Chatbot Response Generation Using Document Sentence Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional chatbot systems rely on predefined query-response pairs, which limits their adaptability and ability to provide suitable responses to diverse messages, especially with the explosive growth of social media data, leading to inaccurate and unsuitable responses.
Innovation Solution
The chatbot system matches received messages with sentences from documents rather than predefined query-response pairs, determining relevance and suitability for the ongoing conversation, allowing for adaptive and meaningful responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional chatbot systems use predefined query-response pairs, then the system structure is simple and easy to implement, but the adaptability to diverse messages and topics is limited
Solution Approach 1:
The patent transforms the static query-response pair system into a dynamic document-based system. Instead of fixed Q-R pairs, the system now dynamically retrieves and generates responses from documents based on message relevance, enabling adaptability to diverse topics while maintaining manageable complexity through automated processing.
Solution Approach 2:
The patent changes the fundamental parameter of response generation from matching predefined queries to retrieving relevant sentences from documents. This parameter change enables the system to handle diverse messages and topics effectively, as documents can cover broader content ranges compared to fixed Q-R pairs.
2Reliability
If chatbot systems store a set of existing query-response pairs in advance, then the response generation is fast and efficient, but the response accuracy and suitability for ongoing conversations deteriorate
Solution Approach 1:
The patent applies preliminary action by pre-processing and storing sentences from documents in a structured format before they are needed for response generation. This allows the system to quickly retrieve and evaluate sentences during conversation without extensive real-time processing, maintaining efficiency while improving accuracy through document-based content.
Solution Approach 2:
The patent implements feedback mechanisms by evaluating the suitability of retrieved sentences in the context of ongoing conversations. The system determines whether selected sentences are appropriate for the conversation flow and adjusts response generation accordingly, improving response accuracy while maintaining efficient processing through targeted evaluation.
3Ease of operation
If the chatbot system matches messages with predefined queries, then the matching process is simple and quick, but the ability to provide meaningful and contextually suitable responses is limited
Solution Approach 1:
The patent introduces an intermediary mechanism by using sentence retrieval and suitability evaluation as intermediate steps between message input and response generation. This intermediary process enables the system to maintain matching simplicity while preserving contextual information, as the system can evaluate multiple sentences and select the most contextually appropriate one.
Data Source
AI summary
In implementations of the subject matter described herein, a new approach for presenting a response to a message in a conversation is proposed. Generally speaking, in response to receiving a message in a conversation, the received message will be matched with one or more documents on the sentence basis. That is, the received message is compared with the sentences from a document(s), rather than predefined query-response pairs. In this way, a whole sentence may be selected from the document as a candidate response. Then the suitability of this sentence with respect to the ongoing conversation will be determined, and the response will be generated and rendered in an adaptive way based on the suitability. As a result, the user experiences may be significantly enhanced in the chatbot scenario.