Artificial neural networks-based domain- and company-specific talent selection processes
Methods and systems are provided for adaptive recruitment computer system. In one novel aspect, the adaptive recruitment computer system generates a question bank based on a job description, selects adaptively questions from the question bank for an interview, and generates a feedback report for the candidate based on the evaluation of the candidate's answer. In one embodiment, the computer system categorizes a job requirement based on a body of knowledge (BOK) skill knowledge base, generates a question bank, selects adaptively a subset of questions from the generated question bank, wherein each question selected is based on evaluations of candidate's answers to corresponding prior questions using a recurrent neural network (RNN) model, and generates a feedback report for the candidate, wherein the feedback report using the RNN model based on evaluations of answers and a BOK candidate knowledge base, wherein the BOK candidate knowledge base receives updates from the computer system.
1. A method, comprising:
categorizing, by a computer system with one or more processors coupled with at least one memory unit, a job requirement into a set of job skills based on a body of knowledge (BOK) skill knowledge base;
obtaining a group of primary candidates through an adaptive selection procedure based on a BOK question knowledge base, wherein the adaptive selection procedure adaptively selecting a subset of questions based on evaluations of one or more answers from the candidate to corresponding prior questions using a recurrent neural network (RNN) model;
generating a candidate-criteria matrix for the group of primary candidates based on corresponding profiles of each primary candidate and a set of selection criteria, wherein each element of the candidate-criteria matrix is a pairwise comparison of each primary candidate for each selection criterion; and
ranking the group of primary candidates by performing an analytic hierarchy process (AHP) using a multi-criteria decision analysis (MCDA) based on a set of hierarchy criteria and the candidate-criteria matrix.
2. The method of claim 1 , wherein the ranking is generated by multiplying the candidate-criteria matrix with a weight vector, and wherein each element of the weight vector is a weight value of a normalized relative importance.
3. The method of claim 2 , wherein the weight vector is a principal eigenvector of a pairwise comparison matrix, and wherein each element of the pairwise comparison matrix is a factor of a pair-wise comparison of two corresponding selection factors.
4. The method of claim 3 , wherein the pairwise comparison matrix is a relative comparison matrix, and wherein each element of the pairwise comparison matrix is a relative value comparing with other elements.
5. The method of claim 3 , wherein the pairwise comparison matrix is an absolute comparison matrix, and wherein each element of the pairwise comparison matrix is a value comparing to an ideal value.
6. The method of claim 1 , wherein the subset of questions is selected from a generated question bank comprising a list of questions based on the set of job skills and the BOK question knowledge base, and the adaptive selection procedure further comprising:
generating a feedback report for the candidate, wherein the feedback report using the RNN model based on evaluations of answers from the candidate and a BOK candidate knowledge base, wherein the BOK candidate knowledge base receives updates from the computer system.
7. The method of claim 6 , wherein each job skill has a set of attributes comprising a multi-level industry taxonomy, a skill level, and cross disciplinary references.
8. The method of claim 7 , wherein each question in the BOK question knowledge base has a skill level attribute, and wherein the generated question bank includes questions of different skill levels based on the skill level of the job skill attributes.
9. The method of claim 6 , further comprising:
obtaining candidate information prior to the interview; and
generating a candidate profile and authentication information.
10. The method of claim 6 , wherein the feedback report includes a deficiency report, and wherein a training recommendation list derived from the deficiency report is included.
11. The method of claim 6 , wherein the feedback report includes a strength report, and wherein a matching opening recommendation list derived from the strength report is included.
12. An apparatus comprising:
a network interface that connects the apparatus to a communication network;
a memory; and
one or more processors coupled to one or more memory units, the one or more processors configured to
categorize a job requirement into a set of job skills based on a body of knowledge (BOK) skill knowledge base;
obtain a group of primary candidates through an adaptive selection procedure based on a BOK question knowledge base, wherein the adaptive selection procedure adaptively selecting a subset of questions based on evaluations of one or more answers from the candidate to corresponding prior questions using a recurrent neural network (RNN) model;
generate a candidate-criteria matrix for the group of primary candidates based on corresponding profiles of each primary candidate and a set of selection criteria, wherein each element of the candidate-criteria matrix is a pairwise comparison of each primary candidate for each selection criterion; and
rank the group of primary candidates by performing an analytic hierarchy process (AHP) using a multi-criteria decision analysis (MCDA) based on a set of hierarchy criteria and the candidate-criteria matrix.
13. The apparatus of claim 12 , wherein the ranking is generated by multiplying the candidate-criteria matrix with a weight vector, and wherein each element of the weight vector is a weight value of a normalized relative importance.
14. The apparatus of claim 13 , wherein the weight vector is a principal eigenvector of a pairwise comparison matrix, and wherein each element of the pairwise comparison matrix is a factor of a pair-wise comparison of two corresponding selection factors.
15. The apparatus of claim 13 , wherein the pairwise comparison matrix is a relative comparison matrix, and wherein each element of the pairwise comparison matrix is a relative value comparing with other elements.
16. The apparatus of claim 15 , wherein the pairwise comparison matrix is an absolute comparison matrix, and wherein each element of the pairwise comparison matrix is a value comparing to an ideal value.
17. The apparatus of claim 12 , wherein each job skill has a set of attributes comprising a multi-level industry taxonomy, a skill level, and cross disciplinary references.
18. The apparatus of claim 12 , wherein each question in the BOK question knowledge base has a skill level attribute, and wherein the generated question bank includes questions of different skill levels based on the skill level of the job skill attributes.
19. The apparatus of claim 12 , wherein the processor is further configured to: obtain candidate information prior to the interview; and generate a candidate profile and authentication information.
20. The apparatus of claim 19 , wherein the candidate profile is generated from the candidate information using the RNN model based on the BOK candidate knowledge base.