Search Query Term Optionalization for Ecommerce Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eCommerce search engines face challenges in accurately processing long queries with noisy terms, leading to reduced search accuracy and increased computing resources, resulting in poor user experience and performance issues as they consider all terms for text match without context or user error.
Innovation Solution
A system that determines the optionalization probability of search terms in a query, allowing for the retention or removal of terms based on their relevance, using algorithms that analyze bigrams, trigrams, and statistical signals to generate more relevant search results, thereby improving search accuracy and reducing computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all search terms in a query are considered for text match, then recall is improved, but search accuracy deteriorates due to noisy terms
Solution Approach 1:
The patent extracts and removes noisy or non-informative terms from the search query before processing. The system identifies terms that contribute little to search accuracy (such as common stop words, grammatical particles, or clearly irrelevant terms) and excludes them from the text matching process, thereby improving search accuracy without losing important information.
Solution Approach 2:
The patent applies different processing qualities to different terms in the query. Instead of treating all terms uniformly, the system evaluates each term's importance and applies selective weighting or filtering. Important terms receive full processing attention while noisy terms are downweighted or removed, creating a non-uniform quality approach that improves overall search accuracy.
2Reliability
If every product record in the database is retrieved and sorted, then search completeness is improved, but computing time and resources increase
Solution Approach 1:
The patent performs preliminary actions by pre-processing and indexing product data before search queries are executed. Product records are pre-tagged with relevant keywords, categories, and metadata that enable rapid filtering. When a search query arrives, the system uses these pre-computed indexes to quickly identify candidate products without scanning the entire database, thus maintaining completeness while reducing computing time.
Solution Approach 2:
The patent extracts only the necessary subset of product records that are relevant to the search query, rather than retrieving and sorting all product records. By using the cleaned and optimized search terms to filter the database immediately, the system extracts only the candidate products that match the query criteria, significantly reducing the number of records that need to be processed and sorted.
3Quantity of substance
If search algorithms consider every term for text match, then recall is improved, but large recall sets are generated reducing search accuracy
Solution Approach 1:
The patent removes noisy terms from the query that would otherwise generate unnecessary recall. By eliminating terms that contribute to large recall sets but add little value (such as common words, stop words, or clearly irrelevant terms), the system reduces the size of the recall set while preserving the important terms that are necessary for accurate search results.
Solution Approach 2:
The patent applies different processing qualities to different terms, with important terms receiving full attention and noisy terms being downweighted or removed. This local quality approach ensures that the recall set is generated based on meaningful terms only, improving the precision of the search results without sacrificing recall of relevant products.
Data Source
AI summary
A system for displaying product information to a customer is described herein. The system includes a database and a server including a processor. The processor is programmed to receive a search query including a plurality of search terms via a computing device. The processor selects a search term from the plurality of search terms and determines an optionalization probability associated with the selected search term. The processor generates a search request as a function of the search query and determines whether to retain the selected search term with the search request as a function of the optionalization probability. The processor retrieves product records from the database as a function of the search request, and generates and displays a product search result including product information associated with the retrieved product records.