Researchers introduce NBQ (Next-Best-Question), a conversational AI framework that dynamically profiles users by asking strategically optimized questions to maximize information gain. The system improves user profiling accuracy by up to 14% and includes QuickMatch, an efficient retrieval layer for reciprocal matching that accelerates search by 22.9x, with applications in hiring, marketplaces, and dating platforms.
NBQ represents a meaningful advancement in conversational AI systems designed to understand people through structured dialogue. Rather than asking random or pre-determined questions, the framework prioritizes questions that generate the highest expected information gain, treating user profiling as an optimization problem. This approach mirrors how skilled interviewers naturally conduct conversations, but automates the decision-making process with algorithmic rigor.
The framework addresses a genuine friction point in digital marketplaces and hiring platforms, where understanding user preferences and compatibility requires nuanced information gathering. Traditional static profiling questionnaires often suffer from low completion rates or shallow insights. NBQ's adaptive approach maintains a continuously updated user state, allowing it to ask contextually relevant follow-up questions rather than fixed forms.
The introduction of QuickMatch demonstrates pragmatic thinking about scalability. Reciprocal matching problems—where mutual compatibility matters—inherently require quadratic computational complexity if solved naively. By recasting the matching problem as approximate vector search, QuickMatch achieves 22.9x speedup while maintaining 0.989 recall, making real-time matching feasible at scale.
The 13.6-14% improvement in profiling metrics (AC@T and AR@T) suggests meaningful gains in both accuracy and coverage. This technology has immediate applications for dating platforms, recruitment systems, and talent marketplaces where effective matching drives user retention and marketplace liquidity. The modular "plug-and-play" design enables integration across different platforms and use cases.
- →NBQ optimizes conversational profiling by asking questions with highest expected information gain, improving profiling accuracy by up to 14%.
- →QuickMatch acceleration technique reduces reciprocal matching complexity from quadratic to approximate vector search, enabling 22.9x faster retrieval.
- →Framework maintains continuous user state updates, enabling adaptive questioning rather than static questionnaires.
- →Technology has direct applications in hiring, dating platforms, and marketplace matching where mutual compatibility is critical.
- →High recall rate of 0.989 in matching retrieval suggests minimal loss of quality candidates despite significant speed improvements.