KR20230060401A - Method for satisfaction-based contents recommendation and electronic device thereof - Google Patents

Abstract

According to an embodiment of the present invention, a lecture table of a specified range is determined among lecture tables sorted based on satisfaction, and m recommended candidate lectures are determined based on the names of lectures included in the determined lecture table and a designated algorithm. and a content recommendation service for determining recommended content based on designated keywords of lectures included in the lecture table sorted based on the level of satisfaction and designated keywords of the m recommended candidate lectures, and providing the determined recommended contents to the user. A method and its electronic device are disclosed.

Description

Satisfaction-based content recommendation method and electronic device thereof

The present invention relates to a satisfaction-based content recommendation method and an electronic device therefor. In detail, a method for recommending content based on the satisfaction level of content consumed by a user and an electronic device thereof are disclosed.

With the development of Internet technology, lectures that were provided only offline began to be distributed to users through the Internet as well. Lectures provided through the Internet are called online lectures. Online lectures have the advantage of allowing users to take lectures at a desired time and place, and require relatively low tuition compared to offline lectures.

Online lectures are dissatisfied with men and women of all ages and are widely used by various age groups, and in particular, the demand for online lectures for various studies continues to increase.

The above background art is possessed or acquired by the inventor in the process of deriving the disclosure of the present application, and cannot necessarily be said to be known art disclosed to the general public prior to the present application.

An object of the embodiment is to provide a method and an electronic device for effectively recommending next content based on a user's satisfaction with various content.

The problems to be solved through various embodiments of the present invention are not limited to the mentioned problems, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

According to an embodiment of the present invention, a method of operating an electronic device providing a content recommendation service includes determining a lecture table of a specified range from among lecture tables sorted based on satisfaction levels; determining m recommended candidate lectures based on names of lectures included in the determined lecture table and a designated algorithm; determining recommended contents based on designated keywords of lectures included in a lecture table sorted based on the level of satisfaction and designated keywords of the m recommended candidate lectures; and providing the determined recommended content to the user.

Here, the designated algorithm may include a BERT4REC language model in which a mask is applied to designated parts of operation steps of a BERT (bidirectional encoder representations from transformers, BERT) language model.

Here, the step of determining m recommended candidate lectures based on the names of lectures included in the determined lecture table and the designated algorithm determines the recommended candidate lectures based on the names of lectures included in the determined lecture table and the designated algorithm. determining n recommended candidate lectures by performing the operation n times; and determining the top m courses as the m recommended candidate courses based on the satisfaction level of each of the n recommended candidate courses, wherein m and n may be natural numbers.

Here, the step of determining the n recommended candidate lectures by performing the operation for determining the recommended candidate lectures n times based on the names of the lectures included in the determined lecture table and the specified algorithm is the time point at which each of the n computations is performed. It may be to determine the table of lectures in the specified range.

Here, the recommended contents are designated keywords of lectures included in a parsing operation result of vectors including at least some of the lectures included in the lecture table sorted based on satisfaction and the names of the m recommended candidate lectures. , and determining among the m recommended candidate lectures based on designated keywords of the m recommended candidate lectures.

Here, the recommended contents are designated keywords of lectures included in a parsing operation result of vectors including at least some of the lectures included in the lecture table sorted based on satisfaction and the names of the m recommended candidate lectures. , And it may be to determine among the instructors included in the instructor satisfaction table sorted based on the specified keywords of the m recommended candidate lectures.

Here, the lecture table of the specified range may be determined based on the satisfaction with each lecture among the lecture tables sorted based on the satisfaction level and the time when each lecture was first provided or first inquired.

According to various embodiments, an electronic device providing a content recommendation service may include a communication unit configured to communicate with at least one user device; a storage unit for storing data transmitted and received through the communication unit; and determining a lecture table of a specified range among the lecture tables sorted based on satisfaction, determining m recommended candidate lectures based on the names of lectures included in the determined lecture table and a designated algorithm, and sorting based on the satisfaction level. and a processing unit configured to determine recommended contents based on designated keywords of lectures included in the lecture table and the designated keywords of the m recommended candidate lectures, and provide the determined recommended contents to a user.

