Network Database Search Entropy Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network database systems face inefficiencies in handling search queries, leading to low precision and recall, with large numbers of irrelevant datasets retrieved, exhaustion of system resources, and failure to provide up-to-date data, which compromises data integrity and network efficiency.
Innovation Solution
A computer-implemented method and apparatus that dynamically generates taxonomy refinements by calculating taxonomy entropy based on attributes, ranking datasets, and aggregating attributes to improve search result relevance, reducing resource consumption and enhancing precision and recall.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional network database search methods are used, then all datasets matching the query are retrieved, but the precision is low and large numbers of irrelevant datasets are returned
Solution Approach 1:
The system pre-calculates taxonomy entropy values for different datasets before the search query is executed. When a search is performed, these pre-computed entropy values are immediately available to filter results, avoiding the need to calculate entropy in real-time and enabling fast precision improvement without adding computational overhead during query execution.
Solution Approach 2:
Taxonomy entropy serves as an intermediary metric that bridges the gap between raw dataset attributes and search result relevance. By introducing this intermediate calculation layer, the system can objectively rank datasets based on their informational content and filter out irrelevant results more effectively than traditional keyword-matching methods.
2Reliability
If comprehensive dataset retrieval is performed to ensure high recall, then up-to-date data is provided, but system resources are exhausted
Solution Approach 1:
The system changes the parameter of dataset evaluation from simple presence/absence to taxonomy entropy-based relevance scoring. By transforming how datasets are evaluated and ranked, the system can identify and retrieve only the most relevant up-to-date datasets rather than all matching datasets, reducing resource consumption while maintaining data freshness and reliability.
Solution Approach 2:
Datasets are automatically evaluated and ranked based on their intrinsic taxonomy attributes without requiring manual intervention or complex external validation. The entropy calculation inherently identifies the most relevant datasets, allowing the system to self-optimize resource allocation by focusing computation on high-value results.
3Measurement precision
If taxonomy entropy calculation is performed on all datasets, then search precision is improved, but calculation complexity increases
Solution Approach 1:
The taxonomy entropy calculation is segmented into discrete steps: normalizing individual taxonomy attributes, calculating logarithmic values for each attribute, determining weight values, computing weighted logarithms, and summing them to get the final entropy. This segmentation makes the complex calculation more manageable and allows for selective application to only the most relevant datasets.
Solution Approach 2:
Instead of calculating taxonomy entropy for all datasets in the database, the system applies the calculation only to datasets that are candidates for the search results. This partial action approach maintains high ranking accuracy for relevant results while avoiding the excessive computational complexity of processing the entire dataset collection.
Data Source
AI summary
Methods, apparatus and computer program products for improving network database functionalities are discussed herein, such as a computer-implemented method including receiving a network database search query associated with one or more network databases; in response to receiving the network database search query, querying the one or more network databases to identify a plurality of datasets; determining a plurality of taxonomy attributes, each of the plurality of taxonomy attributes being associated with at least one dataset of the plurality of datasets; calculating a taxonomy entropy based on the plurality of taxonomy attributes; comparing the taxonomy entropy to an entropy threshold; in response to the taxonomy entropy satisfying the entropy threshold, generating a dataset ranking based on relevance scores associated with the plurality of datasets; determining one or more taxonomy refinements based on the dataset ranking; and outputting the plurality of datasets and the one or more taxonomy refinements.