KR102675644B1 - System and method of recommending custmized course subjects based on deep learning - Google Patents

Abstract

The present invention relates to a technical idea for recommending user-customized courses based on deep learning. Based on deep learning, user behavior information including information related to the user's job, qualifications, language skills, etc., and courses and majors that the user can take. By learning course attribute information, including information related to such things, and learning the learned user deactivation information and course attribute information together, user-customized courses are learned by learning interaction characteristics using each user's participation in individual courses as a target variable. It relates to a technology for recommending courses. According to an embodiment of the present invention, a deep learning-based user-customized course recommendation system includes a data collection unit that collects a plurality of user behavior data and a plurality of course attribute data, and a plurality of the collected data. At least one user behavior characteristic is extracted by learning user behavior data, at least one course attribute characteristic is extracted by learning the plurality of collected course attribute data, and the extracted at least one user behavior characteristic and the extracted A deep learning model processing unit that learns at least one course attribute characteristic together to extract at least one interaction characteristic related to whether a user takes a plurality of courses, and at least one interaction characteristic based on the extracted at least one interaction characteristic It may include a recommendation section recommending courses to be taken.

Description

Deep learning-based user-customized course recommendation system and method {SYSTEM AND METHOD OF RECOMMENDING CUSTMIZED COURSE SUBJECTS BASED ON DEEP LEARNING}

The present invention relates to a technical idea for recommending user-customized courses based on deep learning. Based on deep learning, user behavior information including information related to the user's job, qualifications, language skills, etc., and courses and majors that the user can take. By learning course attribute information, including information related to such things, and learning the learned user deactivation information and course attribute information together, user-customized courses are learned by learning interaction characteristics using each user's participation in individual courses as a target variable. It is about technology for recommending subjects.

Due to the advancement of information technology and the popularization of mobile devices, various types of structured and unstructured data have increased explosively, making it possible to produce and collect a lot of information.

However, due to the rapidly increasing amount of information, users are faced with the problem of information overload, which makes it difficult to make decisions and takes a lot of time to select the necessary product or service.

Accordingly, the importance of personalized recommendation services is emerging, and recommendation systems are being developed in areas such as movies, news, and products to provide various user-customized services.

Recently, the demand for online education platforms is increasing due to COVID-19, which is spreading globally.

However, there is a problem in that users have difficulty selecting the right subject suitable for them among numerous subjects due to differences in knowledge structure.

Therefore, a personalized service that provides customized subjects to users is essential as it plays an important role in increasing the user's learning efficiency.

Currently, most online education platforms provide recommendation services by applying traditional collaborative filtering (CF) algorithms.

However, because the analysis logic of these existing recommendation algorithms is simple, it is difficult to build a model using various user information.

In addition, recommendation services that apply a collaborative filtering algorithm have a "Cold Start" problem in which suitable items cannot be recommended to new users due to insufficient past information or data, and recommendations cannot be provided until the user's preferences are reflected. There is also the non-existent "First Start" problem.

As user preference data continues to increase, there is also a scalability problem that degrades the computational performance of the algorithm. Recently, research has been conducted to complement problems such as data scarcity and scalability.

Recently, research on applying deep learning techniques has been actively conducted in various fields to improve the limitations of existing personalized recommendation service research.

However, in the current research on personalized recommendation services in the field of online education, there are not many studies that apply deep learning techniques to recommend subjects suitable for the user's knowledge structure.

Korean Patent No. 10-2210504, “Customized education program system using big data-based AI education platform” Korean Patent No. 10-2063183, “Learning content recommendation device and method” Korean Patent No. 10-2070783, “A clothing processing device that recommends customized courses and options by learning the patterns of courses and options performed by the user through artificial intelligence, a control method thereof, and an online system including the clothing processing device " Korea Patent Publication No. 10-2020-0103146, “Artificial intelligence learning diagnosis system using ontology-based knowledge map”

The purpose of the present invention is to reduce information overload on curriculum resources and solve the data scarcity problem based on a deep learning algorithm that models user behavior information and course attribute information.

The purpose of the present invention is to solve the problems of high-dimensional data scarcity and scalability, which are problems with existing collaborative filtering algorithms, by simultaneously considering user behavior information and course attribute information to solve the information overload problem in the field of online education.

The present invention utilizes a multi-layer perceptron to construct a high-level nonlinear deep learning algorithm, reduces dimensionality in sparse data of an online education platform, and utilizes the interaction between user behavior information and course attribute information. The purpose is to sufficiently predict the user's interest preferences by precisely extracting characteristic information.

