KR20200052412A - Artificial intelligence employment system and employing method of thereof - Google Patents

Abstract

According to one embodiment of the present invention, an artificial intelligence employment system capable of reducing demand of a job seeker in a recruitment process comprises: an input unit receiving data; a processing unit refining the data to generate learning data and evaluation data, wherein the evaluation data includes at least one evaluation element; and a control unit generating a required talent pattern by performing learning by artificial intelligence based on the learning data and the evaluation data and selecting a measurement result for the degree of conformity to the required talent pattern and the evaluation element determined to have influenced the measurement result by a job seeker.

Description

ARTIFICIAL INTELLIGENCE EMPLOYMENT SYSTEM AND EMPLOYING METHOD OF THEREOF}

The present invention relates to an AI hiring system and a method for employing the system, and more particularly, to an AI hiring system employing job seekers using deep learning based on text data analysis and a method for hiring the system.

Deep learning is a technique used to classify or cluster objects or data. Deep learning is a type of machine learning that trains a computer to sort things, just like the human brain sorts things out.

Deep learning performs a series of tasks to find and evaluate hidden patterns in data through the process of learning data by itself, without any special results analyzed in advance. Text deep learning is a representative field that can analyze unstructured text data among various fields of deep learning.

Recently, various fields have attempted to utilize AI technology based on text deep learning, and various methods have been proposed for the purpose of reducing the efforts of job seekers through an intelligent method in the recruitment process for evaluating job seekers.

However, in the existing methods, the recruitment evaluation criteria are only suggested by job seekers, and artificial intelligence is only learning and evaluating to what extent job seekers meet the employment evaluation criteria. As such, existing recruitment methods use artificial intelligence only for the purpose of processing vast amounts of information.

Therefore, there is a problem that job seekers must prepare new documents according to the company's job application form, or that job seekers must select job applicants' evaluation criteria in advance, thereby requiring additional efforts for job seekers and job seekers.

An object of the present invention is to provide an artificial intelligence hiring system and a method of employing the system, wherein job seekers utilize text data in documents collected from job seekers in the past without modification.

An object of the present invention is to provide a system for hiring artificial intelligence and a method for employing the system, in which job seekers generate a necessary talent pattern by artificial intelligence instead of setting employment standards.

In addition, an object of the present invention is to provide an artificial intelligence recruitment system and a recruitment method of the system to select and provide text that has a major influence on evaluation to improve the reliability of results and to enable job seekers to utilize it when recruiting.

Furthermore, the present invention provides an artificial intelligence recruitment system and a method for employing the system, which divides and combines text on the basis of diversified concrete methods of required talent suitability and job suitability to perform evaluation on documents of new job seekers It aims to do.

The technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned are obvious to those skilled in the art to which the embodiments proposed by the following description belong. Can be understood.

Artificial intelligence employing system according to an embodiment of the present invention, an input unit for receiving data; And a refinement unit to generate learning data and evaluation data, wherein the evaluation data includes at least one evaluation element; And based on the learning data and the evaluation data, learning by artificial intelligence is performed to generate a necessary talent pattern, and it is determined that job seekers have influenced the measurement result and the measurement result for the degree of matching the required talent pattern. And a control unit for selecting the evaluation element.

According to another embodiment of the present invention, a method for employing an AI employing system includes: receiving data; And purifying the data to generate learning data and evaluation data, wherein the evaluation data includes at least one evaluation element; And generating a necessary talent pattern by performing learning by artificial intelligence based on the learning data and the evaluation data. And selecting a measurement result for the degree that job seekers meet the required talent pattern and the evaluation factor determined to have influenced the measurement result.

According to the embodiments of the present invention, since job seekers utilize text data in documents collected from job seekers in the past without modification, it is possible to reduce the effort of job seekers required in the recruitment process.

According to embodiments according to the present invention, it is possible to minimize the effort of the job seekers required in the recruitment process by generating the necessary talent pattern by artificial intelligence instead of setting the hiring criteria.

In addition, according to embodiments of the present invention, by improving the reliability of the results by selecting and providing the text that has a major influence on the evaluation, it is possible for job seekers to utilize this when hiring and to effectively improve the user experience.

Furthermore, according to the embodiments of the present invention, after learning by dividing and combining texts based on a diversified concrete method of required talent mismatch and job suitability, by evaluating documents for new job seekers, more meaningful job seekers It can provide analysis and evaluation results.

