Extrapolative Search Query Expansion for Stock Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tools for searching stock images require significant user effort and often yield limited results due to narrow search scopes, and users are uncertain about the usage rights associated with the images found.
Innovation Solution
The implementation of extrapolative search techniques that expand specific queries into generalized queries by analyzing text to identify theme phrases, extracting keywords, and categorizing named entities, thereby expanding the search scope and providing users with a wider range of relevant image options while ensuring informed selection based on usage rights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users craft specific search queries to find images, then search precision is improved, but search scope becomes too narrow and results are limited
Solution Approach 1:
The search query is segmented into specific keywords and generalized terms. The system identifies named entities in the user's specific query and separates them from other terms, allowing the specific part to maintain precision while the generalized part expands scope.
Solution Approach 2:
The system adds a dimensional expansion by introducing generalized search terms alongside specific keywords. This creates a multi-dimensional search approach where queries operate at both specific and general levels simultaneously, expanding the search space without losing precision.
2Ease of operation
If users manually craft search queries, then search control is improved, but user effort and time consumption increase
Solution Approach 1:
The system performs automatic named entity recognition and query generalization without requiring user intervention. The search engine autonomously identifies entities, retrieves generalized terms, and constructs expanded queries, allowing the system to serve itself rather than requiring continuous user input.
Solution Approach 2:
The system pre-loads and maintains a database of generalized terms for common named entities. This preliminary preparation allows the query expansion to occur rapidly during actual search operations, reducing user waiting time and effort.
3Adaptability or versatility
If generalized terms are used to expand search scope, then result diversity is improved, but search precision may decrease
Solution Approach 1:
The query is segmented into specific keywords that maintain original precision and generalized terms that expand scope. By keeping both components in the search strategy, the system preserves precision through specific terms while gaining diversity through generalized terms.
Solution Approach 2:
Different parts of the search strategy serve different quality requirements: specific keywords provide local precision for the user's intent, while generalized terms provide local diversity for broader results. Each component optimizes for its specific quality attribute.
Data Source
AI summary
Techniques for extrapolative searches are described herein. In one or more implementations, a searches are conducted using an extrapolative and additive mechanism that expands a specific query into one or more generalized queries. To do so, keywords contained in an input search query are extracted to use as a basis for a search related to content. The extracted keywords are expanded by categorization of name entities recognized from the keywords into corresponding generalized terms. Query strings to use for the search are built using combinations of keywords that are extracted from the text and corresponding generalized terms obtained by expanding the keywords. A search is conducted using the expanded queries and images results returned by the search are exposed as suggested images to represent the content.