Personalized Story Recommendations

[Myr]

TL;DR - I am looking into options to provide story recommendations and checking interest.

I am exploring options to provide story recommendations.  The best option currently to provide relevant to you recommendations is something called collaborative filtering.  It is basically what Amazon, Netflix, and every other online video streamer and store uses to provide recommendations.  This is an extremely deep technical topic that I'm going to vastly over simplify for this discussion.  Basically, a mathematical equation can be created based on user interactions (like reactions, recommendations and reviews), that can predict, often much better than a coin flip odds, that having looked at one story, you'd be interested in another, based on how other people behaved.

So, for example, I like Stories A, B, and C.  Cia likes stories B, C, and D.  The system would recommend Story D to me and recommend Story A to Cia because we have shared interests in Stories B and C.  Again, this is simplified to just 4 stories between 2 people.

Amazon offers this system as a service.  It has a cost, of course.  And that cost would be ongoing, since it is a service.  But it also offers a stand alone sort of option, where we could build the equation and then do a batch process and create custom emails for all active members, for example.

I'm currently checking levels of interest, so please vote in the poll and comment below.

Oh, on privacy, the data involved with this is all public stuff. Member #, reactions, recommendations, reviews. 

[Mrsgnomie]

I think this sounds good. 

[ReaderPaul]

@Myr, would we be able to opt out if we change our minds later?  However, I do like the idea.

Edited February 27, 2022 by ReaderPaul

[Myr]

5 hours ago, ReaderPaul said:

@Myr, would we be able to opt out if we change our minds later?  However, I do like the idea.

If we end up going this route, I will most likely set it up in such as a way that you must deliberately opt in such as clicking a button for suggested stories or setting in your profile for receiving recommendations.

I'm still in the exploration phase.

[chris191070]

Sounds good 

[Daddydavek]

I voted for the email option.  I'm disappointed that there are so few responses in total. 

[ReaderPaul]

1 hour ago, Daddydavek said:

I voted for the email option.  I'm disappointed that there are so few responses in total. 

I also voted for the email option.  But I could have gone for the "while viewing other stories option."  The email option, it seems to me, would allow me to read the stories later, if I didn't have time to read the suggested stories for a day or three.

[kbois]

It sounds like ot would not only benefit readers, but authors too. Getting their work(s) out to more readers. 

[Myr]

6 hours ago, kbois said:

It sounds like ot would not only benefit readers, but authors too. Getting their work(s) out to more readers. 

Yes, collaborative filtering is used for that as well.  the software can make a reasonable guess based on the past history that would at least display similar stories that people might then engage with and react to.