Project to develop a user-centric analytics platform for the stackexchange community.
Currently developing a collaborative-filter recommendation engine to better match users (stackexchange members) with items (questions), so that users are presented k-top-questions that are current and which will likely be of interest to to that specific user. Initial prototypes will be based on static data that is publically available in a stackexchange data dump (published March 2016), though plans are to extend this via real-time calls through stackexchange APIs.
Recommendation Engine V1 (proof of concept)
- site: datascience.stackexchange.com
- goal: Item-based collaborative filter
- matrix: m users x n questions
- entities: unary answer response (may treat NA as 0 for binary)
- ui: console and file i/o
Recommendation Engine V2
- site: datascience.stackexchange.com
- goal: Item-based collaborative filter enhanced with item & user tag content
- matrix: m users x n questions
- entities: unary answer response (may treat NA as 0 for binary)
- ui: rudimentary shiny