According to various embodiments of the present disclosure, content optimized for the user may be recommended because the next content recommendation is analyzed based on the user's satisfaction level.

According to one embodiment of the present invention, it is not limited to those mentioned above, and other effects not mentioned above will be clearly understood by those skilled in the art from the description below.

1 is a diagram showing the configuration of a device according to an embodiment of the present invention.
2 illustrates a data table creation operation of an electronic device according to an embodiment of the present invention.
3 illustrates a data table creation operation of an electronic device according to an embodiment of the present invention.
4 illustrates a data processing structure of an electronic device according to an embodiment of the present invention.
5 shows a simple structure of a transformer in an electronic device according to an embodiment of the present invention.
6 is a flowchart of an operation of providing a content recommendation service in an electronic device according to an embodiment of the present invention.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, since various changes can be made to the embodiments, the scope of rights of the patent application is not limited or limited by these embodiments. It should be understood that all changes, equivalents or substitutes to the embodiments are included within the scope of rights.

Terms used in the examples are used for descriptive purposes only and should not be construed as limiting. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "include" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiment belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in the present application, they should not be interpreted in an ideal or excessively formal meaning. don't

In addition, in the description with reference to the accompanying drawings, the same reference numerals are given to the same components regardless of reference numerals, and overlapping descriptions thereof will be omitted. In describing the embodiment, if it is determined that a detailed description of a related known technology may unnecessarily obscure the subject matter of the embodiment, the detailed description will be omitted.

Also, terms such as first, second, A, B, (a), and (b) may be used in describing the components of the embodiment. These terms are only used to distinguish the component from other components, and the nature, order, or order of the corresponding component is not limited by the term. When an element is described as being “connected,” “coupled to,” or “connected” to another element, that element may be directly connected or connected to the other element, but there may be another element between the elements. It should be understood that may be "connected", "coupled" or "connected".

Components included in one embodiment and components having common functions will be described using the same names in other embodiments. Unless stated to the contrary, descriptions described in one embodiment may be applied to other embodiments, and detailed descriptions will be omitted to the extent of overlap.

According to various embodiments of the present disclosure, an electronic device provides a method for recommending next content to a user based on satisfaction with content provided online, and the electronic device.

Here, the contents may include lectures provided online and information related thereto. For example, the content may include lectures provided online or instructors providing lectures. Here, the lecture may include at least one of video and/or image.

Hereinafter, preferred embodiments of the present invention will be described with reference to FIGS. 1 to 6 .

The drawings attached to this specification illustrate a preferred embodiment of the present invention, and together with the detailed description of the invention serve to further understand the technical idea of the present invention, the present invention is limited only to those described in the drawings should not be interpreted.

1 is a diagram showing the configuration of a device according to an embodiment of the present invention.

Referring to FIG. 1 , an electronic device 100 may include a processing unit 110 , a storage unit 120 , and a communication unit 130 .

The processing unit 110 may provide a content recommendation service. According to an embodiment, the processing unit 110 may be connected to a user device or a server through the communication unit 130 and transmit/receive data. At this time, the electronic device may provide a program for content recommendation service (app, application, tool, plug-in, etc., hereinafter referred to as a content recommendation program) to the user device, and recommend content from the electronic device through the content recommendation program installed in the user device. service can be provided.

Hereinafter, in various embodiments, providing a content recommendation service through an electronic device or user device may be understood as providing events of the content recommendation service through a content recommendation program in the electronic device or user device.

Here, the content recommendation service is described as being provided through a content recommendation program installed in the user device, but is not limited thereto, and may be provided through other pre-installed programs or temporarily installed programs. According to an embodiment, the content recommendation service may be provided by including (or connecting to) at least a part of a database provided free of charge or for a fee in an external device of the electronic device.

The processing unit 110 provides the content requested from the user device to the user device according to the user's selection. The electronic device 100 may be connected to a plurality of user devices, and may provide content requested from each user device to each requested user device.