1 is a block diagram showing the configuration of an AI employing system according to an embodiment of the present invention.
2 is a block diagram showing a detailed configuration of a control unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.
3 is a block diagram showing a detailed configuration of a processing unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.
4 is a block diagram showing a detailed configuration of a required human resource matching unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.
5 is a diagram showing a detailed configuration of a CNN learning module included in the artificial intelligence recruitment system according to an embodiment of the present invention.
6 is a diagram illustrating a detailed configuration of the RNN learning module included in the artificial intelligence recruitment system according to an embodiment of the present invention.
7 is a diagram showing the detailed configuration of the job suitability analysis unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.
8 is a diagram showing a detailed configuration of a correlation analysis module included in the artificial intelligence recruitment system according to an embodiment of the present invention.
9 is a view showing an operation process of the artificial intelligence recruitment system according to an embodiment of the present invention.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the technical spirit of the present invention is not limited by the embodiments described below, and other implementations that fall within the scope of the technical spirit of the present invention or other inventions that are regressive by adding, changing, and deleting other components. You can easily suggest examples.

The terminology used in the present invention has been selected, as far as possible, the general terminology widely used in connection with the current technology, but in special cases, the term is arbitrarily selected by the applicant, and in this case, its meaning is described in detail in the description of the applicable invention. Therefore, it is revealed in advance that the present invention should be grasped as the meaning of the term rather than the name of the simple term. In the following description, the word 'comprising' does not exclude the presence of elements or steps other than those listed.

1 is a block diagram showing the configuration of an AI employing system according to an embodiment of the present invention.

The artificial intelligence recruitment system 100 according to an embodiment of the present invention may support the decision making of job seekers for job search using deep learning based on text data analysis. In this case, the artificial intelligence recruitment system 100 is an engine that performs recruitment using artificial intelligence, and may be implemented in the form of a server or device provided on a job offer. Here, job seekers may be defined as job seekers, and job seekers may be defined as job seekers.

Specifically, the AI hiring system 100 may learn and evaluate unstructured text data through a deep learning analysis model. Here, the unstructured text data may be defined as original text data that has not undergone special purification or processing. For example, it may be a self-introduction item in a job seeker's document held by a job applicant.

The AI employing system 100 according to an embodiment of the present invention may include an input unit 110, a processing unit 120, and a control unit 130.

The input unit 110 may receive data.

Specifically, the data may be related to recruitment. In this case, the form of data may include any form such as text, document, image, video, and voice.

According to an embodiment, the data may be unstructured text data. For example, the text may be original data that has not been processed and refined as resumes and self-introduction letters submitted by job seekers.

The input unit 110 may load data stored in the database of the AI hiring system 100, receive data directly from a user, or receive data from an external device connected to the AI hiring system 100 by wire or wireless. have.

The processor 120 may perform data pre-processing.

Here, data pre-processing may be a task of converting data in advance into a state capable of performing learning by artificial intelligence. For example, the processing unit 120 may extract only text by removing images and shapes such as photos and lines from the resume.

According to an embodiment, the processor 120 may perform data preprocessing to be optimized for learning by artificial intelligence. For example, the processing unit 120 may extract only text from the job application and divide it into word units, and then select words necessary for learning and evaluation to determine whether to hire.

The processing unit 120 may refine the data to generate learning data and evaluation data. Here, the evaluation data may include at least one evaluation element.

Specifically, the processing unit 120 may generate learning data and evaluation data based on text data extracted from the data. The processor 120 extracts words from the text data, performs a bottom-up combining process of combining the extracted word list in word-sentence-document order, and constructs a corpus, which is a set of words that contain information about documents and sentences in the data. Then, a word vector and a sentence vector can be generated for the corpus. In this case, the processing unit 120 may generate learning data using the digitized word vector, sentence vector, and corpus as input values.

Meanwhile, in the present invention, evaluation for new job seekers may be performed based on evaluation data. The evaluation data may be a factor that is the basis for recruitment evaluation. The method for generating learning data and evaluation data will be described later in detail in the description of FIG. 3.

The controller 130 may employ job seekers by performing learning by artificial intelligence.

According to one embodiment, the control unit 130 generates a necessary talent pattern by performing learning by artificial intelligence based on the learning data and the evaluation data, and the measurement result and the measurement result for the degree to which job seekers meet the required talent pattern It is possible to select an evaluation factor that is considered to have influenced.

