More personal Aigents analytics on Steemit for @dan, @dantheman and @son-of-satire

last month
50 in steemit

Here we continue to study - how Aigents ( @aigents ) technology powered by artificial intelligence, available at https://aigents.com/ and presented by @akolonin can be used to make sense of people around you in social media.

We just have used Aigents personal social reporting engine as described in video at


(slides can be obtained at http://aigents.com/papers/2017/Aigents-News-Analytics-2017-en.pdf) and created reports for three accounts well known at Steemit: @dan, @dantheman and @son-of-satire.

Here is what we have learned so far.

https://aigents.com/reports/report_steemit_dan.html
Posts of @dan can be split in four groups - involving

  • bitcoin/blockchain/dao markets and communities;
  • smart contracts;
  • aspects of accounting and values;
  • aspects of reality.

People most similar to @dan on Steemit are @sirwinchester, @williambanks, @sabbybibi, @eeks, @wingz and @innuendo - mostly sharing concerns regarding further fate of the Steem and fate of people connected to it.
Best friend of @dan on Steemit is @stellabelle, with next really good friends @tuck-fheman, @krnel and @dana-edwards.
The top fan of @dan is @murh, with @magix, @teamsteem, @robrigo, @boombastic and @smooth following next.
Top authority for @dan is @ozchartart, he also pays substantial attention to @krnel, @mage00000, @dollarvigilante and @tftproject.
Best karma was earned by @dan in July 2016, in discussion regarding Steem community aspects connected to community of people and constitution issues.
Most popular terms used by @dan are "steem", "people", "dao", "account", "community", "blockchain" and "consitution".
Most impression to audience was cause by post "Steemit Releases Groundbreaking Account Recovery Solution" last year, and the best words attracting people's attention were "way", "people" and "use" (no comments are needed for that).
After all, it is really sad seeing @dan resigned however we believe that whatever is happening is happening to the best :-)

https://aigents.com/reports/report_steemit_dantheman.html
Posts of @dantheman can be slit in two groups:

  • about blockchain, market, money and power;
  • regarding voting and content.

People most similar to @dantheman on Steemit are @donkeypong, @son-of-satire, @dragosroua, @xeroc, @anotherjoe and @lukestokes - sharing concerns about "Steem" and "people".
Best three friends of @dantheman are @tuck-fheman, @stellabelle and @krnel (this makes it surprizingly similar to @dan).
There is great fan of @dantheman called @johnathanhenry with may other following far away next.
The top authority for @dantheman is @steemitblog (no great surprise, huh) and then @larkenrose the next authority.
The most karma by @dantheman has been earned in August 2016, while addressing issues people using Steem for voting.
The most popular terms used by @dantheman are "steem", "people", "value" and "consensus".
The most liked posts written by @dantheman are "Why I Flag @ozchartart" this year and "Why do we fight to change the world?" and "Curation Rewards and Voting Incentive" last year and the words attracting audience of @dantheman to greatest extent are "people" and "means".

https://aigents.com/reports/report_steemit_son-of-satire.html
Posts of @son-of-satire, compared to previous accounts seem having more existential nature, less focused on one particular topic, so they can be split in four categories with the following (possibly speculative) interpretations:

  • addressing questions "why?" regarding the "way" of doing things;
  • regarding how to "give" and "keep";
  • concerning what is "wanted";
  • focused on matters connected to "steem".

Accounts most similar to @son-of-satire are @lifeworship, @dreemit, @trevor.george, @stellabelle, @j3dy, @riskdebonair and @the-ego-is-you, sharing interests regarding the "steemit", "people" and "time".
The ultimately best friend of @son-of-satire on Steemit is @thecryptofiend with @rigaronib, @dwinblood and @dreemit following at distance.
He has three consistent fans called @trans-juanmi, @selwi, @juliosalas and @jeff-kubitz.
Ultimate authority for @son-of-satire seem to be @rubens9119, there are also @curie, @wowpeach and @steemsports but relationships seem to be more of mutual nature.
Most karma by @son-of-satire has been earned in February 2017, when he addressed Earth-scale concern of the current situation.
Obviously, the latter is connected with the most attracting post "The Truth About Power -- Seeing Past The Illusion" in February 2017. The next attracting post is about "ow Social Media's Predictive Algorithms Are Perpetuating Division Among The People" (and we like it post so much that we put link to it right here: https://steemit.com/life/@son-of-satire/how-social-media-s-predictive-algorithms-are-perpetuating-division-of-the-people).
Finally, the best words liked by fans of @son-of-satire are "now", "post", "like", "know", "people" - which leaves not doubts regarding passion and confidence of the author.