The processing unit 110 may determine the satisfaction of the provided lecture and/or the instructor's satisfaction based on the request of the user device. According to an embodiment, the processing unit 110 may determine satisfaction with respect to each provided lecture and may determine satisfaction with respect to a group including the lecture.

According to various embodiments of the present invention, satisfaction is the number of views of lectures, the number of views of lecturers, the evaluation score of each lecture, the evaluation score of instructors, the number of evaluations of lectures, the number of evaluations of lecturers, the number of views of lectures, the lectures provided by the instructor It may be determined based on at least some of the cumulative inquiry time of the lectures, the average user evaluation score of the lectures provided by the instructor, the comments about the lecturer, and the comments about the lecture. Here, satisfaction may be determined by at least a part of a score or a percentage.

For example, the processing unit 110 may count the number of views of the first group when eight lectures are included in the first group and all eight lectures are viewed by the same user. In addition, when counting the number of inquiries of the first group, the processing unit 110 may give additional points to at least some of the first group, at least one lecture included in the first group, and the instructors, and reflect them to satisfaction.

The processing unit 110 may store in the storage unit 120 including information of a user who has inquired about the lecture and/or the instructor in storing the satisfaction level determined for the lecture and the instructor.

Here, the user's information may include information on at least one of the user's name, ID, age, date of birth, address, preferred lecture and/or instructor, and current lecture.

The processing unit 110 may recommend content to be consumed next by the user based on the artificial neural network algorithm and accumulated data. Here, data generated based on the content being provided by the processing unit 110 and/or data stored in the storage unit 120 may be defined as accumulated data. According to an embodiment, the processing unit 110 may learn an artificial neural network algorithm based on satisfaction levels of lectures included in the accumulated data, and may determine the user's next content based on the learned artificial neural network.

Here, the artificial neural network algorithm may be expressed as a deep learning algorithm and various algorithms related thereto.

The processing unit 110 may process the operation of the artificial neural network algorithm based on various language models. According to an embodiment, the processor 110 may provide a content recommendation service based on a bidirectional encoder representations from transformers (BERT) language model.

The butt model includes a transformer (TRM), and may include, for example, a multi-layer transformer.

The transformer may include an encoder or decoder including at least one of a multi-head attention mechanism and a position-wise feed forward mechanism. .

According to an embodiment, each of the multi-head attention mechanism and the position-wise feed forward mechanism may be configured as a sublayer of an encoder or decoder.

The processing unit 110 may provide a content recommendation service based on a BERT4REC model that processes a mask in some of the calculation steps of the BERT model. According to an embodiment, the Burt4REC model may apply a mask to a last calculation step among calculation steps.

The processing unit 110 may generate various data tables to be applied to the artificial neural network algorithm based on the accumulated data. For example, processing unit 110 may generate a lecture table containing provided lecture data and an instructor satisfaction table containing instructor data. For each lecture, the processor 110 may include in the lecture table at least some of lecture name (or lecture identification information), lecture satisfaction data, user data that inquired the lecture, and instructor data that provided the lecture.

In addition, the processing unit 110 includes at least a part of the instructor name (or instructor identification information), lecture data provided by the instructor, user data for viewing the lecture, and satisfaction data for each lecture for each instructor in the instructor satisfaction table. can do. In addition, the processing unit 110 may include at least some of characteristics assigned to instructors, categories of lectures provided by instructors, keywords assigned to instructors, keywords assigned to instructors, and keywords assigned to each lecture in the instructor satisfaction table.

According to an embodiment, the processing unit 110 may create a separate table for at least some of the data included in the lecture table and the instructor satisfaction table. For example, processing unit 110 may generate a lecture satisfaction table including lectures and satisfaction data for each lecture.

The processing unit 110 may further include information about users who have watched the lectures for each lecture in the lecture satisfaction table, or may link the lecture table with the lecture table.

The processing unit 110 may process at least some data and operations based on vectors when performing calculations of artificial neural network algorithms or related data for providing content recommendation services.