The present invention uses latent features learned in each of the user behavior network (UBN) layer, course attribution network (CAN) layer, and deep neural network (DNN) layer. The purpose is to perform deep learning learning in a way that minimizes pointwise error by using each user's attendance of individual courses as a target variable.

The purpose of the present invention is to effectively recommend courses suitable for the user's knowledge structure when applied to an online education platform.

The present invention addresses the cold start problem in which suitable courses cannot be recommended to new users due to insufficient past information or data, and the first start problem in which recommendations cannot be provided until the user's preferences are reflected. The purpose is to solve problems simultaneously.

A deep learning-based user-customized course recommendation system according to an embodiment of the present invention includes a data collection unit that collects a plurality of user behavior data and a plurality of course attribute data, and at least one device by learning the plurality of collected user behavior data. Extract user behavior characteristics, extract at least one course attribute characteristic by learning the plurality of collected course attribute data, and learn the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together. It includes a deep learning model processing unit that extracts at least one interaction characteristic related to whether the user has taken a plurality of courses, and a recommendation unit that recommends at least one course based on the extracted at least one interaction characteristic. can do.

The data collection unit may collect a plurality of user behavior data including a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data.

The data collection unit may collect the plurality of course attribute data including a plurality of course data and a plurality of major data.

The deep learning model processing unit first embeds a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data included in the plurality of user behavior data. converting the converted first embedding vectors into vectors, inputting the converted first embedding vectors into a user behavior network (UBN) layer, and learning the input first embedding vectors to extract the at least one user behavior characteristic. there is.

The deep learning model processing unit converts a plurality of course data and a plurality of major data included in the plurality of course attribute data into second embedding vectors, and converts the converted second embedding vectors into course attribute data. The at least one course attribute feature can be extracted by inputting it to a network (course attribution network, CAN) layer and learning the input second embedding vectors.

The deep learning model processing unit inputs the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic to a deep neural network (DNN) layer, and the input at least one user behavior characteristic And the at least one interaction characteristic can be extracted by learning the at least one input course attribute characteristic together.

The deep learning model processing unit uses the user behavior network (UBN) layer and the course attribution network (CAN) layer as a parallel sub-network, and the parallel sub-network and By connecting the deep neural network (DNN) layer, the deep neural network (DNN) layer can be used as a prediction layer.

In learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together, the deep learning model processing unit controls the negative sampling rate of the point wise function to determine the at least one In extracting interaction characteristics, the point wise error of the deep learning model can be reduced.

The recommender may recommend the at least one course based on implicit feedback that indirectly reflects the extracted at least one interaction characteristic in the recommendation of the at least one course.

The recommender may rank the at least one course, list the ranked at least one course, and provide a recommendation list including the listed at least one course.

The recommendation unit may randomly select a course with which the user has not interacted among the at least one course and rank the at least one course.

A deep learning-based user-customized course recommendation method according to an embodiment of the present invention includes collecting a plurality of user behavior data and a plurality of course attribute data in a data collection unit, and collecting the plurality of collected user behavior data and a plurality of course attribute data in a deep learning model processing unit. Extracting at least one user behavior characteristic by learning user behavior data, extracting, in the deep learning model processing unit, at least one course attribute characteristic by learning the plurality of collected course attribute data, the deep learning model In a processing unit, learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together to extract at least one interaction characteristic related to whether the user takes a plurality of courses, and recommending The method may include recommending at least one course based on the extracted at least one interaction characteristic.

The step of collecting a plurality of user behavior data and a plurality of course attribute data includes a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data. It may include collecting the plurality of user behavior data and collecting the plurality of course attribute data including a plurality of course data and a plurality of major data.

The step of extracting at least one user behavior characteristic by learning the plurality of collected user behavior data includes a plurality of user data, a plurality of job data, and a plurality of user data included in the plurality of user behavior data. Convert certificate data and a plurality of language data into first embedding vectors, input the converted first embedding vectors to a user behavior network (UBN) layer, and input the input first embedding vectors to a user behavior network (UBN) layer. 1 It may include extracting the at least one user behavior characteristic by learning embedding vectors.

The step of extracting at least one course attribute characteristic by learning the plurality of collected course attribute data includes a plurality of course data and a plurality of major data included in the plurality of course attribute data. Converting the converted second embedding vectors into embedding vectors, inputting the converted second embedding vectors to a course attribution network (CAN) layer, and learning the input second embedding vectors to extract the at least one course attribute feature. May include steps.

