KR20220075490A - Learning content recommendation method - Google Patents

Abstract

A learning content recommendation method is disclosed. A method for recommending learning content includes: collecting digital content online; generating a category list by applying the collected digital content to a classification model, and classifying each of the digital content according to the generated category list; After analyzing the digital content and extracting each word, a basic word dictionary is generated, a dictionary for each learning level is generated using the basic word dictionary, and a coverage ratio of the dictionary for each learning level of the digital content is derived, and the digital content determining a learning sequence of and recommending learning content in consideration of a user learning result and a learning sequence of the digital content.

Description

Learning content recommendation method

The present invention relates to a method for recommending learning content.

Basic foreign language skills are often required in a variety of job-related situations while working at work, but most professionals are unable to carry out their work properly because they do not speak the basic foreign language despite their excellent work skills. The vocational competence required by the National Competency Standards (NCS) includes job performance ability and vocational basic competence. It is a common skill required to successfully perform a job in the occupation.

As foreign language ability becomes more important, various contents for learning foreign languages are provided. However, most of the conventional learning contents are learning methods in the form of providing study sheets and vocabulary books so that learners can learn voluntarily.

When a learner encounters words that are difficult to memorize, not only will their interest be lowered, but their achievement will also be lowered due to the continuing difficulty in memorizing, which in turn leads to a decrease in their motivation to learn.

Korean Patent Publication No. 10-2018-0000444

An object of the present invention is to provide a method for recommending learning content that can be provided by processing online digital content into learning content.

Another object of the present invention is to provide a learning content recommendation method capable of recommending a user's learning level and content learned by similar learners.

Another object of the present invention is to provide a method for recommending learning content that is similar to a user's learning level and can improve learning ability by distinguishing between difficult content and easy content.

According to one aspect of the present invention, there is provided a learning content recommendation method that can be provided by processing the online digital content into learning content.

By providing the learning content recommendation method according to the embodiment of the present invention, there is an advantage in that it can be processed and provided as learning content using online digital content.

In addition, the present invention also has an advantage in that the user's learning level, the user's learning goals or preferences, and content learned by similar learners can be recommended.

In addition, the present invention is similar to the user's learning level and has an advantage in that it is possible to improve the learning ability by separately providing difficult content and easy content.

1 is a diagram schematically showing the configuration of a content recommendation system for language learning according to an embodiment of the present invention 3 is a diagram illustrating sample data for category classification according to an embodiment of the present invention;
4 is a flowchart illustrating a method of recommending learning content according to an embodiment of the present invention.
5 is a diagram illustrating a diagnostic test set according to an embodiment of the present invention.
6 is a diagram illustrating recommended content for each category according to an embodiment of the present invention.
7 is a diagram illustrating recommended content for all documents according to an embodiment of the present invention.
8 is a diagram illustrating a result of arranging a recommendation order by reflecting weights according to an embodiment of the present invention;
9 is a flowchart illustrating a content recommendation method according to another embodiment of the present invention.
10 is a diagram illustrating data for selecting similar learners according to an embodiment of the present invention.
11 is a diagram illustrating a result of selecting similar learners according to an embodiment of the present invention.
12 is a diagram illustrating a list of contents learned by selected learners according to an embodiment of the present invention.
13 is a diagram illustrating an arrangement result by reflecting weights according to an embodiment of the present invention;
14 is a flowchart illustrating a content recommendation method according to another embodiment of the present invention.
15 is a diagram illustrating a similar content recommendation result according to another embodiment of the present invention.
16 is a diagram illustrating a result of classifying easy learning content reflecting coverage according to another embodiment of the present invention.
17 is a diagram illustrating a result of classifying difficult learning content reflecting coverage according to another embodiment of the present invention.
18 is a flowchart illustrating a content recommendation method according to an embodiment of the present invention.

Since the present invention can apply various transformations and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

In describing the present invention, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, numbers (eg, first, second, etc.) used in the description process of the present specification are only identification symbols for distinguishing one component from other components.

Also, throughout the specification, when an element is referred to as “connected” or “connected” with another element, the one element may be directly connected or directly connected to the other element, but in particular It should be understood that, unless there is a description to the contrary, it may be connected or connected through another element in the middle. In addition, throughout the specification, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated. In addition, terms such as "unit" and "module" described in the specification mean a unit that processes at least one function or operation, which means that it can be implemented as one or more hardware or software or a combination of hardware and software .

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram schematically illustrating a configuration of a content recommendation system for language learning according to an embodiment of the present invention.

Referring to FIG. 1 , a content recommendation system 100 for language learning according to an embodiment of the present invention includes a user terminal 110 and a server 120 .

The user terminal 110 is a device possessed by a user, and is a device that receives content for language learning. The user terminal 110 may be, for example, an electronic device equipped with a communication function, such as a mobile communication terminal, a tablet PC, or a notebook computer.

The user terminal 110 is a device for receiving content for language learning from the server 120 and performing learning and review using the content.

