Search Refinement Scoring Using Term Frequency and Category Hierarchy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Commercial product search devices fail to weight information effectively, leading to improper terms for search refinement, such as generic names and category names being presented equally.
Innovation Solution
An information providing device that obtains a ranking of commercial products, collects relevant terms from sales pages or search queries, calculates a score based on term frequency, and presents suggested keywords for search refinement, using a category hierarchy defined by a tree structure to adjust scores based on term frequency and category similarity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If commercial product search devices present searched information equally without weighting, then the presentation is simple and easy to implement, but the search refinement terms become improper (generic names and category names are presented equally)
Solution Approach 1:
The patent applies parameter changes by introducing a scoring mechanism that assigns different weights to terms based on their collection frequency and category relevance. Terms are scored according to multiple parameters including the number of collections, category hierarchy depth, and term specificity, transforming the flat presentation into a weighted ranking system that prioritizes more relevant terms for search refinement.
2Reliability
If terms are weighted based on collection frequency and category similarity, then appropriate suggested keywords are provided for search refinement, but the calculation and presentation system becomes more complex
Solution Approach 1:
The patent applies local quality by differentiating the treatment of terms based on their specific characteristics and context. Each term is evaluated locally within its category context, with scoring adjusted according to category hierarchy level, term specificity, and collection patterns. This localized evaluation ensures that generic category names receive lower weights than specific product terms, while the overall system maintains a unified scoring framework.
Data Source
AI summary
An obtaining unit (120) obtains a ranking of a commercial product belonging to a category defined on an e-marketplace. A collecting unit (121) collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed. A calculating unit (122) calculates a score of each collected term based on at least a number of collections of the term. A presenting unit (123) presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score in an upper ranking position.