The step of extracting at least one interaction characteristic related to whether a user takes a plurality of courses by learning together the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic, Input at least one user behavior characteristic and the extracted at least one course attribute characteristic into a deep neural network (DNN) layer, and input the at least one user behavior characteristic and the at least one input course attribute It may include extracting the at least one interaction characteristic by learning the characteristics together.

The present invention can reduce information overload on curriculum resources and solve data scarcity problems based on a deep learning algorithm that models user behavior information and course attribute information.

The present invention can solve the problems of high-dimensional data scarcity and scalability, which are problems with existing collaborative filtering algorithms, by simultaneously considering user behavior information and course attribute information to solve the information overload problem in the field of online education.

The present invention utilizes a multi-layer perceptron to construct a high-level nonlinear deep learning algorithm, reduces dimensionality in sparse data of an online education platform, and utilizes the interaction between user behavior information and course attribute information. By precisely extracting characteristic information, the user's interest preferences can be sufficiently predicted.

The present invention uses latent features learned in each of the user behavior network (UBN) layer, course attribution network (CAN) layer, and deep neural network (DNN) layer. Deep learning learning can be performed by minimizing pointwise error by using individual users' attendance at individual courses as the target variable.

The present invention can be applied to an online education platform to effectively recommend courses suitable for the user's knowledge structure.

The present invention addresses the cold start problem in which suitable courses cannot be recommended to new users due to insufficient past information or data, and the first start problem in which recommendations cannot be provided until the user's preferences are reflected. Problems can be solved simultaneously.

1 to 3 are diagrams illustrating a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.
Figure 4 is a diagram illustrating a deep learning-based user-customized course recommendation method according to an embodiment of the present invention.
Figures 5A to 7B are diagrams illustrating evaluation results for recommendations made by a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.

Hereinafter, various embodiments of this document are described with reference to the attached drawings.

The examples and terms used herein are not intended to limit the technology described in this document to a specific embodiment, and should be understood to include various changes, equivalents, and/or substitutes for the embodiments.

In the following description of various embodiments, if a detailed description of a related known function or configuration is judged to unnecessarily obscure the gist of the invention, the detailed description will be omitted.

In addition, the terms described below are terms defined in consideration of functions in various embodiments, and may vary depending on the intention or custom of the user or operator. Therefore, the definition should be made based on the contents throughout this specification.

In connection with the description of the drawings, similar reference numbers may be used for similar components.

Singular expressions may include plural expressions, unless the context clearly indicates otherwise.

In this document, expressions such as “A or B” or “at least one of A and/or B” may include all possible combinations of the items listed together.

Expressions such as “first,” “second,” “first,” or “second,” can modify the corresponding components regardless of order or importance and are used to distinguish one component from another. It is only used and does not limit the corresponding components.

When a component (e.g. a first) component is said to be "connected (functionally or communicatively)" or "connected" to another (e.g. a second) component, it means that the component is connected to the other component. It may be connected directly to a component or may be connected through another component (e.g., a third component).

In this specification, “configured to” means “suitable for,” “having the ability to,” or “changed to,” depending on the situation, for example, in terms of hardware or software. ," can be used interchangeably with "made to," "capable of," or "designed to."

In some contexts, the expression “a device configured to” may mean that the device is “capable of” working with other devices or components.

For example, the phrase "processor configured (or set) to perform A, B, and C" refers to a processor dedicated to performing the operations (e.g., an embedded processor), or by executing one or more software programs stored on a memory device. , may refer to a general-purpose processor (e.g., CPU or application processor) capable of performing the corresponding operations.

Additionally, the term 'or' means 'inclusive or' rather than 'exclusive or'.

That is, unless otherwise stated or clear from the context, the expression 'x uses a or b' means any of the natural inclusive permutations.

Terms such as '..unit' and '..unit' used hereinafter refer to a unit that processes at least one function or operation, and may be implemented as hardware, software, or a combination of hardware and software.

1 to 3 are diagrams illustrating a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.

Figure 1 illustrates the components of a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.

Referring to FIG. 1, the deep learning-based user customized course recommendation system 100 according to an embodiment of the present invention includes a data collection unit 110, a deep learning model processing unit 120, and a recommendation unit 130. You can.

According to one embodiment of the present invention, the data collection unit 110 may collect a plurality of user behavior data and a plurality of course attribute data.

Specifically, the data collection unit 110 collects a plurality of user behavior data including a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data. It can be collected.

Additionally, the data collection unit 110 may collect a plurality of course attribute data including a plurality of course data and a plurality of major data.

