How toxins are tanks compared to other left groups?

Links table

Abstract and 1 introduction

2 background and relevant work

3 data

3.1 Determine Tankie Subreddits

3.2 Determination of sub -ideology and 3.3 groups yet

4 User base analysis and 4.1 Create a graph and community detection

4.2 Community growth

4.3 The user’s deportation over time

5 content analysis and 5.1 What are the tanks talking about?

5.2 Who talks about tanks?

5.3 Disturbance analysis

5.4 toxic analysis

5.5 field analysis

5.6 LEMMYGRAD analysis

6 discussion and conclusion and 6.1 restrictions

6.2 Antiquities, future work and references

Data

B name is entities

C Disable Analysis

Domain analysis d

5.4 toxic analysis

In this section, we use API models to compare online behavior for cabinets and other left societies.

API perspective. API perspective [92] It is widely used [9, 12, 26] Tool to measure toxicity. Although it has restrictions, for example, there are issues about prejudice and performance questions when facing unaccounted conversation patterns, it provides a proper comparison scale of online communities. The application programming interface provides six production models: 1) toxicity, 2) highly toxic, 3) IDENTITY_ATTACK, 5) threats, and 6) profanity (see see [91] For full details on models). We consider a 0.8 threshold, which was defined as “high” for dozens of sharp toxicity by Hoseini et al. [57]. For a basis line for comparisons, we try 0.5 % of Reddit publications during the data group schedule, which represents more than 36 meters.

results. Figure 5 shows the functions of the cumulative distribution (CDFS) for each model within the left block. Our analysis reveals that the tanks tend to get higher degrees than other Farleft communities (except R/Allthelect) for all perspective API models. In addition, all left -left societies have API degrees a higher perspective than the basic Reddit sample.

Specifically, the tanks have the highest percentage of both grades ≥ 0.5 and high degrees (i.e. grades ≥ 0.8) for identity and threat, and the second highest percentage of the remaining forms. Table 5 shows that the tanks have nearly twice the number of high degrees of the average of other left societies.

We confirm that the score distributions of each model varies greatly between tanks and other left societies using a dual-sample KS test (𝑝 <0.01 for everyone after modifying the multiple test using the Benjamini-Hochberg method). These results indicate that the tanks tend to create leaflets with higher levels of toxicity, insult and profanity compared to other left societies, with the exception of R/alltheleft. Tanks also tend to create leaflets with more identity attacks and threats than all other left societies.

Next, we study the entities called Tankies who have high perspective API degrees by removing any specific entities that show less than 100 times. In Table 6, we offer the 10 best entities called the rank in part of the posts that mention the entity and highly record across all perspective models. For all models except identity, the most common entities mentioned mainly are associated with the United States (for example, amerikkkans, Yankee, Charlotte, Qanon),

Public figures and politicians (for example, John Oliver, Elon Musk, Kyle Retnhaus, Anthony Blinkin, Erdogan, Alex Jones, Joe Rogan), or countries/nationalities (for example, Chechnya, Brazil, Iraqis). For the IDENTITY_ATTACK model, the most frequently mentioned entities are religious or ethnic groups. The tanks seem to target Muslims and Jews in the first place, with 41.80 % and 34.03 % of the publications that mention these groups that have high identity degrees; The highest percentage of the high identity degrees of Muslims and Jews from any far left society that we analyze. Moreover, we note that tanks attack the identities of Asians, Arabs, Hindus, Mexicans, Africans, and eggs in more than 20 % posts mentioning these identities.

Fast food. Our analysis shows that the tanks have the highest percentage of high degrees of identity and threats between remote left societies, and they have the second highest percentage of high levels of toxicity, intense-toxic, profanity, insulting, behind R/alltheleft. Although our findings indicate that American policy is not the main focus of tanks, it still expresses strong opinions on events related to the United States, conspiracy, politicians and public figures. Finally, we note that tanks target Muslims and Jews often.