For example, the processing unit 110 may apply a vector to at least a part of data and a table generated based on accumulated data, and at least part of them may be processed by vector operation.

According to various embodiments of the present invention, the term 'inquiry' is not limited to simply checking the content by selecting a lecture or a superior, and may include meanings of watching, learning, and proceeding.

According to one embodiment, the processing unit 110 creates a table for determining recommended content through a content recommendation service or selects data included in the table, lecture or instructor data that satisfies the number of hits of a specified criterion. You can check.

According to various embodiments, the processing unit 110 converts the number of views into scores to check data of lectures or instructors that satisfy the scores of a designated criterion.

According to various embodiments of the present disclosure, the processing unit 110 may check data of lectures or instructors who have received an evaluation of a specified criterion. Here, the evaluation of the designated criterion may include at least a part of the number of evaluations of users satisfying the number of designated criteria and evaluation scores of users satisfying the scores of the designated criterion.

As described above, the processing unit 110 may generate a table based on data satisfying specified criteria or may apply the artificial neural network algorithm. According to one embodiment, when 8 points out of 0 to 10 points are determined as the reference score, the processing unit 110 provides lectures or instructors satisfying 8 points or more (or more than) as data for processing the content recommendation service. can be included in

The storage unit 120 may store various data used by at least one component (eg, the processing unit 110 or the communication unit 130) of the electronic device 110. The data is, for example, a content recommendation service. It may include a program (or software) for, accumulated data, various tables generated based on the accumulated data, and input data or output data for commands related thereto.

The storage unit 120 may include at least some of an artificial neural network algorithm for providing a content recommendation service and a mechanism, operator, and language model related thereto.

The storage unit 120 stores at least some of the data processed through the processing unit 110, such as instructor data, lecture data provided by the instructor, satisfaction data for the instructor and/or lecture, and user data that has viewed the instructor and/or lecture. Accumulated data can be stored.

Accumulated data stored in the storage unit 120 may include at least some data of a time when a user inquired about a lecture, a number of times the same lecture was searched, and a math time.

The storage unit 120 includes at least one data table generated based on the accumulated data, learning data of the artificial neural network algorithm, test data, and recommended lecture data provided to the user, or at least some of the recommended instructor data and data generated during the process. can be saved.

Data stored in the storage unit 120 may be changed, corrected, deleted, or created through the processing unit 110 based on a manager input of the electronic device 100 or a user input of the user device.

The storage unit 120 may include volatile memory or non-volatile memory.

The communication unit 130 may support establishment of a wired communication channel between the electronic device 100 and at least one other electronic device (eg, a user device or a server), establishment of a wireless communication channel, and communication through the established communication channel. there is.

The communication unit 130 may include one or more communication processors that are operated independently or subordinately to the processing unit 110 and support wireless communication. According to an embodiment, the communication unit 130 may be a wireless communication module (eg, a cellular communication module, a short-distance wireless communication module, or a global navigation satellite system (GNSS) communication module) or a wired communication module (eg, a local area network (LAN)) communication module or power line communication module).

The communication unit 130 communicates with an external electronic device through a short-range communication network such as Bluetooth, WiFi direct, or infrared data association (IrDA) and/or a long-distance communication network such as a cellular network, the Internet, or a computer network (eg, LAN or WAN). can communicate

Various types of communication modules constituting the communication unit 130 may be integrated as one component (eg, a single chip) or implemented as a plurality of separate components (eg, multiple chips).

According to various embodiments, the electronic device 100 is not limited to the above-described components and may further include at least one of an input unit (not shown) for inputting data and an output unit (not shown) for outputting data. can

According to various embodiments of the present invention, the electronic device 100 or user device may be applied to all information and communication devices including mobile communication terminals, multimedia terminals, wired terminals, fixed terminals, and internet protocol (IP) terminals. As an example, the user device includes a mobile phone, a personal computer (PC), a portable multimedia player (PMP), a mobile internet device (MID), a smartphone, a tablet PC, a phablet PC, and a laptop computer ( notebook), etc. can be exemplified.

