algorithm for determining similar topics based on votes

[ddfridley]

Each user will see 2 topics that they have created, plus 2 topics from each of 9 other people. (20 total)
The user is asked to group the topics that are similar, and pick the one that is most representative in each group.
There will be 100's of users, and each user will be shown items at random, but each user's input will be seen an equal number of times.

Develop an algorithm for figuring out most representative groupings, based on the user input of what they group and what they select as most representative.

There will be a round based system, where if there are 1000 users, each user will review a group of 20, from 10 different users. Then, the most representative top 200 groups (10%) will be distributed at random in groups of 20 to users again, and they will be asked to group them. If there are 10000 people, another round will be added.

The number of topics a user creates, and the number of other people's input that will be shown to a users should be variables that we can tweak based on users experience. (Basically - is 20 items to many and users burn out?)

[ddfridley]

Derek will investigate and try to draw out a process, and investigate using in javascripts, or python.

[ddfridley]

David and Derek met on 3/30 and generated some example data to use for grouping. But more is needed. Goal to finish process workflow for next week. Hope to have basic javascript for processing initial data.

[ddfridley]

Derek wrote a test of grouping responses, and group by count.
A challenge is how do we keep track how often topics are shown.
Derek and David are working on a google form in order to get initial data to seed the algorithm.

[ddfridley]

@dplem working on algorithm to determine how many times a topic has been seen, and then figure out how often a topic has been voted up.
Here is some data found to help with generating test data.
https://colab.research.google.com/drive/1EJol2OxLwWbgdn4_e98pZeFv4ZQqkOKr?usp=sharing

On friday we will talk about how to read in a csv file.

[ddfridley]

This process workflow was done by @dplem https://docs.google.com/presentation/d/1WI7p2pf2HKeRnRCXMMKaz_svSq7PeHU5SFz_3mk2O88/edit?usp=sharing

[ddfridley]

Ignore the text to the left-this is a whiteboard drawing of an algorithm for figuring it out from the first round.