The controller 130 may generate necessary talent patterns by performing learning by deep learning.

In addition, the controller 130 may learn whether or not a job seeker meets a required talent and whether or not a job seeker is suitable for a hiring job by deep learning. In this case, the controller 130 may perform learning by the deep learning through the CNN learning model and the RNN learning model.

2 is a block diagram showing a detailed configuration of a control unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.

Previously, the control unit 130 described in FIG. 1 performs learning and evaluation by artificial intelligence to determine whether or not a job seeker is hired, and for this purpose, includes a required talent conformity measurement unit 210 and a job fitness analysis unit 220 Can be configured.

The required talent conformity measurement unit 210 may determine whether a job seeker meets a job search condition desired by a job seeker. For example, the recruitment conditions desired by job seekers may include at least one of a job candidate's major, work experience, qualification, gender, and education.

Specifically, the required talent conformity measurement unit 210 may perform a learning by artificial intelligence based on the evaluation data generated by the processing unit 120 to determine whether a job seeker meets the employment conditions desired by a job seeker. have.

The job suitability analysis unit 220 may determine whether a job candidate is suitable for a job in the recruitment field. For example, jobs in the recruitment field may include at least one of research and development, technical support, manufacturing, process, accounting, law, sales and marketing.

Specifically, the job suitability analysis unit 220 may perform learning by artificial intelligence based on the evaluation data generated by the processing unit 120 to determine whether job seekers are suitable for the job in the recruitment field.

In the past, job seekers had to select factors in advance for recruitment evaluation. Thereby, there was a problem in that the job burden of the job seekers and the economic cost and time increased.

In the present invention, the processing unit 120 selects a factor for the employment evaluation in advance, and based on this, the required talent conformity measurement unit 210 determines whether the job seekers meet the employment conditions. In this case, since both the processing unit 120 and the required talent conformity measurement unit 210 perform the corresponding operation by artificial intelligence, there is an advantage that the job of the job seekers in the recruitment process is reduced and cost and time are minimized.

3 is a block diagram showing a detailed configuration of a processing unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The processor 120 may preprocess data. Specifically, the processing unit 120 may preprocess unstructured text data.

To this end, the processing unit 120 includes a text information extraction module 310, a word element extraction module 320, a Corpus generation module 330, a word embedding module 340, a sentence embedding module 350, and a learning data generation module ( 360) and an evaluation data generation module 370.

The text information extraction module 310 may extract text data from data. Here, the data is unstructured text data, and may be original data that has not been processed or refined. For example, the text information extraction module 310 may remove images such as figures and photos of outlines or cells from original documents such as job applications submitted by job seekers, and select and extract only data composed of text.

On the other hand, the text data extracted by the text information extraction module 310 can be utilized by itself. However, when additional analysis is performed by the word element extraction module 320 to be described later, the quality of data for artificial intelligence learning may be improved.

The word element extraction module 320 may extract necessary sentence components from text data through morpheme analysis. Morphological analysis can be defined as an analysis process that grasps the structure of a given language sentence and discovers morphemes, which are the smallest units that have meaning through sentence division, analysis, extraction, and prototype restoration.

According to an embodiment, the sentence component may be a word. In this case, the word element extraction module 320 extracts only the necessary words, and unnecessary words are not included in the extraction target regardless of parts of speech. For example, the word element extraction module 320, from the text data, except for the investigation (ex: ~ ,, this, ~ ,, ~, ~) or definite article (ex: that, above), noun ( ex: Only a technologist, research, engineer, career) or adjective (ex: excellent, committed, responsible) can be extracted.

The Corpus generation module 330 may configure Corpus. Here, Corpus (corpus) is composed of text in a form readable by a computer, and may be configured in various forms or types according to standards such as media, time and space, and annotation steps.

Specifically, the Corpus generation module 330 performs a bottom-up combining process of combining the word list derived by the word element extraction module 320 in word-phrase-document order, and as a result, the document in the full text data and A corpus, which is a set of words containing information about sentences, can be constructed.

In this case, the Corpus generation module 330 may output the generated corpus to each of the word embedding module 340 and the sentence embedding module 350.

The word embedding module 340 may generate a word vector for the corpus.

In addition, the sentence embedding module 350 may generate a sentence vector for the corpus.

