11/9/2022 0 Comments Wechat official account proposal![]() ![]() Chinese-run academic tutoring and career consultancy services also partner with CSSAs to sponsor advertisements disguised as career guides. For example, the logos of major Canadian banking and telecommunication companies were showcased in a University of McMaster CSSA article reporting a Uyghur rights talk to the Chinese embassy, underneath a so-called “nine-dashed line” map presenting disputed islands in the South China sea as Chinese territory. This created potentially reputation-damaging juxtapositions. Local Chinese businesses and even established Western companies paid for advertising on CSSA WeChat pages. Sponsors were vital to each year’s grand production of the Spring Gala, and CSSAs actively courted partnerships. Despite the pandemic lockdowns, the CSSA had high expenditures of roughly $40,000 for the 2019–2020 school year, even though the CSSA claimed it was much less than that of the previous school year.īridging the funding gap is the role of advertisement-related posts - the second-largest category, representing 26 percent of the total. For reference, UC Berkeley CSSA publicly released portions of its financials after accusations of embezzlement by member students in 2021. ![]() These cultural activities are cash drains that require substantial outside financial support. ![]() Events include everything from singing contests to cooking competitions, with the biggest event being the Spring Gala for Chinese New Year. (Source: DFRLab/WeChat)Īrticles in the “Culture” category made up over 40 percent of the total, suggesting CSSA members spend the most time planning and participating in cultural activities. Table 1: total articles published by each country’s CSSAs per article category. The labeled posts then became a starting point for further research. The classifier had an overall accuracy of 70 percent when measured against a testing set articles in the national interest, network, and patriotism categories were reviewed to remove as many false positives as possible. Instead, DFRLab manually labeled 1,700 articles and used their titles to train a logistic regression classifier that categorized the remaining articles into the ten separate categories. It was impractical to individually filter through each of the 44,546 articles to tag those of interest.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |