Conversational Recommendation Planning for Fewer User Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing conversational recommendation systems, such as LLM-based chatbots, are inefficient in determining user intent and require multiple queries to provide satisfactory responses, leading to high computing resource consumption due to limited memory and processing capacity.
Innovation Solution
A conversational recommendation system trained using triplets that include user queries, model reasoning plans, and responses, allowing for multi-stage recommendation processes with planning, conversation, and retrieval stages to accurately determine user intent and reduce query requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If LLM-based chatbots are used for conversational recommendations, then natural language processing capability is improved, but computing resource consumption increases
Solution Approach 1:
The patent segments the recommendation task into multiple stages: intent recognition, recommendation generation, and feedback processing. Each stage is handled by specialized components rather than a monolithic LLM, reducing overall computational requirements while maintaining natural language processing capabilities.
Solution Approach 2:
The system performs preliminary actions by pre-processing user queries to extract key features and intents before passing them to the recommendation engine. This preliminary processing reduces the complexity of subsequent LLM operations and minimizes computing resource consumption.
2Power
If LLM-based systems are deployed at scale, then processing capacity is improved, but efficiency deteriorates due to multiple queries required
Solution Approach 1:
The patent implements feedback mechanisms where the system learns from user interactions and refines its understanding of user intent over time. This feedback loop enables the system to accurately determine user intent in fewer queries, improving efficiency while maintaining high processing capacity.
Solution Approach 2:
The system dynamically adjusts parameters such as context window size, model temperature, and processing depth based on query complexity and user behavior patterns. This adaptive parameter adjustment optimizes the balance between processing capacity utilization and query efficiency.
3Loss of information
If multiple queries are required to fulfill user intent, then conversation understanding is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary intent recognition and context extraction from the first query, forming an initial understanding that guides subsequent interactions. This preliminary action reduces the number of queries needed while preserving conversation understanding through structured context management.
Solution Approach 2:
The patent maintains continuous context tracking across conversation turns, preserving useful information from previous queries without requiring users to repeat themselves. This continuity enables accurate intent determination in fewer interactions while preventing information loss.
Data Source
AI summary
Aspects of the disclosed technology include computer-implemented systems and methods for conversational recommendation systems, such as conversational chatbots that are configured to process user queries and generate responses. A recommendation system includes a conversational user interface configured to receive a user query and provide a recommendation response and a machine-learned sequence processing model that has been trained on training data including a plurality of triplets. Each triplet includes an example query, an example model reasoning plan associated with the example query, and an example response associated with the example query and the example model reasoning plan. The sequence processing model can be trained to provide conversational-based recommendations using a multi-stage recommendation process that includes a planning stage, a conversation stage, and a retrieval stage.