Here, embedding is a method of quantifying text into numerical information that can be expressed in a specific dimension by analyzing the corpus and learning distribution information such as location information and frequency in which each word or sentence is utilized in documents and sentences. That is, the word embedding module 340 and the sentence embedding module 350 perform statistical inference of words and sentences used in the context based on the artificial neural network learning of the corpus, and generalize the pattern.

The learning data generation module 360 may generate learning data. In this case, the learning data may be for learning required human resource conformity measurement and job suitability analysis through artificial intelligence. According to one embodiment, artificial intelligence learning may be deep learning.

Specifically, the learning data generation module 360 may generate learning data using the digitized word vector and sentence vector and corpus data as input values. In this case, word data included in the corpus is replaced with numerical data based on the results of word embedding and sentence embedding, and then learning data usable by the artificial intelligence analysis model used in the controller 130 can be derived.

The evaluation data generation module 370 may generate evaluation data.

Here, the evaluation data can provide a reference standard for deriving a result from the learned AI analysis model.

4 is a block diagram showing a detailed configuration of a required human resource matching unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The required talent conformity measurement unit 210 is included in the control unit 130, and through a learning and evaluation process using artificial intelligence, a job seeker can determine whether a job seeker meets a desired recruitment condition. To this end, the required talent conformity measurement unit 210 includes a data input module 410, a CNN learning module 420a, an RNN learning module 420b, an analysis result integration module 430, and a main text selection module 440. And it may include a required talent conformity evaluation module 450.

The data input module 410 may receive learning data and evaluation data from the processing unit 120.

The CNN learning module 420a and the RNN learning module 420b may perform a learning process of an artificial intelligence analysis model for inferring required talent matching with respect to input data. In this case, the learning process can be repeatedly performed.

Specifically, the CNN (Convolutional Neural Network) learning module 420a may grasp patterns of input data by performing segmentation and combination analysis of input words according to data analysis purposes. In addition, the Recurrent Neural Network (RNN) learning module 420b may analyze data while grasping the structure and order of words.

In this case, the CNN learning module 420a and the RNN learning module 420b may apply evaluation data to the trained AI analysis model to derive numerical results for calculating the required talent matching for job seekers. Subsequently, the numerical result may be input to the analysis result integration module 430 and used to determine whether the required talent is met.

Meanwhile, FIG. 4 illustrates a case where the required talent conformity measurement unit 210 uses two learning modules 420a and 420b, but the present invention is not limited thereto. The number of learning modules used by the required talent conformity measurement unit 210 may be variously set according to embodiments.

The analysis result integration module 430 may derive a measurement result for the required talent fit for each job seeker based on the inputted numerical result, and generate a result for the required talent fit from this.

The main text selection module 440 may select text elements that are mainly used in the learning process by the AI analysis model.

Specifically, the main text selection module 440, after the learning process of the AI analysis model is performed based on the learning data, the evaluation data of new job seekers is referred to by referring to the pattern information for required talents stored in the analysis model. When entered, the text that is judged to have a great influence on the result of matching the required talents can be selected as the main text. In this case, the main text selection module 440 may perform main text selection using artificial intelligence.

The required talent conformity evaluation module 450 may integrate the measurement result for the required talent conformity and the selected key text information and transmit it to the job applicant. For example, the information may be output to a mobile terminal (not shown) possessed by a job seeker or a display device (ex: monitor, etc.) provided on the job seeker's side. Here, the selected main text can help a job seeker to determine which word elements the AI analysis model derives the required talent matching. In this case, the job seekers may perform an evaluation for job seekers in the recruitment process based on the numerical information of the job candidates' required talent conformity and the selected main text information that is the criteria for determining AI.

Hereinafter, the CNN learning module 420a and the RNN learning module 420b included in the required talent matching unit 210 will be described in more detail with reference to FIGS. 5 and 6.

5 is a diagram showing a detailed configuration of a CNN learning module included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The CNN learning module 420a may perform a learning process of an artificial intelligence analysis model for inferring required talent matching with respect to input data. Specifically, the CNN learning module 420a may grasp a pattern of input data by performing segmentation and combination analysis of input words according to data analysis purposes. Here, the purpose of data analysis may include hiring required talent or hiring job seekers.

