User Embedding Message Suggestions for Faster Mobile Text Entry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of entering messages, especially on mobile devices with limited input capabilities, is cumbersome and prone to errors, requiring a more efficient and accurate method for message suggestion.
Innovation Solution
A system that suggests messages based on a set of designated messages, tailored to user preferences using user embedding vectors, which are trained from user history and adapted to the context of the conversation, ensuring appropriate and contextually relevant suggestions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual message entry is used, then message accuracy can be maintained, but message entry efficiency deteriorates due to cumbersome input process
Solution Approach 1:
The system pre-processes user history data to generate embedding vectors that capture user communication patterns, preferences, and style. These pre-computed representations enable the message suggestion system to quickly retrieve and rank relevant suggestions without performing complex analysis during the actual message entry process, thereby improving efficiency while maintaining accuracy.
Solution Approach 2:
The patent replaces manual mechanical typing with an automated AI-based suggestion system. The system uses machine learning models to automatically generate and rank message suggestions based on user embeddings and conversation context, substituting the manual input process with an intelligent automation system that learns from user behavior patterns.
2Adaptability or versatility
If generic message suggestions are provided, then system complexity can be reduced, but message relevance to user preferences deteriorates
Solution Approach 1:
The system transforms user history data into embedding vectors that capture essential characteristics of user communication patterns. By changing the representation parameters from raw text data to compressed vector representations, the system achieves personalized message suggestions without proportionally increasing system complexity. The embedding dimension acts as a controllable parameter to balance personalization quality and computational resources.
Solution Approach 2:
User embedding vectors serve as an intermediary between raw user history data and the message suggestion process. These embeddings act as a compact representation that mediates between the complexity of analyzing entire conversation histories and the need for personalized suggestions, enabling the system to adapt to user preferences without directly processing all historical data during suggestion generation.
3Measurement precision
If user embedding vectors are computed from extensive user history, then suggestion accuracy improves, but computational time and resources increase
Solution Approach 1:
The system extracts only the essential features from extensive user history data by computing embedding vectors that capture key patterns in user communication. Instead of processing and storing all historical messages, the system extracts compressed representations that retain the most important information for generating accurate suggestions, thereby reducing computational time while maintaining suggestion accuracy.
Solution Approach 2:
User embedding vectors are pre-computed from user history data before the actual message suggestion process. This preliminary computation allows the system to capture user patterns in advance, so that during real-time message entry, the system can quickly retrieve and apply these pre-analyzed representations without performing time-consuming analysis on the full conversation history.
Data Source
AI summary
A message may be suggested to a user participating in a conversation using one or more neural networks where the suggested message is adapted to the preferences or communication style of the user. The suggested message may be adapted to the user with a user embedding vector that represents the preferences or communication style of the user in a vector space. To suggest a message to the user, a conversation feature vector may be computed by processing the text of the conversation with a neural network. A context score may be computed for one or more designated messages, where the context score is computed by processing the user embedding vector, the conversation feature vector, and a designated message feature vector with a neural network. A designated message may be selected as a suggested message for the user using the context scores. The suggestion may then be presented to the user.