 9 (18%) were indexed in scite but not in dimensions (Supplementary Table 2). Interestingly, DOIs that were not indexed in Dimensions were almost uniformly descriptions of endangered animals, such as the Sunda pangolin [10]. The discrepancy between the number of citations in scite and the dimensions can be explained by the difference in approach between a traditional citation index and an intelligent citation index, whereas scite requires access to the full text of scientific articles to extract extracts from citations, whereas traditional citation indexes do not.