Back to results
US 10,242,345 B2Granted
Automatic interview question recommendation and analysis
AssigneeHireVue, Inc.
InventorsBenjamin Taylor, Loren Larsen
Life of this patent
Estimating the term…
Abstract
Described herein are methods and systems for interview question or prompt recommendation and analysis to improve the quality and efficacy of subsequent evaluation campaigns by combining data sets are described herein. In one method, processing logic selects a first prompt from a first data set of a first candidate evaluation campaign and a second prompt from a second data set of a second candidate evaluation campaign. The processing logic determines whether a degree of similarity between the first prompt and the second prompt exceeds a threshold and combines data from the first data set with data from the second data set to create a combined data set associated with the first prompt and with the second prompt based on the determination.
Claims18
A method comprising:
- receiving, by a candidate evaluation system executed by a processing device, a request to create a current candidate evaluation campaign for a position sector;
- selecting, by the candidate evaluation system, a first prompt from a first historical evaluation campaign associated with the position sector, the first historical evaluation campaign comprising a first data set stored in a database in a data store of the candidate evaluation system;
- selecting, by the candidate evaluation system, a second prompt from a second historical evaluation campaign associated with the position sector, the second historical evaluation campaign comprising a second data set stored in the database;
- accessing, within memory of the candidate evaluation system, a synonym lookup table represented as a sparse binary matrix that minimizes a memory footprint of the synonym lookup table;
- identifying, by the candidate evaluation system, at least one synonym listed in the synonym lookup table that is present in the first prompt and the second prompt;
- counting, by the candidate evaluation system, a number of word-based edits to transform the first prompt into the second prompt;
- discounting, by the candidate evaluation system, the number of word-based edits in view of a number of synonyms of the at least one synonym, to generate a discounted number of word-based edits;
- calculating, by the candidate evaluation system, a degree of similarity between the first prompt and the second prompt based on the discounted number of word-based edits;
- determining, by the candidate evaluation system, whether the degree of similarity between the first prompt and the second prompt exceeds a threshold;
- combining, by the candidate evaluation system in response to a determination that the degree of similarity exceeds the threshold, data from the first data set with data from the second data set to create a related cluster of prompts associated with the first prompt and with the second prompt;
- analyzing, by the candidate evaluation system, the related cluster to assess a first correlation between the first prompt and an evaluation result and a second correlation between the second prompt and the evaluation result;
- ranking, by the candidate evaluation system, the first prompt and the second prompt to obtain a highest ranked prompt according to the first correlation and the second correlation;
- and
- selecting, by the candidate evaluation system, the highest ranked prompt as a prompt for the current candidate evaluation campaign.
11 more elements
A computing system comprising:
- a data storage device;
- and
- a processing device, coupled to the data storage device, to execute a candidate evaluation system to:
- receive a request to create a current candidate evaluation campaign for a position sector;
- select a first prompt from a first historical evaluation campaign associated with the position sector, the first historical evaluation campaign comprising a first data set stored in a database of the data storage device;
- select a second prompt from a second historical evaluation campaign associated with the position sector, the second historical evaluation campaign comprising a second data set stored in the database;
- access, within memory, a synonym lookup table represented as a sparse binary matrix that minimizes a memory footprint of the synonym lookup table;
- identify at least one synonym listed in the synonym lookup table that is present in the first prompt and the second prompt;
- count a number of word-based edits to transform the first prompt into the second prompt;
- discount the number of word-based edits in view of a number of synonyms of the at least one synonym, to generate a discounted number of word-based edits;
- calculate a degree of similarity between the first prompt and the second prompt based on the discounted number of word-based edits;
- determine whether the degree of similarity between the first prompt and the second prompt exceeds a threshold;
- combine, in response to a determination that the degree of similarity exceeds the threshold, data from the first data set with data from the second data set to create a related cluster of prompts associated with the first prompt and with the second prompt;
- analyze the related cluster to assess a first correlation between the first prompt and an evaluation result and a second correlation between the second prompt and the evaluation result;
- rank the first prompt and the second prompt to obtain a highest ranked prompt according to the first correlation and the second correlation;
- and
- select the highest ranked prompt as a prompt for the current candidate evaluation campaign.
14 more elements
A non-transitory storage medium storing instructions that when executed by a processing device cause the processing device to perform operations comprising:
- receiving, by the processing device, a request to create a current candidate evaluation campaign for a position sector;
- selecting, by the processing device, a first prompt from a first historical evaluation campaign associated with the position sector, the first historical evaluation campaign comprising a first data set stored in a database in a data store;
- selecting, by the processing device, a second prompt from a second historical evaluation campaign associated with the position sector, the second historical evaluation campaign comprising a second data set stored in the database;
- accessing, within memory, a synonym lookup table represented as a sparse binary matrix that minimizes a memory footprint of the synonym lookup table;
- identifying, by the processing device, at least one synonym listed in the synonym lookup table that is present in the first prompt and the second prompt;
- counting, by the processing device, a number of word-based edits to transform the first prompt into the second prompt;
- discounting, by the processing device, the number of word-based edits in view of a number of synonyms of the at least one synonym, to generate a discounted number of word-based edits;
- calculating, by the processing device, a degree of similarity between the first prompt and the second prompt based on the discounted number of word-based edits;
- determining, by the processing device, whether the degree of similarity between the first prompt and the second prompt exceeds a threshold;
- combining, by the processing device in response to a determination that the degree of similarity exceeds the threshold, data from the first data set with data from the second data set to create a related cluster of prompts associated with the first prompt and with the second prompt;
- analyzing, by the processing device, the related cluster to assess a first correlation between the first prompt and an evaluation result and a second correlation between the second prompt and the evaluation result;
- ranking, by the processing device, the first prompt and the second prompt to obtain a highest ranked prompt according to the first correlation and the second correlation;
- and
- selecting, by the processing device, the highest ranked prompt as a prompt for the current candidate evaluation campaign.
11 more elements
Dependencies are inferred from the claim text.