Referring to FIG. 5, the input layer may receive learning data generated based on word vectors and sentence vectors from the CNN model, which is one of the artificial intelligence analysis models, and transmit it to the convolutional layer of the text segmentation analysis unit.

The convolutional layer analyzes data by dividing and combining learning data into a specific number of consecutive words. In this case, the result derived from the convolutional layer may be a result of analyzing a partial word set composed of different numbers of words.

The Dropout Layer may drop some information from the analysis results. By the elimination process, the AI analysis model prevents overfitting phenomenon that is excessively adjusted to the learning data, and by adding randomness of atypical text that may occur in reality, the quality of the analysis result can be improved.

Subsequently, the data is sequentially processed by the Hidden Layer and the Output Layer, and through the Output Layer, one numerical information for constructing the required talent conformity is derived. In this case, an iterative learning process may be performed by comparing the result values given to the learning data and reflecting the degree of error in the analysis model. While the entire learning process is being performed, the CNN learning module 420a may perform repetitive learning on the text pattern of required talents while adjusting the weight value inside the analysis model based on the learning data.

6 is a diagram illustrating a detailed configuration of the RNN learning module included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The RNN learning module 420b may perform a learning process of an artificial intelligence analysis model for inferring required talent matching with respect to input data. Specifically, the RNN learning module 420b may analyze the data while grasping the structure and order of words.

The RNN learning module 420b may process numerical text data through an artificial intelligence analysis model. The CNN learning module 420a utilizes the meaning of each word and a combination thereof for learning, while the RNN learning module 420b performs learning by focusing on the association between the preceding word and the succeeding word from the structural point of view of text data.

The RNN learning module 420b grasps the composition and order of words, and may change the weight value in the AI analysis model to derive an analysis result in the process of repeatedly learning the required talent pattern.

The input layer can receive learning data. The input layer can transmit the input data to each of the text structure analysis unit and the attention layer.

The text structure analysis unit analyzes text. In this case, the text structure analysis unit may use a long short-term memory (LSTM) structure for effective text analysis.

The Attention Layer can improve the quality of the analysis results by integrating and analyzing the input learning data together with the analysis results of the text structure analysis unit.

Subsequently, through the dropout layer and output layer, additional numerical information for constructing the required talent conformity is obtained, and the process of comparing with the actual result value expressed in the learning data is the same as the process performed in the CNN learning module 420a. Through this, an iterative analysis model learning process is performed to minimize errors.

Furthermore, both models use the learning data to modify the information of the AI analysis model, and to activate nonlinearity in the analysis results of each layer in the iterative learning process to improve the quality of the analysis results ( ex: ReLU). Therefore, the AI analysis model can learn not only linear information but also nonlinear information. Analysis model based on learning data After the learning process is over, learning no longer occurs in the process of analyzing new job seekers data using evaluation data, and only deriving the result of matching the required talents through performing the analysis model is performed.

7 is a diagram showing the detailed configuration of the job suitability analysis unit included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The job suitability analysis unit 220 may be included in the control unit 130 to receive refining data and perform learning and evaluation of job suitability based on this to evaluate job suitability for job seekers. To this end, the job suitability analysis unit 220 may include a data input module 710, a Cosine Similarity measurement module 720, a correlation analysis module 730, and a job suitability evaluation module 740.

The data input module 710 may receive learning data and evaluation data from the processing unit 120.

The Cosine Similarity Measurement Module 720 performs similarity determination between words and sentences by using the input data. In this case, based on the word vector and the sentence vector generated by the word embedding module 340 and the sentence embedding module 350 described in FIG. 3, similarity between the digitized word or the sentence may be calculated.

The correlation analysis module 730 may analyze the correlation based on the measured similarity result and perform similar sentence determination. This will be described in more detail in the description of FIG. 8.

The job suitability evaluation module 740 evaluates the job fit based on the similar sentence determination result.

8 is a diagram showing a detailed configuration of a correlation analysis module included in the artificial intelligence recruitment system according to an embodiment of the present invention.

The correlation analysis module 730 may perform sentence correlation analysis.

Specifically, referring to FIG. 8, the correlation analysis module 730 includes sentences Sentence ₁ and m words (W ₂₁ , W ₂₂ , consisting of n words W ₁₁ , W ₁₂ , ... W _1n ) ... W _2m ) can analyze the similarity of sentence Sentence ₂ .

