(Speculative) Data Intensive Dating and Friendship App

1 votes

Create a thorough, sophisticated personal profile based upon data imports and individual user input to connect with potential dating partners and friends. Could include data from Meyers-Briggs personality tests, movies/TV/content streaming, music streaming, sleep time/patterns/duration, activity levels/fitness level/types of activity, work hours/types of work, book interests, location and travel data, education level/areas of interest, fashion interests/consumer purchases, volunteer activity, and other material the user could input/adjust manually. This info could then be processed into a synthetic profile, or shared via encryption, such that profiles can match against each other without anybody but the individual user actually knowing specific data. Profile matches could get an overall percentage match + overlapping areas of interest matches, say a T.V. show, movie, or band each person likes, which would serve as a conversation starter. Example of Valto's microcosm social network idea

Under consideration Suggested by: Matt Eva (14 Feb, '21) • Upvoted: 14 Feb, '21 • Comments: 4

Comments: 4

14 Feb, '21
Matt Eva
Uncertain about the feasibility of this kind of app/how matching percentages would work. Would either rely upon direct sharing of profile with another user to compare data and get a literal percent match result based upon overlapping areas - which would also require a flexible definition of "match" parameters - or rely upon the ability to create synthetic profiles of user data that become more sophisticated over time to generate better matches. Markus pointed me toward the federated learning model in regard to this latter idea.

The format - in my mind - would function similar to tinder, and could go in two ways. Either a person would automatically share their data profile with each new profile that appeared, meaning the percent match and areas of interest would appear alongside the immediate viewing of each new profile, or data profiles would be shared upon the instance of a mutual match - each person "swipes right" (or is it left?).
14 Feb, '21
Matt Eva
Ideally, the percent match and overlapping areas of interest would appear immediately upon viewing a profile, which would encourage people to make "affirmative" decisions based more upon overlapping qualities/areas of interest rather than visual appeal (which is how Tinder primarily functions), but this could present a greater data security/privacy risk, as profiles would be automatically shared with each user. Getting matches after both people have "swiped right" would have another layer of data sharing consent, meaning that people would have explicitly agreed to sharing their synthetic or encrypted profile with a specific user. In either case, once the match has been generated via comparison, the synthetic or encrypted data profile should be automatically deleted from wherever the comparison was made, which would likely be an individual's local environment, as that's where apps in our ecosystem will run.
14 Feb, '21
Matt Eva
Developing this kind of app seems like it would present a significant challenge/could require a lot of resources, but if it functioned I could see it making a huge wave in the dating app market, as it could provide the most accurate, private, and secure dating and friend-making app around.
14 Feb, '21
Matt Eva
Alternatively, the app could simply give each individual user a recommendation on potential partners based upon their own personal data, which would be a kind of match-making app that points individuals toward potential ideal partners/friends rather than connecting directly to them via a social network. That would be much simpler and less complicated. User's could still come up with some kind of abstract "score" or representation of their data, and get recommendations to connect with other individual's with a certain "score" or representation. User's could then share that score with prospective partners or friends to assess compatibility, which wouldn't require any kind of actual data sharing.