Here, the server is one entity existing on the network and serves as a web server, database server, and application server. According to a preferred embodiment, the server provides various contents to the user device based on the processing of the electronic device 100, stores the user's search history for the provided contents, and based on the search history, the user prefers Content can be recommended.

2 illustrates a data table creation operation of an electronic device according to an embodiment of the present invention.

The electronic device 100 may create a data table for providing a content recommendation service based on data of the storage unit 120 or a server connected to the electronic device 100 . According to an embodiment, the electronic device 100 may create a lecture table for pre-processing based on data about lectures that users searched for from accumulated data.

The electronic device 100 includes a lecture table for preprocessing including data in which at least two or more of the user-inquired lecture name, the ID (or user identification information) of the user who inquired the lecture, and the hash value of the user are matched for each user. can create

In addition, the electronic device 100 may create a user table for preprocessing including user information from accumulated data. The electronic device 100 may generate data matching the user ID and user name of each user as a user table for preprocessing.

The electronic device 100 may generate lecture data based on the preprocessing lecture table and the preprocessing user table. The electronic device 100 may create a lecture table by including data obtained by matching the user name and the lecture name searched by the user for each user.

The electronic device 100 may include, for each user, information about at least a part of the first provided (generated) time of a lecture inquired by the user, the first inquired time, and satisfaction with the corresponding lecture in the lecture table.

3 illustrates a data table creation operation of an electronic device according to an embodiment of the present invention.

The electronic device 100 may generate an instructor satisfaction table for preprocessing to provide a content recommendation service based on the storage unit 120 or accumulated data of a server connected to the electronic device 100 .

The electronic device 100 may create a creator table for preprocessing including whether or not a creator is a creator in the user table for preprocessing. The electronic device 100 may add and record a new column for a user having a true value in the creator table for preprocessing. The electronic device 100 may determine a creator's user name having a true value as the instructor's name, and may create an instructor satisfaction table for each instructor including the determined instructor's name.

The electronic device 100 may generate an instructor satisfaction table based on the pre-processing lecture table and the pre-processing creator table. At this time, the electronic device 100 may include in the instructor satisfaction table at least some of the instructor name, the number of lectures provided by the instructor, satisfaction with all and/or each lecture, and keywords designated for the instructor.

The electronic device 100 may generate a lecture satisfaction table including satisfaction data for each lecture based on the lecture table for preprocessing. The electronic device 100 may include keywords designated for each lecture in the lecture satisfaction table. When generating the lecture satisfaction table, the electronic device 100 may include the user's lecture in the lecture satisfaction table based on whether the user included in the preprocessing creator table is a creator or not.

The electronic device 100 may determine an instructor or a cafe category of a lecture provided by the instructor based on the instructor satisfaction table and the keywords of the lecture satisfaction table.

4 illustrates a data processing structure of an electronic device according to an embodiment of the present invention.

The electronic device 100 may apply an artificial neural network algorithm including various language models in order to provide a content recommendation service that meets the user's needs.

Referring to FIG. 4 , the electronic device 100 shows a simple processing structure of a vertex language model including a plurality of transformers.

The electronic device 100 may process an artificial neural network algorithm including a BERT language model, and more specifically, a BERT4REC language model masked to a designated position of a BERT operation may be applied.

According to an embodiment, the Bert4REC language model may be processed in a state in which a position (eg, the last operation) specified in the transformer's attention matrix operation is masked.

In FIG. 4, v={v1, . . . , vt-1, vt} may be applied including at least a part of the data table for preprocessing. According to an embodiment, the electronic device 100 checks information (eg, view history or watched history) on lectures arranged according to the time point viewed in the lecture table, and includes at least a part of v The electronic device 100 may determine vt based on the determined components of v and an artificial neural network algorithm, and may determine a recommended lecture or a recommended instructor based on vt.

The electronic device 100 may perform learning of an artificial neural network algorithm in order to provide a content recommendation service.

