Recommendation Systems

From MIT Technology Roadmapping
Revision as of 03:55, 30 October 2020 by 10.244.0.1 (talk)
Jump to navigation Jump to search

Roadmap Overview

Recommendation systems have become a critical engine of the modern digital economy, allowing businesses to exploit user behaviors and similarities to develop specific notions of preference and relevance for their customers. Today, recommendation systems can be found “in the wild” in many different services ubiquitous to daily digital life, filtering the content we see (eg Spotify, Tik Tok, Netflix), products we are advertised (eg Instagram, Amazon), and humans we connect to (eg Tinder, LinkedIn). Against the backdrop of Level-1 commercial user understanding or personalization, we develop a Level-2 roadmap for Recommendation Systems (2RS) below.

OPM Model

OPD of a Recommendation System

DSM Allocation

Recsys tech hierarchy.png
Recsys dsm.png

Figures of Merit

FOM Definition Units Trends (dFOM/dt)
User Base Size Number of users and datapoints used to generate a recommendation [users] The number of users and datapoints has increased over time as databases of users and their preferences has grown
Accuracy Degree to which the recommendation is relevant to the user. This depends on the implementation of the recommender system, but one example could be match percentage (Netflix, OkCupid) [%, relevant content recommended / total content recommended] It is not clear that the quality of recommendations has increased over time.
Quantity Number of recommendations generated by the system for each user [recommendations] The number of recommendations systems can generate has increased as the amount of content in the system has increased
Engagement Number of users who display interest in the recommendation provided. This depends on the implementation of the recommender system. An example of this could be clicks or views. [user clicks] Unknown if this has changed over time


Technical Model

https://github.mit.edu/ebatman/Recommender-System