In this case, the sentence correlation analysis unit receives the sentence Sentence ₁ and the sentence Sentence ₂ , and determines the similarity between words included in the sentence. Thereafter, the correlation analysis module 730 derives a correlation analysis result between two sentences by integrating similarity information analyzed between words.

Based on the analysis result of the correlation analysis module 730 as described above, the job suitability evaluation module 740 shown in FIG. 7 above may analyze job evaluation data for each job candidate and derive a job fit result. Therefore, job seekers can judge the degree of similarity between job applicants' documents and job fields, and use them for job evaluation.

9 is a view showing an operation process of the artificial intelligence recruitment system according to an embodiment of the present invention.

The AI employing system 100 according to an embodiment of the present invention may learn and evaluate a job seeker's pattern through an AI analysis model.

The text data in the job seeker's document is purified (S901).

The processing unit 120 of the artificial intelligence recruitment system 100 may preprocess the original data by processing and purifying it. Specifically, the processing unit 120 may extract text data from original data that is unstructured text data, and extract words from the extracted text data. In this case, the processor 120 may perform a bottom-up combining process of combining the extracted word list in word-sentence-document order, and construct a corpus, which is a set of words including information about documents and sentences in the entire text data. have. Thereafter, the processor 120 may generate word vectors and sentence vectors for the corpus.

Generate learning data (S902).

The processing unit 120 of the artificial intelligence recruitment system 100 may generate learning data. Here, the learning data may be for learning required human resource conformity measurement and job suitability analysis through artificial intelligence.

Specifically, the learning data generation module 360 of the processing unit 120 may generate learning data using the digitized word vector and sentence vector and corpus data as input values. In this case, the word data included in the corpus is replaced with numerical data based on the results of word embedding and sentence embedding, and then learning data usable by the artificial intelligence analysis model used in the controller 130 can be derived.

Deep learning model learning is performed (S903).

According to an embodiment of the present disclosure, the controller 130 of the artificial intelligence recruitment system 100 may perform a learning process of a deep learning analysis model for inferring required talent matching with respect to the input data.

Specifically, the required talent matching unit 210 of the controller 130 may perform a learning process of the CNN learning model and the RNN learning model. In this case, the CNN learning module 420a of the required talent matching unit 210 may grasp the input data according to the purpose of data analysis and grasp the input data pattern. Here, the purpose of data analysis may include hiring required talent or hiring job seekers. In addition, the RNN learning module 420b of the required talent matching unit 210 grasps the composition and order of words, and in the artificial intelligence analysis model for deriving an analysis result in the process of repeatedly learning the required talent pattern. The weight value can be changed.

While the CNN learning module 420a utilizes the meaning of each word and a combination thereof for learning, the RNN learning module 420b performs learning by focusing on the association between the preceding word and the following word from the structural point of view of text data.

Evaluation data is generated (S904).

The processing unit 120 of the artificial intelligence recruitment system 100 may generate evaluation data. Specifically, the evaluation data generation module 370 of the processing unit 120 may generate evaluation data. Here, the evaluation data can provide a reference standard for deriving a result from the learned AI analysis model.

An evaluation result for each job candidate is derived (S905).

The control unit 130 of the artificial intelligence recruitment system 100 may derive a measurement result for the required talent matching for each job seeker based on the numerical result derived by the learning process of the deep learning analysis model. When the evaluation data of new job seekers is input by referring to the pattern information on required talents stored in the analysis model, the control unit 130 selects texts that are determined to have a great influence on the required talents matching result as the main texts Can be. In this case, the control unit 130 may integrate the measurement result for the required talent matching with the selected key text information and deliver it to the job applicant.

In this case, the job seekers may perform an evaluation for job seekers in the recruitment process based on the numerical information of the job candidates' required talent conformity and the selected main text information that is the criteria for determining AI.

On the other hand, in this specification, for convenience of explanation, the case of evaluating and learning job seekers using deep learning, which is one of artificial intelligence techniques, is described as an example, but the present invention is not limited thereto. Evaluation and learning for job seekers performed in the present invention can be performed by various artificial intelligence techniques.

Meanwhile, the above-described method may be written in a program executable on a computer, and may be implemented in a general-purpose digital computer that operates the program using a computer-readable medium. Further, the structure of the data used in the above-described method can be recorded on a computer-readable medium through various means. It should be understood that a recording medium that records an executable computer program or code for performing various methods of the present invention does not include temporary objects such as carrier waves or signals. The computer-readable medium may include a storage medium such as a magnetic storage medium (eg, ROM, floppy disk, hard disk, etc.) and an optical read medium (eg, CD ROM, DVD, etc.).