Referring to FIG. 4 , the electronic device 100 may replace lecture names of lectures included in data tables with d-dimensional vectors in an embedding layer. Here, the electronic device 100 may apply at least one algorithm of one-hot encoding and word2vec.

The electronic device 100 uses vectors {v1, . . . input in the substitution process. , vt-1}, positional encoding for marking position information (or order) of each vector may be performed.

The electronic device 100 may process the result value of the embedding layer as data for deep learning learning of an artificial neural network algorithm. The electronic device 100 may perform artificial neural network learning and recommended content determination by applying the result value of the embedding layer to the verte language model as an input value.

As shown in FIG. 4, the Burt language model includes a plurality of layers composed of a plurality of transformers, and each transformer includes at least one head attention as shown in FIG. 5. It can be.

5 shows a simple structure of a transformer in an electronic device according to an embodiment of the present invention.

Referring to FIG. 5 , each transformer may include an encoder including multi-head attention. The electronic device 100 may process to learn different information for each multi-head attention of the Burt language model, and may determine a processing result close to user needs by repeatedly performing an operation based on layers composed of a plurality of transformers.

The electronic device 100 may minimize an error by applying negative sampling and/or backpropagation in the learning operation of the artificial neural network algorithm.

The electronic device 100 may provide a content recommendation service based on the learned artificial neural network algorithm.

6 is a flowchart of an operation of providing a content recommendation service in an electronic device according to an embodiment of the present invention.

The electronic device 100 may provide a content recommendation service based on an artificial neural network algorithm including a Bert4REC language model.

In step 601, the electronic device 100 determines a table of lectures of a specified range period sorted based on the level of satisfaction. According to an embodiment, the electronic device 100 may determine, for example, a lecture table for the last year by referring to the lecture table.

According to various embodiments, the lecture table for the specified range period may be determined based on various values, such as the last 6 months, the recent 2 years, or the last year or 2021 to the present time, rather than being limited to the last year.

According to an embodiment, the electronic device 100 selects lectures included in a lecture table of a designated range period 1, . . . , can be determined by t-1 lectures.

In step 603, the electronic device 100 recommends m candidates based on the names (or lecture names) of lectures (or lectures 1, ..., t-1 times) included in the lecture table of the specified range period and the specified algorithm. decide on a lecture

As described above, the electronic device 100 may provide a content recommendation service based on an artificial neural network algorithm including BERT4REC. The electronic device 100 is 1, . . . , it is possible to determine t recommendation candidate lectures by using the lecture names of t-1 lectures as an input value. At this time, the electronic device 100 1, . . . , in inputting t-1 lectures, positional encoding can be performed.

The electronic device 100 may determine m recommended candidate lectures by repeatedly performing the operation of determining the recommended candidate lectures t times. At this time, the electronic device 100 may determine the recommended candidate lecture in the same way as the previously performed operation, or may determine the recommended candidate lecture by applying negative sampling and/or backpropagation.

The electronic device 100 may perform the operation of determining recommended candidate lectures m times to determine m recommended candidate lectures, or perform the operation n times to determine m recommended candidate lectures among the determined n recommended candidate lectures. can be determined by

Here, n and m are integers greater than or equal to 1, and n may be greater than or equal to m.

In step 605, the electronic device 100 may determine recommended lectures or instructors based on designated keywords of lectures included in the lecture table and designated keywords of determined m recommended candidate lectures.

In determining the recommended lecture as the recommended content, the electronic device 100 may determine the specified number of keywords having the largest number among the specified keywords of the m recommended candidate lectures. The electronic device 100 may determine designated keywords of lectures included in the lecture table that match the designated number of keywords, and may determine lectures of the determined keywords as recommended content. At this time, the electronic device 100 may determine a plurality of lectures as recommended content.

In determining the recommended instructor as the recommended content, the electronic device 100 may determine the specified number of keywords having the largest number among the specified keywords of the m recommended candidate lectures. The electronic device 100 may determine designated keywords of instructors included in the instructor satisfaction table that match the designated number of keywords, and may determine instructors of the determined keywords as recommended content. At this time, the electronic device 100 may determine a plurality of instructors as recommended content.

