[versus feature] map opinion properties of versus threads by ratings of users and display results as net-line-graph

[infinite-dao]

Summary, Idea

I became an idea for the future to implement in the versus feature. Because forcing consensus can lead to have groups manipulating public opinion we need to ensure to say multiple opinions, different perspectives etc., but these perspectives can also be ratet by (rather more or less objective) properties to see the intensions or characteristics of a versus debate, to visualize somehow the ongoing debate not primarily by emotions but by properties of valid good journalism characteristics (sure there are some guidance rules out there) that can indicate the intent of a versus. These properties have to be developed yet if there is no opinion-ontology out there yet.

Details

So one can

• define opinion properties, communication properties that can visualise the sincerity, truthfulness, honesty, aiming at independence and communication culture of that versus thread, to visualise where the thread is anchored in the “opinion space, opinion spectre”, e.g.
  • candid (* to *****)
  • aimed to the truth (* to *****)
  • substantiation (* to *****)
  • well researched (* to *****)
  • communication culture (* to *****)
  • etc.
• let people rate these properties of opinion and communication
• display the result in a net like display

The net display could look like a spider map, in the middle the actual versus or so:

Steps to start with

Properties

• Ontology of http://gsi.dit.upm.es/ontologies/onyx/ is focused on emotions but may give a hunch to develop reasonable, rather objective opinion properties.
• look into the “investigative handbook of journalism” at the rules section (I don’t know if there is such a thing ;-) — to conceptualize the properties

Technical examples

See also the plot of coocgraph of occurrences of a word

• the svg graph can provide something to start with, e.g. “debate” http://corpora.uni-leipzig.de/de/res?word=debate&corpusId=eng-za_web_2014
• http://wortschatz.uni-leipzig.de/de/documentation/faq#corporaportal_coocgraph

cheers, Andreas

[infinite-dao]

some subjects of properties to conceptualise (just came to mind):

• truth (German: Wahrhaftigkeit)
  • aimed to the truth (German: wahrheitsgetreu)
• perspective view
  • broaden vs. narrow intend/mindedness
• culture of communication
  • the arguing, agitation vs. argumentation, sympathetic
• validity
  • substantiation, valid, justified
  • well researched vs. mere surmised
• the heat of the debate itself (see example image above: red (hot), orange, yellow, light green, blue (cool)
  • one could measure that by the speed of inputs sent to the versus (average of 10 sent inputs recalculated gives the heat measure or so)
• … (perhaps further properties)

To implement it,

• I would aim to quality, to something I would call “opinion-quality-o-meter“, so to say, not map emotions, arguments, the content or opinons itself, you know, but the quality of the debate
• it is imperative to ensure the freedom of opinion ant not frame the opinion space but measure quality parameters
• not too much properties 5 max. or so

[realB12]

Maybe we could start with "just" one attribute. Let's say "truthiness" (the quality of seeming or being felt to be true, even if not necessarily true). Recently there have been big debates, books haben been written and public discussions in the "Fake News" context. Just THIS alone, demonstrates the dimensions of such an undertaking and how complex it is to find common ground and clarity that could be managed by simple attributes. However, there is some progress in this area and other projects may profit heavily when we would come up with a working solution prototype. On the other hand, addressing more than one attribute I personally consider too much of scope that most probably will leed to endless discussions and never ending confusion. I would prefer to focus on ONE thing, find a solution and add other "social aspects" later.

[realB12]

Along my current work in the AI driven BotTech field I have found that practial, narrow, real live user activity focused approaches and designs might bring us farther and might widen our solution portfolio faster than building from abstract theories.

[realB12]

Another viewpoint: Are we really sure, that people want their input to be categorized and rated by others or algorithms? Do we have some evidence that this feature is urgently needed? Or is it just nice to have? How and who is putting this feature in an overall context?

[ghost]

This is a general discussion on the feature, if and when we need it will be discussed in a developers meeting at some later point. For now it's not urgent.