Query Suggestions from User-Generated Content Bigrams
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online shopping search engines rely on historical search entries for query suggestions, which are ineffective for users unfamiliar with products, leading to repetitive and inefficient searches.
Innovation Solution
A computer-implemented method that generates query suggestions by extracting frequently repeated bigrams from user-generated content such as product reviews and blogs, combining them to form composite phrases, and presenting these suggestions to users for more accurate searches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If query suggestions are generated from historical search entries, then the system can provide suggestions based on past user behavior, but the suggestions become ineffective for users unfamiliar with products leading to repetitive searches
Solution Approach 1:
The patent introduces user-generated content (product reviews, questions, answers) as an intermediary data source between the search engine and query suggestions. Instead of directly using historical search entries, the system uses UGC as a mediator to infer user needs and generate more accurate suggestions that reflect actual product knowledge and user perspectives.
Solution Approach 2:
The patent changes the fundamental parameter used for generating query suggestions from 'historical search entries' to 'user-generated content'. This parameter change enables the system to move from pattern-matching past searches to understanding actual user needs through product information and user experiences, thereby improving suggestion effectiveness.
2Loss of information
If the search engine uses historical search entries for suggestions, then it can leverage existing data, but it fails to account for product-specific information and user knowledge gaps
Solution Approach 1:
The patent segments the information gathering process into distinct components: product information extraction from UGC, user query analysis, and suggestion generation. This segmentation allows the system to process and combine different types of information (product specs, user reviews, search patterns) to generate more accurate and context-aware query suggestions.
Solution Approach 2:
The system incorporates feedback loops where user-generated content about products is continuously collected, analyzed, and used to improve query suggestions. This feedback mechanism ensures that the suggestions evolve based on actual user experiences and product information, making them increasingly accurate and relevant over time.
Data Source
AI summary
An improved query suggestion system can allow a user to more effectively search for items they would like to purchase online. The present invention provides a computer-implemented method for generating query suggestions based on user-generated content. A user may begin an online shopping venture by entering a keyword into the search function on a shopping website or other search engine interface. The search engine can launch a search of user generated content on the Internet, such as product related blogs, product reviews, or question-and-answer articles. The search engine can extract frequently repeated bigrams from the content. The search engine may combine the bigrams to generate candidate composite words and select a final query suggestion list from the candidate words. The query suggestion list is displayed to the user in the search interface. The user may select one of the suggested queries if it more accurately defines his search.