Synthetic Feedback Generation for ML Recommendation Cold Start
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
New digital items on online recommendation platforms often lack user feedback, leading to the 'cold start' problem, where they are not recommended to users, thereby compromising user experience and the ability of item providers to reach their target audience.
Innovation Solution
A method and system that generate synthetic user feedback for new digital items by training a scoring machine-learning model to mimic the feedback distribution of existing items, allowing these new items to be considered and recommended by the platform's machine-learning algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the platform uses a machine-learning algorithm trained on user feedback to generate recommendations, then recommendation accuracy is improved, but new digital items without feedback are excluded from recommendations
Solution Approach 1:
The system performs preliminary action by generating synthetic user feedback for new digital items before they accumulate actual user feedback. This allows the machine-learning algorithm to immediately incorporate new items into recommendations based on synthesized feedback patterns derived from existing item-feedback relationships, resolving the cold start problem while maintaining recommendation accuracy
Solution Approach 2:
The system introduces an intermediary component that synthesizes fictional user feedback as a bridge between new digital items and the machine-learning recommendation algorithm. This intermediary generates plausible feedback patterns based on existing feedback distributions, enabling new items to be included in recommendations without compromising the algorithm's accuracy on established items
2Reliability
If the platform excludes new items without feedback from recommendations, then recommendation quality is maintained, but item provider visibility and user discovery are reduced
Solution Approach 1:
The system creates copies of realistic user feedback patterns and applies them to new digital items. By copying the structure and distribution characteristics of actual user feedback from similar items, the system enables new items to gain visibility in recommendations while maintaining the quality and reliability of the recommendation system through authentic feedback-like data
3Device complexity
If the platform introduces new items without synthetic feedback, then system complexity is reduced, but user experience and item provider satisfaction deteriorate
Solution Approach 1:
The system implements self-service by automatically generating synthetic feedback for new digital items without requiring manual intervention. The machine-learning model autonomously creates plausible feedback patterns based on existing data, eliminating the need for complex manual curation processes while improving user experience and item discovery
Data Source
AI summary
A method and a server for generating a training set of data for training a Machine-Learning Algorithm (MLA) to generate digital item recommendations for users of an online recommendation platform are provided. The method comprises generating synthetic user feedback for new digital items, devoid of any actual user feedback. The generating comprises: (i) during a first stage, training, by the server, a scoring machine-learning model to determine a predicted indication of the user feedback for a given digital item; and (ii) during a second stage: acquiring an indication of a given new digital item of a respective new item provider; applying the scoring machine-learning model to the given new digital item to determine a respective indication of the synthetic user feedback therefor; and generating, based on the new digital items with respective indications of the synthetic user feedback assigned thereto, the training set of data for training the MLA.