According to one embodiment of the present invention, the deep learning model processing unit 120 may extract at least one user behavior characteristic by learning a plurality of collected user behavior data.

Additionally, the deep learning model processing unit 120 may extract at least one course attribute characteristic by learning a plurality of collected course attribute data.

In addition, the deep learning model processing unit 120 learns at least one extracted user behavior characteristic and at least one extracted course attribute characteristic together to determine at least one interaction characteristic related to whether the user takes a plurality of courses. It can be extracted.

For example, the deep learning model processing unit 120 uses the user behavior network (UBN) layer and the course attribution network (CAN) layer as a parallel sub-network, and the parallel sub-network By connecting a network and a deep neural network (DNN) layer, the deep neural network (DNN) layer can be used as a prediction layer.

According to an embodiment of the present invention, the deep learning model processing unit 120 includes a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of user data included in the plurality of user behavior data. Convert language data into first embedding vectors, input the converted first embedding vectors into a user behavior network (UBN) layer, and learn the input first embedding vectors to create at least one user Behavioral characteristics can be extracted.

For example, the deep learning model processing unit 120 converts a plurality of course data and a plurality of major data included in the plurality of course attribute data into second embedding vectors, and uses the converted second embedding vector At least one course attribute feature can be extracted by inputting them into a course attribution network (CAN) layer and learning the input second embedding vectors.

According to one embodiment of the present invention, the deep learning model processing unit 120 inputs at least one extracted user behavior characteristic and at least one extracted course attribute characteristic to a deep neural network (DNN) layer, and inputs At least one interaction characteristic may be extracted by learning together at least one input user behavior characteristic and at least one input course attribute characteristic.

That is, the deep learning model processing unit 120 may extract at least one interaction characteristic by learning at least one user behavior characteristic and at least one course attribute characteristic in combination.

According to one embodiment of the present invention, the deep learning model processing unit 120 uses negative sampling of a point wise function when learning at least one extracted user behavior characteristic and at least one extracted course attribute characteristic together. In extracting at least one interaction characteristic by controlling the ratio, the point wise error of the deep learning model can be reduced.

According to one embodiment of the present invention, the recommender 130 may recommend at least one course based on at least one interaction characteristic.

For example, the recommender 130 may recommend at least one course based on implicit feedback that indirectly reflects at least one extracted interaction characteristic in the recommendation of the at least one course.

According to one embodiment of the present invention, the recommendation unit 130 ranks at least one course, lists the ranked at least one course, and creates a recommendation list including the listed at least one course. can be provided.

For example, the recommendation unit 130 may rank at least one course by randomly selecting a course with which the user has not interacted among at least one course.

Here, courses that have not been interacted with may be related to courses that the user has not taken.

Therefore, the present invention can reduce information overload on curriculum resources and solve data scarcity problems based on a deep learning algorithm that models user behavior information and course attribute information.

In addition, the present invention can solve the problems of high-dimensional data scarcity and scalability, which are problems with existing collaborative filtering algorithms, by simultaneously considering user behavior information and course attribute information to solve the information overload problem in the online education field.

Figure 2 illustrates the framework of a deep learning model used by a deep learning-based user customized course recommendation system according to an embodiment of the present invention, and shows the framework of a deep learning model according to an embodiment of the present invention ( 200) may be related to the deep learning model processing unit described in FIG. 1.

Referring to Figure 2, the framework 200 of a deep learning model according to an embodiment of the present invention includes an input network layer 210 into which embedding vectors are input, a user behavior network layer 220 which extracts user behavior characteristics, It includes a course attribute network layer 230 and a deep neural network layer 240. For example, a deep learning model may be referred to as a DECOR (deep learning-based approach to course recommender) model.

According to one embodiment of the present invention, the input network layer 210 is a hidden layer, and user behavior data related to the user, occupation, certification, and language is converted into an embedding vector and input into the input network layer 210, Course attribute data related to the course and major may be converted into an embedding vector and input to the input network layer 210.

According to one embodiment of the present invention, data input to the input network layer 210 may be transmitted to the user behavior network layer 220 and the course attribute network layer 230.

For example, the user behavior network layer 220 may be referred to as a user behavior network (UBN) layer, and the course attribute network layer 230 may be referred to as a course attribute network (CAN) layer.

For example, the user behavior network layer 220 may use a feed-forward neural network (FFNN) to extract user behavior characteristics.

