Chatbot System Training with Semantic Fact Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Chatbot systems generate unnatural and bland answers, leading to user dissatisfaction due to their inability to consider relevant facts in response to user requests, which affects the user experience and overall satisfaction with online services.
Innovation Solution
A chatbot system is configured with a semantic similarity machine-learning model to generate vector embeddings of user requests and a generative model that incorporates relevant facts to produce more natural and context-aware answers, trained using a method that involves acquiring dialogue data, identifying relevant facts, and optimizing the difference between machine-generated and human responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a chatbot system generates answers based on standard retrieval and generation processes, then the system can provide relevant information, but the answers appear unnatural and bland
Solution Approach 1:
The system performs preliminary retrieval of relevant passages before generating the answer. These passages are then used to enhance the generative model's output, allowing the chatbot to incorporate factual context and nuance from retrieved information while maintaining natural conversation flow
Solution Approach 2:
The patent introduces an intermediary mechanism that combines retrieval-based information with generative modeling. The generative model acts as a mediator that processes both the user query and retrieved passages, synthesizing them into natural-sounding responses that reflect human conversation patterns
2Reliability
If the chatbot system uses simple retrieval and generation, then the system complexity is low, but the answers lack context-awareness and relevance
Solution Approach 1:
The patent merges two previously separate approaches: retrieval-based information extraction and generative language modeling. By combining these systems, the chatbot achieves both the factual accuracy of retrieval methods and the naturalness of generative models, while improving context-awareness through integrated processing
Solution Approach 2:
The unified model serves multiple functions simultaneously: it retrieves relevant information, understands context, generates natural language responses, and maintains conversation coherence. This multi-functionality is achieved through a single integrated architecture that processes both query and context in one pass
Data Source
AI summary
A method and server for training a chatbot system to generate machine-generated answers to users' requests of users are provided. The method comprises: acquiring dialogue data including textual representations of dialogue pairs; acquiring fact data including textual representations of facts; identifying, from the fact data, for a given dialogue pair, a fact relevant thereto; generating a training set of data including a plurality of training digital objects, a given one of which includes the textual representations of: (i) a human request of the given dialogue pair; (ii) the fact; and (iii) a human answer of the given dialogue pair, responsive to the human request; feeding the training set of data to the chatbot system, causing the chatbot system to generate a machine-generated answer; and optimizing a difference between the machine-generated answer and the human answer, thereby training the chatbot system to generate the machine-generated answers to the users' requests.