Recommendation Support Data Selection Balancing KPI Accuracy and Data Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing recommendation systems fail to consider the adequacy of data amount and granularity for KPI-specific recommendations, leading to potential excess or deficiency, which can burden customers or vendors with unnecessary data loads.
Innovation Solution
A recommendation support device and method that determines the suitability of data for recommendations by assessing the amount and granularity of model data candidates through a KPI determination unit, model data selection unit, contribution determination unit, and application determination unit to ensure appropriate data application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data amount and granularity are increased to improve recommendation accuracy for different KPIs, then recommendation accuracy is improved, but data load and business burden on customers and vendors increase
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting data granularity and amount based on the specific KPI being evaluated. Different KPIs (e.g., availability, maintenance cost, energy consumption) require different levels of data detail, so the system adapts data parameters to match each KPI's requirements rather than using a fixed data structure for all recommendations.
Solution Approach 2:
The patent implements local quality by applying different data granularity levels to different data elements based on their relevance to specific KPIs. Critical data elements related to the target KPI are maintained with high granularity, while less relevant elements use coarser granularity, optimizing the balance between recommendation accuracy and data load.
2Ease of operation
If data amount is reduced to decrease business burden on customers and vendors, then data load is reduced, but recommendation accuracy may deteriorate
Solution Approach 1:
The patent extracts only the essential data elements required for each specific KPI from the overall data set. By identifying and extracting only the relevant data needed for accurate recommendation, the system eliminates unnecessary data collection and processing, thereby reducing business burden while maintaining recommendation accuracy for the target KPI.
Solution Approach 2:
The patent applies partial action by collecting and processing only the sufficient amount of data needed for each KPI rather than all possible data. This avoids excessive data collection while ensuring that the minimum required data quality and quantity are maintained for accurate recommendations.
Data Source
AI summary
It is determined whether sufficient data can be prepared for a recommendation related to business. A recommendation support device includes a storage unit configured to store model data of data related to the business; a KPI determination unit configured to determine an amount of data for calculating an auxiliary variable for a KPI of a related person of the business; a model data selection unit configured to determine a granularity of a model data candidate selected from the storage unit; and a contribution determination unit configured to determine whether the model data candidate contributes to achieving the KPI. An application determination unit is configured to determine whether the selected model data candidate is applicable to the recommendation related to the business according to a determination result of the granularity, a determination result of the contribution, and a possibility of changing the KPI and the selected model data candidate.