Recursive Query Expansion for Multi-Metric Driver Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based CRM systems struggle to efficiently expand the scope of user queries to determine underlying factors affecting data metrics, often resulting in significant processing overhead and inefficiencies.
Innovation Solution
Implementing a system that uses a large language model (LLM) to generate summaries of natural language queries, vectorizes these summaries for database analysis, and recursively expands query scopes to adjacent data metrics while limiting processing overhead through a set recursion depth, supported by offline indexing and vector databases for secure multi-tenant environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If the system searches the full set of CRM data to detect driving factors, then the completeness of query scope expansion is improved, but the processing overhead increases significantly
Solution Approach 1:
The system performs preliminary action by generating summaries of data metrics and their relationships in advance, storing them in a vector database. When a user query arrives, the system can quickly retrieve pre-computed summaries instead of searching the entire CRM data set, thus maintaining completeness while reducing processing overhead.
Solution Approach 2:
The system extracts essential information from large volumes of CRM data by generating concise summaries that capture driving factors and relationships. These extracted summaries are stored and can be quickly retrieved, avoiding the need to process the full data set for every query while maintaining the ability to detect relevant driving factors.
2Productivity
If the system fails to expand the scope of queries, then the processing overhead is reduced, but the ability to detect underlying driving factors is compromised
Solution Approach 1:
The system performs preliminary analysis by generating summaries of data metrics and their relationships in advance. These pre-computed summaries contain information about driving factors, enabling the system to quickly respond to queries without performing extensive real-time searches, thus maintaining both efficiency and detection capability.
Solution Approach 2:
The system introduces summaries as an intermediary layer between the user query and the underlying CRM data. These summaries act as a bridge that contains pre-analyzed information about driving factors, allowing the system to detect underlying factors efficiently without directly searching the full data set.
3Measurement precision
If the system recursively expands query scope to adjacent data metrics, then the accuracy of identifying driving factors is improved, but the processing overhead increases
Solution Approach 1:
The system performs preliminary action by pre-computing and storing summaries of data metrics and their relationships in a vector database. This allows the recursive expansion to operate on pre-analyzed summaries rather than raw data, improving accuracy while reducing the complexity and overhead of recursive processing.
Solution Approach 2:
The system segments the complex task of recursive query expansion into manageable steps: generating summaries of individual metrics, storing them in a vector database, and then recursively querying these summaries. This segmentation reduces processing complexity while maintaining the ability to accurately identify driving factors through multi-level expansion.
Data Source
AI summary
Methods, systems, apparatuses, devices, and computer program products are described. A processing device may receive a natural language query asking a question about a data metric. The processing device may use a large language model (LLM) to generate a summary of the natural language query for vector embedding. The processing device may determine one or more query response portions indicating possible answers to the query based on the summary and a vector database including vector representations of data summaries. To expand the scope of the answers, the processing device may recursively expand a set of data metrics for analysis. For example, the processing device may determine additional data metrics adjacent to the data metric of the query and may search the vector database for additional query response portions based on the additional data metrics. The processing device may use the query response portions to answer the natural language query.