The first embedding vector input to the user behavior network layer 220 is a vector that has low-dimensional dense real values as it includes highly sparse and high-dimensional categorical and continuous values.

The first embedding vector may be fed to the hidden layer of the neural network in the forward pass.

The user behavior network layer 220 according to an embodiment of the present invention can provide user behavior characteristics as output as shown in Equation 1 below.

[Equation 1]

In Equation 1, e may represent the inclusion of user, task, certificate, and language, respectively, and a ⁽⁰⁾ may be associated with the process fed into the deep neural network layer 240.

The process supplied to the deep neural network layer 240 can be defined through Equation 2 below.

[Equation 2]

In Equation 2, l represents the depth of the layer represents the activation function, , and may represent the output, model weight, and bias of the lth layer, respectively.

For example, the course attribute network layer 230 may use a feed-forward neural network (FFNN) to extract user behavior characteristics.

The second embedding vector input to the course attribute network layer 230 is a vector that has low-dimensional dense real values as it includes highly sparse and high-dimensional categorical and continuous values.

The second embedding vector may be fed to the hidden layer of the neural network in the forward pass.

The course attribute network layer 230 according to an embodiment of the present invention can provide course attribute characteristics as shown in Equation 3 below as output.

[Equation 3]

In Equation 3, e may represent the inclusion of courses and majors, respectively, and a ⁽⁰⁾ may be related to the process fed into the deep neural network layer 240.

The process supplied to the deep neural network layer 240 can be defined through Equation 2 above.

According to one embodiment of the present invention, the deep neural network layer 240 may integrate the user behavior network layer 220 and the course attribute network layer 230.

The deep neural network layer 240 can extract the user behavior characteristics of the user behavior network layer 220 and the course attribute characteristics of the course attribute network layer 230 in an end-to-end process.

The deep neural network layer 240 can jointly learn all parameters and network parameters for a deep learning model based on Equation 4 below.

[Equation 4]

In equation 4, can represent a binary class label, may represent a sigmoid function, N may represent the output of the user behavior network layer 220, and CAN may represent the output of the course attribute network layer 230.

Therefore, the present invention utilizes a multi-layer perceptron to construct a high-level nonlinear deep learning algorithm, reduces the dimensionality in sparse data of an online education platform, and generates user behavior information and course attribute information. By sophisticatedly extracting interaction characteristic information, the user's interest preferences can be sufficiently predicted.

Figure 3 explains in more detail a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.

Referring to FIG. 3, in the deep learning-based user-customized course recommendation system 300 according to an embodiment of the present invention, the data pre-processing unit 320 collects a plurality of user behavior data from the first database 310 and , a plurality of course attribute data are collected from the second database 311.

For example, the data pre-processing unit 320 classifies and defines data related to user information, job information, qualification information, and language ability information in relation to a plurality of user behavior data, and course information and information in relation to a plurality of course attribute data. Classify and define data related to major information.

According to one embodiment of the present invention, the data mapping unit 321 maps user information, job information, qualification information, and language ability information to user behavior data, and maps course information and major information to course attribute data.

For example, the training data set generator 322 generates mapped user behavior data and course attribute data as a training data set.

According to one embodiment of the present invention, the data collection unit described in FIG. 1 may include a data preprocessor 320, a data mapping unit 321, and a training data set generator 322.

For example, the training data set generated in the training data set generator 322 is transmitted to the deep learning model processing unit 330, and the deep learning model processing unit 330 generates a plurality of user behavior data and a plurality of user behavior data included in the training data set. By learning course attribute data, user behavior characteristics and course attribute characteristics are extracted, and user behavior characteristics and course attribute characteristics are learned together to extract interaction characteristics related to whether the user has taken the course.

According to one embodiment of the present invention, the recommender 340 recommends courses based on interaction characteristics, and may recommend courses using a ranked list of recommended courses.

For example, the test data set providing unit 350 provides a test data set, and the test data set may be defined as shown in Table 1 below.

For example, the test dataset consists of user behavior data and course attribute data collected from an online recruitment website and an online certificate website, respectively, and the data set may be structured as shown in Table 1 below.

[Table 1]

Referring to Table 1, the dataset was collected from 891 users across 11,099 courses with a sparsity of 99.73%, and the dataset contains 26,115 interactions for five criteria: occupation, certificate, language, and major, in addition to the interactions. It is included.

Additionally, there are 25 job information, 15 language information, 235 certification information, and 45 major information, and each interaction can indicate whether the user has taken a course or not.

Additionally, the evaluation meter providing unit 351 provides an evaluation metric for testing the test dataset and recommendation results.

