Techniques to use machine learning for risk management

Abstract

Techniques to use machine learning for risk management are described. Some techniques build and train a machine learning model with features of which at least one feature corresponds to risk indicia and at least one other feature corresponds to a data source. These features, in general, provide data (e.g., values) indicating a degree of relevance between a particular record and a risk assessment of that particular record's subject matter. User refinement in the form of user selections and other interactions with the particular record and other records provide insights into proper risk management. The techniques leverage such user refinement to improve upon an accuracy of the machine learning model. Driven by such insights, these techniques enhance the machine learning model with accurate feature values and feature weights to perform risk assessment. Other embodiments are described and claimed.

1 . An apparatus, comprising:
 a processing circuit; and   
       logic stored in computer memory and executed on the processing circuit, the logic operative to cause the processing circuit to:
 perform feature extraction on records across a plurality of datasets, features corresponding to the feature extraction comprise at least one feature associated with risk indicia and at least one feature associated with a particular data source of at least one of the plurality of datasets;
 train a machine learning model based upon data generated during the feature extraction on the records; 
 apply the machine-learning model to determine one or more records from the records having risk assessment data that exceeds a baseline threshold; and 
 provide at least one record of the one or more records in response to a search query. 
 
 
     
     
         2 . The apparatus of  claim 1 , the processing circuit to:
 receive a user selection corresponding to the at least one record of the one or more records returned in the response to the search query; and   modify the machine-learning model in response to the user selection.   
     
     
         3 . The apparatus of  claim 2 , the processing circuit to:
 process, during the search query, the user selection; and   update risk assessment values for each of the at least one record for the machine-learning model, each of the risk assessment values indicative of a level of risk associated with subject matter of a particular record.   
     
     
         4 . The apparatus of  claim 3 , wherein updating the risk assessment value comprises adjusting feature values to increase or decrease the risk assessment value and/or adjusting feature weights associated with a particular record. 
     
     
         5 . The apparatus of  claim 2 , wherein the user selection corresponding to the at least one record is risk indicia indicating risk of a product corresponding to a particular record. 
     
     
         6 . The apparatus of  claim 2 , the processor circuit to modify, in response to the user selection, the search query by adding, removing, or replacing one or more search terms of the search query and perform the search with the modified search query. 
     
     
         7 . The apparatus of  claim 2 , the processor circuit to add or remove content from a return record of the provided at least one record in response to the user selection. 
     
     
         8 . A computer-implemented method, comprising:
 performing feature extraction on records across a plurality of datasets, features corresponding to the feature extraction comprise at least one feature associated with risk indicia and at least one feature associated with a particular data source of at least one of the plurality of datasets;   applying a trained machine-learning model to determine one or more records from the records having risk assessment data that exceeds a baseline threshold; and   providing at least one record of the one or more records in response to a search query.   
     
     
         9 . The computer-implemented method of  claim 8 , comprising:
 receiving a user selection corresponding to the at least one record of the one or more records returned in the response to the search query; and   modifying the machine-learning model in response to the user selection.   
     
     
         10 . The computer-implemented method of  claim 9 , comprising:
 processing the user selection; and   updating risk assessment values for each of the at least one record for the machine-learning model, each of the risk assessment values indicative of a level of risk associated with subject matter of a particular record.   
     
     
         11 . The computer-implemented method of  claim 10 , wherein updating the risk assessment value comprises adjusting feature values to increase or decrease the risk assessment value and adjusting feature weights associated with a particular record. 
     
     
         12 . The computer-implemented method of  claim 9 , wherein the user selection corresponding to the at least one record is risk indicia indicating risk of a product corresponding to a particular record. 
     
     
         13 . The computer-implemented method of  claim 9 , comprising modifying, in response to the user selection, the search query by adding, removing, or replacing one or more search terms of the search query and perform the search with the modified search query. 
     
     
         14 . The computer-implemented method of  claim 9 , comprising adding or removing content from a return record of the provided at least one record in response to the user selection. 
     
     
         15 . The computer-implemented method of  claim 9 , comprising train a machine learning model based upon data generated during the feature extraction on the records. 
     
     
         16 . At least one non-transitory computer-readable storage medium comprising instructions that, when executed, cause a system to:
 perform feature extraction on records across a plurality of datasets, features corresponding to the feature extraction comprise at least one feature associated with risk indicia and at least one feature associated with a particular data source of at least one of the plurality of datasets;   train a machine-learning model based upon data generated during the feature extraction on the records;   perform clustering on the records across the plurality of datasets to identify a set of records, each record of the set of records having risk assessment data that exceeds a baseline threshold; and   return one or more records of the set of records in response to a search query.   
     
     
         17 . The non-transitory computer-readable storage medium of  claim 16 , comprising instructions that when executed cause the system to:
 receive a user selection corresponding to the at least one record of the one or more records returned in the response to the search query; and   modify the machine-learning model in response to the user selection.   
     
     
         18 . The non-transitory computer-readable storage medium of  claim 17 , comprising instructions that when executed cause the system to:
 process, during the search query, the user selection; and   update risk assessment values for each of the at least one record for the machine-learning model, each of the risk assessment values indicative of a level of risk associated with subject matter of a particular record.   
     
     
         19 . The non-transitory computer-readable storage medium of  claim 18 , wherein updating the risk assessment value comprises adjusting feature values to increase or decrease the risk assessment value and/or adjusting feature weights associated with a particular record. 
     
     
         20 . The non-transitory computer-readable storage medium of  claim 16 , comprising instructions that when executed cause the system to identify relevant search terms for the search query using the machine learning model.