Sticker Recommendation via ML Relevance Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of messaging applications face challenges in identifying appropriate stickers when responding to messages, due to the vast number of stickers available and the lack of standardized character coding systems.
Innovation Solution
A machine learning model is trained to generate relevance scores for stickers based on attributes and characteristics of the received message, sender, and recipient, allowing for the recommendation of relevant stickers in a reply interface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a messaging application provides a vast number of stickers for user expression, then the expressiveness and engagement of messages are improved, but it becomes difficult for users to identify appropriate stickers when generating replies
Solution Approach 1:
The system automatically generates sticker recommendations by analyzing the received message content, sender characteristics, and recipient profile without requiring manual browsing. The machine learning model self-services the sticker selection process by predicting relevant stickers and presenting them to the user in a ranked list, eliminating the need for users to manually search through vast sticker libraries.
Solution Approach 2:
The patent replaces the mechanical manual browsing and selection process with an automated machine learning-based recommendation system. Instead of relying on users to manually navigate and search through stickers, the system uses NLP and ML models to automatically analyze message context and generate personalized sticker recommendations, substituting human cognitive effort with computational intelligence.
2Adaptability or versatility
If stickers are customized and personalized beyond standardized emoji, then user expression flexibility is improved, but the lack of standardized character coding makes it harder to systematically organize and retrieve stickers
Solution Approach 1:
The system transforms the sticker selection problem from a categorical organization challenge to a contextual relevance problem. Instead of organizing stickers by fixed categories or codes, the machine learning model dynamically evaluates multiple parameters including message content semantics, sender-receiver relationships, conversation context, and user preferences to generate relevance scores. This parameter-based approach handles customized stickers without requiring predefined organizational structures.
Solution Approach 2:
The patent introduces a machine learning recommendation system as an intermediary between the vast sticker library and the user. This intermediary analyzes message context and automatically filters and ranks stickers based on predicted relevance, bridging the gap between unorganized customized stickers and user needs. The intermediary translates complex sticker attributes into simplified relevance scores that guide user selection.
3Measurement precision
If a machine learning model is trained on historical user behavior data, then recommendation accuracy is improved over time, but the system requires ongoing data processing and model updates
Solution Approach 1:
The system implements feedback loops where user interactions with recommended stickers (selections, skips, modifications) are captured and used to retrain and refine the machine learning model. This continuous feedback mechanism improves recommendation accuracy over time by learning from actual user behavior patterns. The model is periodically updated with new training data derived from accumulated user interactions, creating a self-improving recommendation system.
Data Source
AI summary
Described herein is a technique for processing a received media content item (e.g., a message), received at a messaging application of a first end-user of a messaging service, to generate a selection of some predetermined number of recommended stickers. The recommended stickers are then presented in a user interface to the first end-user, allowing the first end-user to select a sticker for use in replying to the received media content item. To generate the selection of recommended stickers, in response to receiving the media content item, the messaging application processes the media content item to identify specific attributes and characteristics (e.g., text included with the message, stickers used with the message, and other contextual metadata). The identified attributes and characteristics of the received message are then processed by a scoring model to identify the predetermined number of stickers for presenting in the reply interface as recommended reply stickers.