For example, evaluation metrics may relate to hit ratio (HR) and normalized discounted cumulative gain (NDCG).

According to one embodiment of the present invention, the evaluation result providing unit 360 may provide evaluation results using a test data set and an evaluation meter for the courses recommended by the recommendation unit 340.

Figure 4 is a diagram illustrating a deep learning-based user-customized course recommendation method according to an embodiment of the present invention.

Referring to FIG. 4, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention collects a plurality of user behavior data and a plurality of course attribute data in step 401.

In other words, the deep learning-based user-customized course recommendation method involves multiple user behaviors including multiple user data, multiple job data, multiple certificate data, and multiple language data. Data may be collected, and multiple course attribute data including multiple course data and multiple major data may be collected.

In step 402, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention extracts user behavior characteristics by learning user behavior data, extracts course attribute characteristics by learning course attribute data, and extracts course attribute characteristics. Interaction characteristics can be extracted by learning the user behavior characteristics and course attribute characteristics together.

In other words, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention includes multiple user data, multiple job data, and multiple certificates included in the multiple user behavior data. Convert data and a plurality of language data into first embedding vectors, input the converted first embedding vectors into a user behavior network (UBN) layer, and learn the input first embedding vectors. At least one user behavior characteristic can be extracted.

In addition, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention uses a plurality of course data and a plurality of major data included in the plurality of course attribute data as second embedding vectors. After converting, the converted second embedding vectors are input to a course attribution network (CAN) layer, and the input second embedding vectors are learned to extract at least one course attribute feature.

In addition, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention inputs at least one user behavior characteristic and at least one course attribute characteristic into a deep neural network (DNN) layer, and inputs At least one interaction characteristic may be extracted by learning together at least one input user behavior characteristic and at least one input course attribute characteristic.

In step 403, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention may recommend user-customized courses based on the interaction characteristics extracted in step 402.

That is, the deep learning-based user-customized course recommendation method according to an embodiment of the present invention ranks at least one course recommended based on interaction characteristics, lists the at least one ranked course, and A recommendation list containing at least one listed course can be provided.

Therefore, the present invention can be applied to an online education platform to effectively recommend courses suitable for the user's knowledge structure.

In addition, the present invention addresses the cold start problem in which suitable courses cannot be recommended to new users due to insufficient past information or data, and the first start problem in which recommendations cannot be provided until the user's preferences are reflected. start) The purpose is to solve problems simultaneously.

Figures 5A to 7B are diagrams illustrating evaluation results for recommendations made by a deep learning-based user-customized course recommendation system according to an embodiment of the present invention.

5A and 5B show the number of user behavior characteristics, the number of course attribute characteristics, and the number of interaction characteristics corresponding to predictive factors in the deep learning-based user-customized course recommendation system according to an embodiment of the present invention. The results of the change are compared and explained with a course recommendation system based on a recommendation algorithm according to the prior art.

The graph 500 of FIG. 5A may represent a hit ratio (HR), and the graph 510 of FIG. 5B may represent a normalized discounted cumulative gain (NDCG).

The graph 500 compares and explains the first prior art (501), the second prior art (502), and the third prior art (503) with the present invention (504), and the graph (510) illustrates the first prior art (501), the second prior art (502), and the third prior art (503). 511), the second prior art 512, and the third prior art 513 and the present invention 514 will be compared and explained.

The first prior art 501 and the first prior art 511 may represent an item-based collaborative filtering (itemKNN) recommendation algorithm, and the second prior art 502 and the second prior art 512 may represent an MF (matrix). factorization) recommendation algorithm, and the third prior art 503 and the third prior art 513 may represent a Neural collaborative filtering (NCF) recommendation algorithm.

According to the graph 500 and graph 510, as the predictive factor increases, the representation of the model improves, but at the same time, more noise is involved and the performance of the model deteriorates due to overfitting.

As a result of measuring model performance according to changes in predictive factors, it can be seen that the best performance is achieved when the value is 32.

Figures 6a and 6b illustrate the results of changes in the number of recommendations in a deep learning-based user-customized course recommendation system according to an embodiment of the present invention, compared with a course recommendation system based on a recommendation algorithm according to the prior art. Here, the change in the number of recommendations is related to the size of the recommendation list. As the number of recommendations increases, the size of the recommendation list increases, and as the number of recommendations decreases, the size of the recommendation list decreases.

The graph 600 of FIG. 6A may represent a hit ratio (HR), and the graph 610 of FIG. 6B may represent a normalized discounted cumulative gain (NDCG).

