Natural Language Dialogue System Using Preference-Based Answer Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional natural language dialogue systems face challenges in accurately recognizing user intentions due to rigid input rules and the inability to handle diverse user inputs, leading to incorrect recognitions and inefficient information retrieval, as they rely on predefined phrase lists and full-text searches that yield non-structured data.
Innovation Solution
A natural language dialogue system that includes a properties database and a speech sampling module, which parses keywords from user inputs to generate a candidate list of report answers, selects the appropriate answer based on user preferences and habits, and outputs a response, while also capturing and storing user preference data for future interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional full-text search is used to retrieve information, then the system can handle diverse user inputs, but the search results are non-structured and require users to browse multiple pages to find useful information
Solution Approach 1:
The patent segments the unstructured search results into structured data fields (e.g., title, content, source, time) and organizes them in a tabular format. This segmentation allows the system to present only the most relevant information prominently, eliminating the need for users to browse through multiple pages of unstructured results.
Solution Approach 2:
The patent transforms the traditional linear browsing experience into a multi-dimensional structured presentation. By organizing results with hierarchical fields and attributes, the system adds dimensions of organization (categories, metadata, relevance scoring) that allow users to quickly locate information without sequential browsing.
2Measurement precision
If predefined phrase lists are used for intention recognition, then the system can accurately recognize user intentions, but the system cannot handle colloquial or diverse user expressions
Solution Approach 1:
The patent implements a dynamic intention recognition system that adapts to different user expressions. Instead of relying on static predefined phrase lists, the system uses machine learning models that can dynamically interpret various colloquial expressions and map them to intended meanings, allowing both accurate recognition and handling of diverse inputs.
Solution Approach 2:
The system incorporates feedback mechanisms where user interactions and corrections are used to continuously improve intention recognition accuracy. The system learns from user behavior patterns and adjusts its interpretation of diverse expressions, maintaining high accuracy while becoming more adaptable to new colloquialisms.
3Reliability
If multiple possible interpretations are provided for ambiguous user inputs, then the system can avoid incorrect recognitions, but the user has to make further selections which is redundant and inefficient
Solution Approach 1:
The patent applies preliminary action by using contextual information and user preference data to pre-select the most likely intention before presenting options to the user. The system analyzes conversation history, user profile, and contextual cues to narrow down possible interpretations, presenting only the top candidate or a limited set of plausible options rather than all possible interpretations.
Solution Approach 2:
The system performs self-service by automatically resolving ambiguous intentions using contextual analysis and user preference data. When the system can confidently determine the user's intention through contextual clues, it directly executes the intended action without requiring user confirmation, thereby maintaining reliability while improving efficiency.
Data Source
AI summary
A natural language dialog method and a natural language dialog system are provided. In the method, a first speech input is received and parsed to generate at least one keyword included in the first speech input, so that a candidate list including at least one report answer is obtained. According to a properties database, one report answer is selected from the candidate list, and a first speech response is output according to the report answer. Other speech inputs are received, and a user's preference data is captured from the speech inputs. The user's preference data is stored in the properties database.