The lecture table processed in step 605 may be a parsed lecture table. For example, it may be a lecture table parsed after performing an m x d matrix operation of vectors included in the lecture table.

In step 607, the electronic device 100 provides information on the determined lecture or instructor.

The electronic device 100 may filter and provide the determined plurality of contents. For example, when the recommended content is a lecture, the electronic device 100 sorts the determined plurality of lectures based on at least one column of the first provided time, the first viewed time, and satisfaction, and provides a designated part as the recommended content. can

According to various embodiments of the present disclosure, when the recommended content is a lecturer, the electronic device 100 selects the determined plurality of lectures as the instructor's name, the instructor's category, the number of lectures provided by the instructor, the total number of views of the lectures, the number of views of each lecture, and the number of lectures provided by the instructor. Sorting based on at least one column of the inquiry time for each lecture, the cumulative inquiry time of lectures provided by the instructor, the evaluation score of each lecture provided by the instructor, and the average user evaluation score of lectures provided by the instructor, A designated portion may be provided as recommended content.

The electronic device 100 may provide recommended content as a result of filtering by applying user needs specified in a plurality of aligned lectures or instructors, for example, a keyword, a specified level of satisfaction with lectures or instructors, and the like.

According to various embodiments, a procedure (or procedure) for learning an artificial neural network algorithm of the electronic device 100 may be performed identically or similarly to a content recommendation procedure. Similarly, the process of the electronic device 100 recommending content based on the artificial neural network algorithm may be a process of learning the artificial neural network algorithm at the same time.

As described above, as the electronic device 100 performs an operation of determining recommended content through a content recommendation service, learning of an artificial neural network algorithm is also progressed, and recommended content that meets user needs can be provided.

According to an embodiment of the present invention, a method of operating an electronic device providing a content recommendation service includes determining a lecture table of a specified range from among lecture tables sorted based on satisfaction levels; determining m recommended candidate lectures based on names of lectures included in the determined lecture table and a specified algorithm; determining recommended contents based on designated keywords of lectures included in a lecture table sorted based on the level of satisfaction and designated keywords of the m recommended candidate lectures; and providing the determined recommended content to the user.

Here, the designated algorithm may include a BERT4REC language model in which a mask is applied to designated parts of operation steps of a BERT (bidirectional encoder representations from transformers, BERT) language model.

Here, the step of determining m recommended candidate lectures based on the names of lectures included in the determined lecture table and the designated algorithm determines the recommended candidate lectures based on the names of lectures included in the determined lecture table and the designated algorithm. determining n recommended candidate lectures by performing the operation n times; and determining the top m courses as the m recommended candidate courses based on the satisfaction level of each of the n recommended candidate courses, wherein m and n may be natural numbers.

Here, the step of determining the n recommended candidate lectures by performing the operation for determining the recommended candidate lectures n times based on the names of the lectures included in the determined lecture table and the specified algorithm is the time point at which each of the n computations is performed. It may be to determine the table of lectures in the specified range.

Here, the recommended contents are designated keywords of lectures included in a parsing operation result of vectors including at least some of the lectures included in the lecture table sorted based on satisfaction and the names of the m recommended candidate lectures. , and determining among the m recommended candidate lectures based on designated keywords of the m recommended candidate lectures.

Here, the recommended contents are designated keywords of lectures included in a parsing operation result of vectors including at least some of the lectures included in the lecture table sorted based on satisfaction and the names of the m recommended candidate lectures. , And it may be to determine among the instructors included in the instructor satisfaction table sorted based on the specified keywords of the m recommended candidate lectures.

Here, the lecture table of the specified range may be determined based on the satisfaction with each lecture among the lecture tables sorted based on the satisfaction level and the time when each lecture was first provided or first inquired.