The graph 600 compares and explains the first prior art 601, the second prior art 602, and the third prior art 603 with the present invention 604, and the graph 610 illustrates the first prior art (601). 611), the second prior art 612, and the third prior art 613 and the present invention 614 will be compared and explained.

The first prior art 601 and the first prior art 611 may represent an item-based collaborative filtering (itemKNN) recommendation algorithm, and the second prior art 602 and the second prior art 612 may represent an MF (matrix). factorization) recommendation algorithm, and the third prior art 603 and the third prior art 613 may represent a Neural collaborative filtering (NCF) recommendation algorithm.

According to the graph 600 and graph 610, it can be seen that as the number of recommendation lists corresponding to the number of recommendations increases, the present invention provides improved performance compared to the prior art.

The traditional neighborhood-based prior art 1 has lower performance than the model-based approach, which confirms the need for a model-based approach rather than a memory-based approach in personalized recommendation.

Figures 7a and 7b illustrate the results of changes in negative sampling in a deep learning-based user-customized course recommendation system according to an embodiment of the present invention, compared with a course recommendation system based on a recommendation algorithm according to the prior art.

The graph 700 of FIG. 7A may represent a hit ratio (HR), and the graph 710 of FIG. 7B may represent a normalized discounted cumulative gain (NDCG).

The graph 700 compares and explains the first prior art 701 and the second prior art 702 and the present invention 703, and the graph 710 illustrates the first prior art 711 and the second prior art ( 712) and the present invention 713 will be compared and explained.

The first prior art 701 and the first prior art 711 may represent a matrix factorization (MF) recommendation algorithm, and the second prior art 702 and the second prior art 712 may represent a neural collaborative filtering (NCF) recommendation algorithm. It can represent a recommendation algorithm.

According to the graph 700 and the graph 710, it can be seen that the advantage of the point wise function over the pair wise function lies in flexible sampling for negative instances.

For example, it can be seen that the pair wise function can be paired with only one negative case, and the point wise function has the advantage of being able to set the sampling rate appropriately for the model.

By checking the performance for changes in the negative sampling rate in the graph 700 and 710, it can be seen that the optimal sampling rate is about 3 to 6.

That is, the deep learning-based user-customized course recommendation system according to an embodiment of the present invention extracts at least one interaction characteristic by controlling the negative sampling rate of the point wise function, It has the advantage of reducing point wise error.

Therefore, the present invention is a latent feature learned in each of the user behavior network (UBN) layer, the course attribution network (CAN) layer, and the deep neural network (DNN) layer. Through this, deep learning can be performed by minimizing pointwise error by using each user's attendance at an individual course as the target variable.

The device described above may be implemented with hardware components, software components, and/or a combination of hardware components and software components. For example, devices and components described in embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), It may be implemented using one or more general-purpose or special-purpose computers, such as a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. A processing device may execute an operating system (OS) and one or more software applications that run on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device includes multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include a plurality of processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are possible.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. Software and/or data may be used on any type of machine, component, physical device, virtual equipment, computer storage medium or device to be interpreted by or to provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

Although the embodiments have been described with limited drawings as described above, various modifications and variations can be made by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

100: Deep learning-based user-customized course recommendation system
110: data collection unit 120: deep learning model processing unit
130: Recommendation section

Claims (16)

