Conversational Shopping Bot with External Knowledge Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing e-commerce chatbots lack the capability to provide meaningful product recommendations due to restricted data and inability to understand complex user demands, resulting in a fragmented and impersonal user experience.
Innovation Solution
A deep learning method with improved search and dialogue properties that connects via a model to search engines for external knowledge, finds relevant product content, and fuses this knowledge with dialogue history to provide personalized recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If existing e-commerce chatbots use restricted data from previous purchases within a specific store, then the system complexity is low, but the recommendation quality and personalization capability deteriorate
Solution Approach 1:
The patent segments the data source into multiple components: user profile data, product knowledge base, dialogue history, and external knowledge. This segmentation allows the system to access diverse data sources without creating a monolithic complex structure, enabling better recommendations while maintaining manageable system architecture through modular data access layers
Solution Approach 2:
The patent introduces an intermediary retrieval-augmented generation layer that mediates between the language model and multiple data sources. This intermediary component processes and integrates information from user profiles, product databases, and dialogue history, allowing the system to leverage diverse data without directly complicating the core recommendation engine
2Loss of information
If existing chatbots provide basic product listings instead of tailored recommendations, then the information processing requirement is low, but the user experience personalization deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-building user profiles from purchase history and pre-organizing product knowledge bases before the recommendation process. This preliminary structuring of information allows the system to quickly retrieve and personalize recommendations without processing raw data in real-time, enabling personalized experiences while controlling information processing requirements
Solution Approach 2:
The patent implements feedback mechanisms where dialogue history and user responses are continuously incorporated into the user profile and preference model. This feedback loop enables the system to progressively improve personalization accuracy by learning from interactions, transforming basic listings into tailored recommendations through iterative refinement
3Device complexity
If existing chatbots cannot understand complex user demands, then the dialogue processing complexity is low, but the recommendation accuracy deteriorates
Solution Approach 1:
The patent employs dynamic dialogue state tracking that adapts to the complexity of user demands in real-time. The system dynamically adjusts its understanding depth and information retrieval strategies based on the detected complexity of the current query, enabling accurate interpretation of complex demands while managing dialogue processing complexity through adaptive rather than uniformly complex processing
Solution Approach 2:
The patent adds another dimension to dialogue processing by incorporating contextual information from multiple sources: user profile attributes, product knowledge dimensions, and dialogue history context. This multi-dimensional approach enables the system to understand complex user demands by viewing them through multiple contextual lenses simultaneously, improving recommendation accuracy without merely increasing processing complexity in a single dimension
4Device complexity
If existing chatbots provide fragmented and impersonal user experience, then the system integration complexity is low, but the user engagement and satisfaction deteriorate
Solution Approach 1:
The patent merges multiple previously separate systems into an integrated conversational recommendation system: combining user profile management, product search, dialogue processing, and recommendation generation into a unified framework. This merging creates a seamless personal experience by integrating data flows and processing stages that work together cohesively rather than as fragmented separate components
Data Source
AI summary
Provided is a deep learning method with improved search and dialogue properties connecting. The method includes using, via a model, search engines to look for external and factual knowledge across the internet, finding relevant content for any product's aspect, such as price, reviews, and features; and completing the process of fusing the retrieved knowledge with the dialogue history in order to provide the final response. The using, finding, and completing to connect the Alexa socialbot with the Amazon Store, opening novel functions like better recommendations, conversational shopping guidance, automatic seeking of new product types, and personalization.