According to various embodiments, an electronic device providing a content recommendation service may include a communication unit configured to communicate with at least one user device; a storage unit for storing data transmitted and received through the communication unit; and determining a lecture table of a specified range among the lecture tables sorted based on satisfaction, determining m recommended candidate lectures based on the names of lectures included in the determined lecture table and a designated algorithm, and sorting based on the satisfaction level. and a processing unit configured to determine recommended contents based on designated keywords of lectures included in the lecture table and the designated keywords of the m recommended candidate lectures, and provide the determined recommended contents to a user.

As described above, by providing content recommendation services according to various embodiments, it is possible to recommend lectures and instructors optimized for a user based on satisfaction.

As described above, although the embodiments have been described with limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, the described techniques may be performed in an order different from the method described, and/or components of the described system, structure, device, circuit, etc. may be combined or combined in a different form than the method described, or other components may be used. Or even if it is replaced or substituted by equivalents, appropriate results can be achieved. Therefore, it should be said that other implementations, other embodiments, and equivalents of the claims are within the scope of the following claims.

100: electronic device 110: processing unit
120: storage unit 130: communication unit

Claims (8)

In the operating method of an electronic device providing a content recommendation service,
Determining a lecture table of a specified range among lecture tables sorted based on satisfaction;
determining m recommended candidate lectures based on names of lectures included in the determined lecture table and a designated algorithm;
determining recommended contents based on designated keywords of lectures included in a lecture table sorted based on the level of satisfaction and designated keywords of the m recommended candidate lectures; and
A method of operating an electronic device that provides a content recommendation service, including providing the determined recommended content to a user. According to claim 1,
The specified algorithm includes a BERT4REC language model in which a mask is applied to a specified part of the operation steps of a BERT (bidirectional encoder representations from transformers, BERT) language model. Electronic providing content recommendation service How the device works. According to claim 1,
Determining m recommended candidate lectures based on the names of lectures included in the determined lecture table and a specified algorithm,
determining n recommended candidate lectures by performing an operation for determining recommended candidate lectures n times based on the names of lectures included in the determined lecture table and a designated algorithm; and
Determining the top m lectures as the m recommended candidate lectures based on the satisfaction level of each of the n recommended candidate lectures;
Here, the operating method of an electronic device providing a content recommendation service, characterized in that m and n are natural numbers. According to claim 3,
Determining n recommended candidate lectures by performing an operation for determining recommended candidate lectures n times based on the names of lectures included in the determined lecture table and a designated algorithm,
A method of operating an electronic device providing a content recommendation service, characterized in that determining a lecture table in the specified range at the time of performing each of the n operations. According to claim 1,
The recommended contents may include lectures included in a lecture table sorted based on satisfaction and designated keywords of lectures included in a parsing operation result of vectors including at least some of the names of the m recommended candidate lectures, and An operating method of an electronic device providing a content recommendation service, characterized in that determining among the m recommended candidate lectures based on designated keywords of the m recommended candidate lectures. According to claim 1,
The recommended contents may include lectures included in a lecture table sorted based on satisfaction and designated keywords of lectures included in a parsing operation result of vectors including at least some of the names of the m recommended candidate lectures, and An operating method of an electronic device providing a content recommendation service, characterized in that determining among instructors included in an instructor satisfaction table sorted based on designated keywords of the m recommended candidate lectures. According to claim 1,
Characterized in that the lecture table of the specified range is determined based on the satisfaction with each lecture among the lecture tables sorted based on the satisfaction and the time when each lecture was first provided or first inquired, content recommendation service A method of operating an electronic device that provides In an electronic device providing a content recommendation service,
a communication unit that communicates with at least one user device;
a storage unit for storing data transmitted and received through the communication unit; and
Determines a lecture table in a specified range among lecture tables sorted based on the number of hits, determines m recommended candidate lectures based on the names of lectures included in the determined lecture table and a specified algorithm, and sorts based on the number of hits A processing unit configured to determine recommended content based on designated keywords of lectures included in the lecture table and designated keywords of the m recommended candidate lectures, and provide the determined recommended contents to a user; providing a content recommendation service including electronics.

Images