a data collection unit that collects a plurality of user behavior data and a plurality of course attribute data;
Extract at least one user behavior characteristic by learning the plurality of collected user behavior data, extract at least one course attribute characteristic by learning the plurality of collected course attribute data, and extract the at least one user behavior a deep learning model processing unit that learns the characteristics and the extracted at least one course attribute characteristic together to extract at least one interaction characteristic related to whether the user has taken a plurality of courses; and
A recommendation unit that recommends at least one course based on the extracted at least one interaction characteristic,
The deep learning model processor implements flexible sampling for a negative instance of a point wise function in learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together. In extracting the at least one interaction characteristic by controlling the negative sampling rate of the point wise function in accordance with the deep learning model based on, the point wise error of the deep learning model is reduced.
Deep learning-based user-tailored course recommendation system. According to paragraph 1,
The data collection unit is characterized in that it collects the plurality of user behavior data including a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data. doing
Deep learning-based user-tailored course recommendation system. According to paragraph 1,
The data collection unit is characterized in that it collects the plurality of course attribute data including a plurality of course data and a plurality of major data.
Deep learning-based user-tailored course recommendation system. According to paragraph 1,
The deep learning model processing unit first embeds a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data included in the plurality of user behavior data. Converting the converted first embedding vectors into vectors, inputting the converted first embedding vectors into a user behavior network (UBN) layer, and learning the input first embedding vectors to extract the at least one user behavior characteristic. characterized by
Deep learning-based user-tailored course recommendation system. According to paragraph 4,
The deep learning model processing unit converts a plurality of course data and a plurality of major data included in the plurality of course attribute data into second embedding vectors, and converts the converted second embedding vectors into course attribute data. Characterized in extracting the at least one course attribute characteristic by inputting it to a network (course attribution network, CAN) layer and learning the input second embedding vectors.
Deep learning-based user-tailored course recommendation system. According to clause 5,
The deep learning model processing unit inputs the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic to a deep neural network (DNN) layer, and the input at least one user behavior characteristic And extracting the at least one interaction characteristic by learning the at least one input course attribute characteristic together.
Deep learning-based user-tailored course recommendation system. According to clause 6,
The deep learning model processing unit uses the user behavior network (UBN) layer and the course attribution network (CAN) layer as a parallel sub-network, and the parallel sub-network and Characterized by connecting the deep neural network (DNN) layer and using the deep neural network (DNN) layer as a prediction layer.
Deep learning-based user-tailored course recommendation system. delete According to paragraph 1,
The recommender is characterized in that it recommends the at least one course based on implicit feedback that indirectly reflects the extracted at least one interaction characteristic in the recommendation of the at least one course.
Deep learning-based user-tailored course recommendation system. According to paragraph 1,
The recommendation unit ranks the at least one course, lists the ranked at least one course, and provides a recommendation list including the listed at least one course.
Deep learning-based user-tailored course recommendation system. According to clause 10,
The recommendation unit randomly selects a course with which the user has not interacted among the at least one course and ranks the at least one course.
Deep learning-based user-tailored course recommendation system. In a data collection unit, collecting a plurality of user behavior data and a plurality of course attribute data;
In a deep learning model processing unit, extracting at least one user behavior characteristic by learning the plurality of collected user behavior data;
extracting at least one course attribute characteristic by learning the plurality of collected course attribute data, in the deep learning model processing unit;
In the deep learning model processing unit, the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic are learned together to extract at least one interaction characteristic related to whether the user takes a plurality of courses. steps; and
In a recommendation unit, recommending at least one course based on the extracted at least one interaction characteristic,
The step of extracting at least one interaction characteristic related to whether a user takes a plurality of courses by learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together,
In jointly learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic, a deep learning model is based on a flexible sampling implementation for a negative instance of a point wise function. In extracting the at least one interaction characteristic by controlling the negative sampling rate of the point wise function in accordance with, reducing the point wise error of the deep learning model.
Deep learning-based user-tailored course recommendation method. According to clause 12,
The step of collecting a plurality of user behavior data and a plurality of course attribute data includes,
collecting the plurality of user behavior data including a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data; and
Characterized in that it includes the step of collecting the plurality of course attribute data including a plurality of course data and a plurality of major data.
Deep learning-based user-tailored course recommendation method. According to clause 12,
The step of extracting at least one user behavior characteristic by learning the collected plurality of user behavior data,
Converting a plurality of user data, a plurality of job data, a plurality of certificate data, and a plurality of language data included in the plurality of user behavior data into first embedding vectors, Inputting the converted first embedding vectors into a user behavior network (UBN) layer, and extracting the at least one user behavior characteristic by learning the input first embedding vectors. doing
Deep learning-based user-tailored course recommendation method. According to clause 14,
The step of extracting at least one course attribute characteristic by learning the plurality of collected course attribute data,
A plurality of course data and a plurality of major data included in the plurality of course attribute data are converted into second embedding vectors, and the converted second embedding vectors are connected to a course attribution network. CAN) layer, and extracting the at least one course attribute feature by learning the input second embedding vectors.
Deep learning-based user-tailored course recommendation method. According to clause 15,
The step of extracting at least one interaction characteristic related to whether a user takes a plurality of courses by learning the extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic together,
The extracted at least one user behavior characteristic and the extracted at least one course attribute characteristic are input to a deep neural network (DNN) layer, and the input at least one user behavior characteristic and the input at least one course attribute characteristic are input to a deep neural network (DNN) layer. Characterized in that it comprises the step of extracting the at least one interaction characteristic by learning together the course attribute characteristics of
Deep learning-based user-tailored course recommendation method.