Insight and algorithmic clustering for automated synthesis

• US 10,318,503 B1
• Filed: 03/27/2017
• Issued: 06/11/2019
• Est. Priority Date: 07/20/2012
• Status: Active Grant

• Current Assignee: OOL LLC
• Original Assignee: OOL LLC
• Inventors: Swamy, Gitanjali
• Primary Examiner(s): Hoang, Son T
• Application Number:
  US15/470,298
• Time in Patent Office: 806 Days
• Field of Search: None
• CPC Class Codes: G06F 16/217 Database tuning G06F16/2282 takes precedence; database performance monitoring G06F11/3409
• CPC Class Codes: G06F 16/2428 Query predicate definition using graphical user interfaces, including menus and forms G06F16/2423 takes precedence
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/3326 using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages
• CPC Class Codes: G06F 16/3328 using graphical result space presentation or visualisation
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 18/24317 Piecewise classification, i.e. whereby each classification requires several discriminant rules
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding

Abstract

A decision support system and method, which receives user inputs comprising: at least one user criterion, and at least one user input tuning parameter representing user tradeoff preferences for producing an output; and selectively produces an output of tagged data from a clustered database in dependence on the at least one user criterion, the at least one user input tuning parameter, and a distance function; receives at least one reference-user input parameter representing the at least one reference-user'"'"'s analysis of the tagged data and the corresponding user inputs, to adapt the distance function in accordance with the reference-user inputs as a feedback signal; and clusters the database in dependence on at least the distance function, wherein the reference-user acts to optimize the distance function based on the user inputs and the output, and on at least one reference-user inference.

Claims

1. A method for selecting at least one data element of a data set from a data space, comprising:

receiving a plurality of data elements from a plurality of reference users, each respective reference user having a reference user profile defining respective contextual domains of reference user expertise within the data space;

receiving feedbacks relating to prior selections of the at least one data element;

receiving a user input relating to a desired selection of the at least one data element;

determining a context of the user input;

synthesizing a synthetic reference user by selecting at least two, and less than all, of the reference users from the plurality of reference users, in dependence on at least the feedbacks relating to the at least one data element and the respective reference user profiles, to represent aggregate expertise in the determined context;

defining an optimized multidimensional distance function operative within the data space to define a reduced effective dimensionality of the data set, selectively dependent on at least the synthetic reference user, and the received feedbacks, wherein the optimized multidimensional distance function differs in dependence on at least which of the plurality of reference users are selected for synthesis of the synthetic reference user; and

processing the received user input in conjunction with the data set, dependent on the defined optimized multidimensional distance function, to select the at least one data element of the data set.

2. The method according to claim 1, further comprising clustering the data set dependent on the optimized multidimensional distance function, wherein the processing the received user input comprises processing the received user input in conjunction with the clustered data set, and selecting at least one cluster of the clustered data set.

3. The method according to claim 1, wherein the synthesizing the synthetic reference user is further dependent on at least one fitness criterion, the at least one fitness criterion being independent of the optimized multidimensional distance function.

4. The method according to claim 1, wherein each of the received plurality of data elements comprises an inference.

5. The method according to claim 1, wherein each of the received plurality of data elements comprises a recommendation.

6. The method according to claim 1, wherein each of the received plurality of data elements comprises a prediction.

7. The method according to claim 1, wherein the data set comprises a set of natural language documents, and the user input comprises a semantic query of the natural language documents.

8. The method according to claim 1, wherein the data set comprises unstructured data having context-dependent ambiguity, and the user input comprises a semantic query.

9. The method according to claim 1, wherein the selecting of at least one, and less than all, of the reference users from the plurality of reference users is in dependence on at least a relationship between the determined context and the defined respective contextual domains of reference user expertise within the data space.

10. The method according to claim 1, further comprising receiving:

at least one user input tuning parameter representing a user tradeoff preference between at least two competing criteria; and

defining the optimized multidimensional distance function operative within the data space of the data set based on the user tradeoff preference.

11. The method according to claim 1, wherein the selection of the at least one data element of the data set further comprises ranking the plurality of data elements of the data set.

12. The method according to claim 1, further comprising:

receiving second feedback relating a user concurrence with the selected at least one data element of the data set; and

adaptively clustering the plurality of data elements of the data set in dependence on at least the second feedback.

13. The method according to claim 12, wherein the second feedback comprises a user valuation or rating of at least one respective data element, and wherein the selection of the at least one data element of the data set further comprises ranking the plurality of data elements of the data set in dependence on at least the valuation or rating of the at least one respective data element.

14. A system for selecting a record in a database, comprising:

an automated database configured to store a plurality of records, each record corresponding to an identifier of a reference user of a plurality of reference users, an input received from a respective reference user, and feedback relating to the input received from the reference user, each respective reference user having a reference user profile defining respective contextual domains of expertise;

an input port configured to receive a user input relating to a desired selection of a subset of the plurality of records, and context information associated with the user input; and

at least one automated processor, configured to;

determine a context associated with the user input based on at least the context information;

synthesize a synthetic reference user representing an aggregate expertise in the determined context based on a selection of at least two reference users, and less than all of the reference users from the plurality of reference users, in dependence on at least the feedbacks of the plurality of reference users relating to the respective received inputs, the determined context, and the reference user profiles of the plurality of reference users;

define an optimized multidimensional distance function to define a reduced effective dimensionality of the plurality of inputs, based on at least the plurality of records, the synthetic reference user, and the feedbacks, wherein the optimized multidimensional distance function differs in dependence on at least which of the plurality of reference users are selected; and

process the received user input in conjunction with the plurality of records, and dependent on the defined optimized multidimensional distance function, to select at least one record.

15. The system according to claim 14, wherein the at least one automated processor is further configured to cluster the plurality of records dependent on the optimized multidimensional distance function, and to output at least one of the selected one record and an identification of selected one record.

16. The system according to claim 14, wherein the at least one automated processor is further configured to rank the selected plurality of records dependent on at least the optimized multidimensional distance function.

17. A method for selecting a data record within a data space, comprising:

receiving information from a plurality of reference user sources, each respective reference user source having a reference user source profile defining respective contextual domains of expertise of a respective reference user within the data space;

storing the received information in data records of a database in conjunction with an identification of the respective reference user, feedback relating to a quality of the received information from the respective reference user, and a context domain of the received information from the respective reference user;

receiving a user query for retrieval of information from the database;

determining a context domain of use for the user query;

synthesizing a synthetic reference user from at least two, but not all, of the plurality of reference user sources, in dependence on at least the feedback, the reference user source profile of the plurality of reference user sources, and a relation of the context domain of the respective information and the determined context of use;

defining an optimized multidimensional distance function, selectively dependent on at least the synthetic reference user, wherein the optimized multidimensional distance function differs in dependence on at least which of the respective reference user sources are omitted from the synthetic reference user;

clustering the database using the optimized multidimensional distance function in a context domain dependent manner, to reduce an effective dimensionality of the data records in the database;

and

processing the received user query in conjunction with the clustered database, to select a respective data record from the database as a result of the received user query.

18. The method according to claim 17, wherein the synthesizing the synthetic reference user is further dependent on at least one fitness criterion, the at least one fitness criterion being independent of the optimized multidimensional distance function.

19. The method according to claim 17, wherein the received information comprises at least one of an inference, a recommendation, a prediction, a natural language document, and unstructured data having context-dependent ambiguity, and the processing comprises ranking a plurality of the data records in the database.

20. The method according to claim 17, wherein the optimized multidimensional distance function operative is further dependent on a user tradeoff preference between at least two competing criteria.