Entity Detection for Selective Query Generalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search systems fail to return relevant results for highly-specific or 'noisy' spoken queries, often resulting in null queries or irrelevant results due to over-specified search terms, leading to user frustration.
Innovation Solution
Implementing an entity detection and recognition model based on statistical analysis of prior search queries to selectively generalize queries, identifying and categorizing entities and their attributes, and using these to refine search criteria, thereby improving the relevance and quality of search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional search systems use exact match search for spoken queries, then search precision is maintained, but search recall deteriorates when queries are over-specified or contain noise
Solution Approach 1:
The query is segmented into multiple search terms and attributes through entity detection and recognition. The system identifies which terms are essential entities versus descriptive attributes, allowing selective matching that maintains precision for core search terms while being flexible with descriptive ones, thus resolving the contradiction between exact match precision and relevance for noisy queries
Solution Approach 2:
The search matching parameters are changed from strict exact match to selective generalization based on entity importance. By detecting and recognizing entities with different levels of specificity, the system adjusts matching strictness dynamically - using exact match for critical entities and partial match for attributes, thereby maintaining both precision and relevance
2Reliability
If search systems generalize queries to handle noisy input, then search recall improves, but search precision deteriorates due to irrelevant results
Solution Approach 1:
Different parts of the query are treated with different matching strictness based on their entity type and importance. Critical entities receive exact match treatment while descriptive attributes receive flexible matching, allowing the system to maintain precision for core search terms while being accommodating to noisy input, thus resolving the contradiction between generalization and precision
3Quantity of substance
If search systems use all search terms from the query, then search completeness is maintained, but search efficiency deteriorates for null queries
Solution Approach 1:
The system extracts and identifies which query terms are essential entities versus optional attributes through entity detection. By separating these components, the system can construct search queries that include only the necessary terms for finding relevant results, avoiding the pitfall of using all terms (including noise) that would result in null queries, thus improving both efficiency and result quantity
Data Source
AI summary
Features are disclosed for processing user queries into a form that can produce relevant results. Spoken user queries can be transcribed into textual queries. Textual queries can be processed using a statistical model to identify entities within the queries. Running searches using the entities rather than the original search query can produce relevant results even when no result would have been obtained by running the original search query. In some embodiments, attributes may be identified and used during the search to narrow the results and potentially produce results that are more relevant.