Although the embodiments have been mainly described above, these are merely examples, and do not limit the present invention, and those skilled in the art to which the present invention pertains are not exemplified above, without departing from the essential characteristics of the embodiments. It will be appreciated that various modifications and applications are possible. For example, each component specifically shown in the embodiment may be modified. And differences related to these modifications and applications should be construed as being included in the scope of the invention defined in the appended claims.

100: artificial intelligence recruitment system 110: input
120: processing unit 130: control unit
210: required talent conformity measurement unit 220: job suitability analysis unit
310: text information extraction module 320: word element extraction module
330: Corpus generation module 340: Word embedding module
350: sentence embedding module 360: learning data generation module
370: evaluation data generation module 410: data input module
420a: CNN learning module 420b: RNN learning module
430: Analysis result integration module 440: main text selection module
450: Required talent matching evaluation module 710: Data input module
720: Cosine Similarity Measurement Module 730: Correlation Analysis Module
740: Job Fit Evaluation Module

Claims (17)

In the AI adoption system,
An input unit that receives data;
A processing unit that refines the data to generate learning data and evaluation data, wherein the evaluation data includes at least one evaluation element; And
Based on the learning data and the evaluation data, learning by artificial intelligence is performed to generate a necessary talent pattern, and it is determined that job seekers have influenced the measurement result and the measurement result for the degree of matching the required talent pattern. Artificial intelligence recruitment system including a control unit for selecting an evaluation element. According to claim 1,
The data is
AI recruitment system, which is unstructured text data that is the original data before performing the purification process. According to claim 1,
The processing unit,
After performing a bottom-up combining process of extracting words from the data and combining the extracted word list in word-phrase-document order, and constructing a corpus, which is a word set including information about the document and the sentence in the data, AI recruitment system for generating word vectors and sentence vectors for the corpus. According to claim 3,
The processing unit,
An artificial intelligence recruitment system that generates the learning data using the digitized word vector, the sentence vector, and the corpus as input values. According to claim 1,
The control unit,
Artificial intelligence recruitment system that performs the learning by deep learning to generate the required talent pattern. According to claim 1,
The control unit,
AI recruitment system that learns whether the job seekers meet the required talents by deep learning. The method of claim 6,
The control unit,
AI recruitment system that performs learning by the deep learning through CNN learning model and RNN learning model. According to claim 1,
The control unit,
AI recruitment system that learns whether the job seekers are suitable for job opening by deep learning. In the recruitment method of the artificial intelligence recruitment system,
Receiving data;
Refining the data to generate learning data and evaluation data, wherein the evaluation data includes at least one evaluation element;
Generating a necessary talent pattern by performing learning by artificial intelligence based on the learning data and the evaluation data; And
A method of employing an artificial intelligence recruitment system, comprising the step of selecting a measurement result for the degree that job seekers meet the required talent pattern and the evaluation factor determined to have influenced the measurement result. The method of claim 9,
The data is
A method of employing an artificial intelligence recruitment system, which is unstructured text data, which is original data before performing a purification process. The method of claim 9,
After performing a bottom-up combining process of extracting words from the data and combining the extracted word list in word-phrase-document order, and constructing a corpus, which is a word set including information about the document and the sentence in the data, A method of employing an artificial intelligence recruitment system that generates word vectors and sentence vectors for the corpus. The method of claim 11,
A method of employing an artificial intelligence recruitment system that generates the learning data using the digitized word vector, the sentence vector, and the corpus as input values. The method of claim 9,
A method of employing an artificial intelligence recruitment system that generates the necessary talent pattern by performing the learning by deep learning. The method of claim 9,
A method of employing an artificial intelligence recruitment system that learns whether the job seeker meets the required talents by deep learning. The method of claim 14,
A method of employing an artificial intelligence recruitment system that performs learning by the deep learning through a CNN learning model and a RNN learning model. The method of claim 9,
A method of employing an artificial intelligence recruitment system that learns whether the job seeker is suitable for a job opening by deep learning. A computer-readable recording medium on which a program for implementing the method of any one of claims 9 to 16 is recorded.

Images