Search engine results system using entity density
View Patent ↗Architecture that enables search engines to meet user expectations for search results (e.g., questions-answers) by improving on result consistency. This is attained by declining to answer queries when it is known by the system that the system is unable to answer more or equivalently prominent queries in the same query class in a vast majority of cases. To achieve consistency, queries are categorized into classes and then the queries of a specific class are segmented into clusters. The answer density in each cluster is then computed to determine the consistency of that class of queries. Clusters with a relatively low answer density are then suppressed to improve consistency for the user.
1. A system, comprising:
a query processing component configured to deconstruct queries of a class into chunks of entities and to cluster the chunks into clusters;
a density component configured to compute result densities for the corresponding clusters and compare the result densities of the clusters to density thresholds to determine result consistency of the clusters;
a results component configured to return results for a query based on favorable comparison of the result densities to the corresponding density thresholds and suppress results for the query based on unfavorable comparison of the result densities to the corresponding density thresholds, to maintain the result consistency; and
at least one hardware processor configured to execute computer-executable instructions in a memory, the instructions executed to enable the query processing component, the density component, and the results component.
2. The system of claim 1 , wherein the query processing component is configured to identify and extract the chunks of the entities from query logs.
3. The system of claim 1 , wherein the density thresholds are empirically defined based on a rate of abandonment or re-queries within a same query cluster.
4. The system of claim 1 , wherein the query is a question and the results are answers, and the system is a question-answer system.
5. The system of claim 1 , wherein the results component maintains result consistency by declining to answer more or equivalently prominent queries in a same query class.
6. The system of claim 1 , further comprising a threshold component configured to determine and adjust the density thresholds of the clusters to maintain the result consistency.
7. The system of claim 6 , wherein the threshold component is configured to determine the density thresholds on an absolute basis.
8. The system of claim 6 , wherein the threshold component is configured to determine the density thresholds on a cross-sectional basis.
9. The system of claim 6 , wherein the threshold component is configured to determine the density thresholds expressed as a percentage of a density of clusters having a greater saliency.
10. A method, comprising acts of:
computing result densities of clusters of entity chunks, the chunks derived from a class of queries;
comparing the result densities of the clusters to density thresholds to determine a result consistency of the clusters;
returning results for a query based on favorable comparison of the result densities to the corresponding density thresholds; and
suppressing results for the query based on unfavorable comparison of the result densities to the corresponding density thresholds, to maintain the result consistency.
11. The method of claim 10 , further comprising suppressing a cluster with unacceptable result densities.
12. The method of claim 10 , further comprising computing the result consistency of a class and adjusting the result consistency to improve predictability.
13. The method of claim 10 , further comprising expanding the class of queries into chunks of the entities using a lexicographic expansion method.
14. The method of claim 13 , further comprising identifying and extracting actual chunks of the entities from query logs.
15. The method of claim 10 , further comprising defining suitable answers of a cluster according to at least one of an absolute basis or a cross-sectional basis.
16. A method, comprising acts of:
deconstructing queries of a class into chunks of entities and clustering the chunks into clusters;
computing answer densities for the corresponding clusters;
comparing the answer densities of the clusters to answer density thresholds to determine answer consistency of the clusters; and
returning answers for a query based on favorable comparison of the answer densities to the corresponding answer density thresholds and suppressing answers for the query based on unfavorable comparison of the answer densities to the corresponding answer density thresholds to maintain answer consistency.
17. The method of claim 16 , further comprising adjusting a density threshold to maintain answer consistency of an associated cluster.
18. The method of claim 16 , further comprising defining suitable answers of a cluster according to at least one of an absolute basis or a cross-sectional basis.
19. The method of claim 16 , further comprising identifying and extracting the chunks of the entities from query logs.
20. The method of claim 16 , further comprising clustering the queries according to relative frequencies as relate to identification of the queries in query logs, and separating the clusters based on an expected number of requests of information from the clusters.