As a summary, @dan and @dantheman share a lot of properties and connections and are concerned more about financial aspects of Steem crypto-currency while @son-of-satire and his environment are rather more about social aspects of Steemit as a social network.

It is well understood that current Aigents analytics is not perfect. We are working to add more features to it and make it more precise (thanks to @akolonin, @dark.sun and @lehard). We are also looking for more partners on Steemit side and outside to provide more values to end users - people around the Earth (quite in sync with passion of @son-of-satire).

More details about how to use Aigents can be found at @akolonin or https://aigents.com/. Complete level of details could be available in terms of specific partnership.

@akolonin
Skype:akolonin

Authors get paid when people like you upvote their post.
Join our amazing community to comment and reward others.
Sort Order:  trending
69
  ·  last month

You may wish to indicate that this post was entirely machine generated. Thanks for the mention though!

·
50
  ·  last month

Huh, I wish the post could be generated entirely automatically by Aigents robot. Not, yet. The reports available via links are generated, indeed. But the interpretation was created by @akolonin. Thank you for idea, though ;-)

60
  ·  last month

Top authority for @dan is @ozchartart

I see. Dan left because ozch told him to.

·
69
  ·  last month

Agreed, I think authority is not the word they are looking for.

·
·
50
  ·  last month

Well, the term "authority" may be not perfect. The "Best friend" is evaluated as power of symmetric interactions between accounts. The "Fan" means person actively votes and comments the account owner not being voted and commented in return - asymmetrically. The "Authority" is the asymmetric opposition to "Fan". We will be publishing the all math under the hood in couple months in the paper. For the time being, there is a presentation describing the idea of the social environment assessment engine:

·
·
·
69
  ·  last month

Yes I understand, but words have meanings and that one has connotations that are sort of the opposite of the interactions we all watched them have. I understand what you mean, but that particular choice of words is causing an adverse impact on the credibility of the study. Simple solution find a better word to describe the thing measured. :)

·
·
·
·
50
  ·  last month

Ok, have talked to another English speaker and agreed on either "Authority (opinion leader)" or "Authority (thought leader)" - hope it will be better although a bit longer. Maybe just replace with "Opinion leader" or "Thought leader" entirely... Thank you!

·
·
·
·
·
69
  ·  last month

Neither is really correct.

"Thought leader" implies someone you rely on to do heavy thinking freeing you of the burden, whereas "Opinion Leader" can be thought of as someone who's advice you regularly take into consideration.

You can just drop the word Authority from there, it's vague and confuses what is being measured here.

Best term would be "Listens to..." or more accurately...
"Responds to",

Which is what you've actually measured here, "Who has Dan most frequently mentioned by name?"

Dan almost never mentions anyone by name, especially in the title of a posting. Your method added extreme weight to this exchange...
https://steemit.com/steemit/@ozchartart/my-response-to-dantheman-s-downvoting-campaign-against-me
https://steemit.com/steemit/@dantheman/why-i-flag-ozchartart

Most likely you are counting @mentions like citations.
Notice how many times he @mentions @ozchartart in that posting?
What about changing the algorithm slightly to only count each @mention in a single posting once?

I am curious to know what happens if you eliminate outliers like that and rerun your report.

Strange you didn't pick up @thedollarvigilante in the "Thought Leader" category, since @dantheman seems to have the utmost respect for that guy.

By the way I'm curious to know the algorithm that placed me on a social similarity index with @dantheman so closely. I'm flattered, but so much of our opinion on things are in direct, polar opposition.

I do agree we are concerned with a lot of the same things. However I wonder if it's providing too much weight to the fact that several of my postings are direct responses to things @dantheman has posted about, vs topics of my own choosing which are not "this is my response to dan", types of subjects.

One last tip... Make sure to apply a bit more weight to tags contained in the body of a message, during the context parsing phase. Then ditch the tags consideration you're applying for the top 50 or so tags, found in the tags section of the post (not the body) and apply a lot more weight on the unique tags for subject ranking.

What you're not considering here is that the top 3 tags attached to any post are generally something off the trending list that day and the subject may only be cursorily related to it's tags.

Keep up the great work BTW! You're fulfilling a prediction I made back in August that steemit would make an ideal training ground for ground breaking concepts in AI.