Conversation System Reducing FAQ Turns via Key Accumulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional interactive-type FAQ systems require a large number of conversations to narrow down question sentence candidates effectively, as they rely on keyword presentation and response, leading to inefficiencies in specifying user question details.
Innovation Solution
A conversation system that accumulates keys from positive user responses and uses a key selecting unit to choose inquiry keys from candidate response details, reducing the number of conversations needed by focusing on high-priority keys and dependency relations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a conventional interactive-type FAQ system presents keywords and answers through matching, then the system can provide structured response options, but a large number of conversations are required to narrow down question sentence candidates effectively
Solution Approach 1:
The system performs preliminary analysis of the user's input sentence to extract candidate response details and their associated keys before the interactive questioning begins. This preliminary action allows the system to proactively identify the most relevant inquiry keys and potential answers, reducing the need for extensive back-and-forth conversations to narrow down the user's intent.
Solution Approach 2:
The patent replaces the conventional mechanical matching system (where users sequentially select from presented keywords) with an intelligent information processing system that uses natural language analysis, key extraction, and relevance scoring. This substitution enables the system to automatically understand and interpret user intent more efficiently, significantly reducing the conversation turns needed to specify question details accurately.
2Ease of operation
If the system presents only one keyword at a time in an interactive manner, then the user can provide focused responses, but the process requires many conversation turns to accumulate sufficient information
Solution Approach 1:
The system segments the user's input sentence into multiple candidate response details, each with associated keys and relevance scores. Instead of presenting information in a single linear sequence, the system divides the inquiry into multiple parallel candidate paths, allowing it to efficiently explore and evaluate different interpretation directions simultaneously, thereby reducing the total number of conversation turns needed.
Solution Approach 2:
The system dynamically adjusts its inquiry strategy based on the accumulated information and relevance scores. It flexibly changes the number and type of inquiry keys presented in each turn, adapting to the user's responses in real-time. This dynamic approach allows the system to optimize the information-gathering process, maintaining ease of user operation while significantly improving the efficiency of specifying response details.
Data Source
AI summary
A conversation device includes: an inquiry unit configured to output an inquiry including presentation of a key; an acquisition unit configured to acquire an answer; a key accumulating unit configured to accumulate keys included in the inquiry as accumulation keys in a case in which a positive answer for the inquiry is acquired; a candidate question sentence extracting unit configured to extract candidate question sentences including the accumulation keys by referring to a question sentence storing unit configured to store a question sentence and a set of keys included in the question sentence in association with each other; a key selecting unit configured to select an inquiry key that is caused to be included in a next inquiry output by the inquiry unit among a plurality of keys included in the candidate question sentences; and an output unit configured to output one question sentence among the candidate question sentences.