ON A MISSION TO KEEP STEEMIT CLEAN ||

Source: www.pixabay.com

Repository

https://github.com/steemit/steem

Introduction

If you have been on steemit for at least 6 months, this heading/topic won't be a surprise to you. There is so much garbage going round on steemit and it is almost choking like taking away our oxygen. People spamming everywhere asking for upvotes, automated posting bots posting comments which have value to the post, people copying other people's work without making reference there by claiming it to be their original works and posting of phishing links or websites in order to defraud unsuspecting users of their earnings and reputation.

All this vices above does not speak well of steemit if such acts are perpetrated here and nothing is being done about it, in this analysis i would analyzing about 11 accounts which specialize in making sure steemit is clean and safe to use. They do this by warning via comments or down voting a post or user who is found to breach the rules of engagement on steemit

STEEM CLEANERS

@cheetah, @mack-bot, @steemcleaners, @spaminator, @blacklist-a, @adm, @prowler, @abusereports and @sadkitten

Outline

• Scope
• Results
• Cheetah Analysis
• Steemcleaners
• Other Cleaners
• Conclusion
• Tools and Scripts
• Relevant Links and Resources

Scope

• Date of the analysis: 18th July 2018
• Timeframe of the analysed data: May 2016 to 18th July 2018
• Components of the analysis: Account activities towards cleaning steemit of spam, plagiarism, abuse etc via upvotes, downvotes, blacklist and comments

Results

CHEETAH ANALYSIS

Since joining steemit in July 2016, Cheetah has made more than 500,000 posts and votes (upvotes and downvotes). A cheetah comments or votes does not automatically pronounce you guilty because in cases of plagiarism you might be rightful owner of the material somewhere else on the internet as identity is difficult to verify on steemit.

For more about Cheetah visit https://steemit.com/steemit/@cheetah/faq-about-cheetah

• There were 392,332 cases of plagiarism or finding similar content somewhere else on the internet

• 110,273 warnings about a user on the blacklist who is possibly a plagiarist, spammer or ID thief

Steemit is based on a crypto, so it is not surprising to see that bitcoin is the category with the highest "warnings" or "cautions". A lot of updates go on in that category with automated posting services,many of this reports on crypto prices are copied from other websites and pasted on steemit with no modification or editing.

News on steemit is another area people a lot of copy and paste, that is why it is high up there with bitcoin and cryptocurrency

The reputation of a user on steemit is a pointer to many things one of which is the rules of engagement on steemit. The average reputation of accounts receiving "warnings" from Cheetah for one thing or the other is 29.9

The most common "crime" among users is "finding similar content to theirs". The most reputation with warnings is 25, i am not surprised because with the learning curve of steemit there might be ignorant of the rules or laws. Most come with their knowledge of other social media where such things are not checked e.g Facebook

The monthly voting pattern indicates cheetah has more upvotes than down votes, this is because a user is up voted when similar contents are found and user down voted only when an abuse is established.

The growth with voting correlates with increase in steem price which in turn affects steemit activities as we see rapid increase in July last year right about the price boom of steem and January at its peak rise

The average upvote weight is 4.42% while the average down vote weight is -4.23%.

The voting pattern shows an almost evenly distributed number of votes within the 24hrs, with votes peaking between 6am and 6pm

STEEMCLEANERS

Steemcleaners joined steemit in August 2016 and it comments on your post notifying you of why your post fell under its radar, so far this is some results from its work

• 4,864 comments reporting the use of irrelevant tags. The use of tags is to filter contents so users don't see things they are not interested in. Some users in a bid to make their work have more views, make use of tags that have nothing to do with their contribution and that is abuse

• 78 comments reporting for spamming. when you go around posting your links or words around steemit repeatedly, you are bound to get caught and spamming can be so annoying.

• 18,145 comments reporting for plagiarism. Plagiarism is using a work without making reference to it.

STEEMCLEANERS VOTING

Steemcleaners only downvotes and the weight is in the negative. The average voting weight is -20%

OTHER CLEANERS

CLEANER	DOWNVOTES	AVG VOTING WEIGHT
ABUSEREPORTS	28717	-3.16
ADM	26942	-30.85
BLACKLIST-A	58065	-24.11
MACKBOT	199763	-0.43
PROWLER	44526	-0.96
SADKITTEN	24680	-4.76
SPAMINATOR	41520	-4.72

Conclusion

Steemit must be cleaned so as to keep the credibility of the platform, so far i believe this accounts are doing a great job seeing from the number of down votes and blacklist. People are scared of decentralized blockchain technology because of its anonymity, people have complained of their account being hacked with no trace of the evil doers.

Cleaners can warn users of impending dangers, phishing links and others. Increasing number of post and comments mean the job is becoming hectic in fishing out spammers, abusers and their like.

Tools and Scripts

Microsoft SQL Server Management Studio 17 was used to access the data from STEEMSQL ( a publicly managed database by @arcange )

Microsoft Excel used to plot graphs and Charts.

SCRIPTS

SELECT
author,
category,
parent_author,
body,
CONVERT (int, (SELECT MAX (v) FROM (VALUES(log10(ABS(CONVERT(bigint, reputation)-1)) -9), (0)) T(v)) * SIGN (reputation) * 9 + 25) as Reputation,
comments.created
FROM COMMENTS (NOLOCK)
INNER JOIN ACCOUNTS (NOLOCK) ON [PARENT_AUTHOR] = [NAME]
WHERE

depth > 0
AND
author = 'cheetah'

I ran different queries by different users by changing the author field

The INNER JOIN function is used to get data from two different tables that have something in common

CONVERT (int, (SELECT MAX (v) FROM (VALUES(log10(ABS(CONVERT(bigint, reputation)-1)) -9), (0)) T(v)) * SIGN (reputation) * 9 + 25) as Reputation

Is used to view reputation in human readable form

No time frame was indicated in the query because i wanted to see data from when the account was created up till date.
For July 2018, data is valid till 18th July 2018.

PIVOT FUNCTION and SEARCH function was used in excel from the body table of the result to filter comments by the bot

SELECT *

FROM TxVotes
WHERE voter = 'cheetah'

Was used to get the Votes made by the cleaners, the voter field was changed to different cleaners and the query was run.

Proof of Authorship

https://github.com/jingis07