In addition, the user terminal 110 may learn by receiving various learning contents from the server 120 based on the user's learning taste and learning information.

The server 120 is a device capable of processing various digital contents on the Internet and recommending and providing learning contents based on a user's learning taste and learning information.

According to an embodiment of the present invention, the server 120 may collect digital contents in various fields online, analyze and process them, and generate contents for learning. Next, the server 120 may recommend and provide learning content based on the user's learning preference and learning information among the learning content generated in response to the access of the user terminal 110 .

This will be more clearly understood by the following description.

2 is a flowchart illustrating a method of constructing a learning content database according to an embodiment of the present invention, and FIG. 3 is a diagram illustrating sample data for category classification according to an embodiment of the present invention.

In step 210, the server 120 collects digital content online.

The server 120 collects various digital contents published online. Here, the digital content may be various types of content posted online. For example, all digital contents posted online such as news, papers, blogs, etc. may be applied.

In step 215, the server 120 extracts the metadata of the digital content. Here, the metadata may include title, source, creation date, total length, number of sentences, average sentence length, number of words, core word set, non-core word set, category, content summary, content level, level coverage, and the like.

In step 220, the server 120 classifies the categories of digital content.

This will be described in more detail.

For example, the server 120 may classify a category of digital content using a machine learning technique. In more detail, categories of digital content can be classified using classification techniques such as RoBerta, GTP-3, RandomForest, Bayesian, SVM, LSTM, RNN, CNN, Transformer, and BERT, or machine learning techniques, deep learning techniques, etc. .

Hereinafter, for the convenience of understanding and explanation, a method of classifying a category of digital content using a Bayesian Classifier will be described.

The Bayesian classification model is a text classification model using Bayesian theorem to calculate conditional probabilities. A brief description of the Bayesian theorem is as follows.

P(A) indicates the probability that A will occur, P(B) indicates the probability that B will occur, P(B|A) indicates the probability that A will occur and B will occur, and P(A|B) indicates that B is It represents the probability that A will occur.

P(A|B) can be expressed as Equation 1 below.

If such Bayesian theorem is applied to category classification of digital content, it can be expressed as follows.

For example, P(science|input document) may be defined as the probability that the input document is classified as science.

If the article is "science news", then

For example, the category list can be classified as follows using about 22,000 sample data sets in which each category is classified. Examples of categories are as follows.

For example, a list of categories might look like business, technology, science, art, sports, world, politics, life, education, environment.

The above-described category list is only an example, and it is natural that the category list may be changed. It can also be used as a test set by inferring a category using the sample data set.

The sample data set is as shown in FIG. 3 . Even if a data set in which categories are classified, such as a sample data set, is used, the final category classification result using the above-described classification model may be different.

In step 225, the server 120 classifies each digital content based on the category classification result. That is, after configuring a category classification result based on the collected digital content analysis result, the digital content may be classified according to the classified category list.

To this end, the server 120 may vectorize each digital content (eg, a document). For example, the document may be vectorized using Word2Vec, Count Vectorization, TF/IDF, or the like. In an embodiment of the present invention, the count vectorization technique will be mainly described. However, it is not necessarily limited to the count vectorization technique, and it is natural that well-known techniques capable of vectorizing data may be applied.

I will explain assuming a sentence such as "They refuse to permit us to obtain the refuse permit".

The server 120 may extract each word from the corresponding sentence through the part-of-speech tagging technique. That is, if the part-of-speech tagging technique is applied in "They refuse to permit us to obtain the refuse permit", ('They', 'PRP'), ('refuse', 'VBP'), ('to', 'TO' ), ('permit', 'VB'), ('us', 'PRP'), ('to', 'TO'), ('obtain', 'VB'), ('the', 'DT' ), ('refuse', 'NN'), ('permit', 'NN').

In this way, if each word is separated by applying the part-of-speech tagging technique and only the word is extracted, it can be extracted as follows.

they, refuse, to, permit, us, to, obtain, the, refuse, permit

In this way, when each word is extracted, the server 120 may configure a feature data set for the document. In this case, the server 120 may configure the feature data set using the frequency of occurrence of the extracted word. That is, after deriving the frequency of appearance of each extracted word, the server 120 may exclude words having an appearance frequency less than the reference value from the feature data set.

That is, the feature data set may be configured only with words having an appearance frequency equal to or greater than a predetermined reference value. In an embodiment of the present invention, it is assumed that a feature data set is configured by excluding words having an appearance frequency of 10 times or less, but it is natural that the reference value can be flexibly changed during implementation.

For example, it is assumed that each word and the frequency of appearance of the word are, for example, as follows.

{'hundreds': 10, 'thousands': 9, 'people': 128, 'marched': 2, 'central': 14, 'london': 15, 'calling': 6, 'referendum': 11, ' mps': 19, 'search': 2, 'way': 45, 'brexit': 37, ??}

The server 120 excludes words having an appearance frequency of 10 or less ['hundreds', 'people', 'central', 'london', 'referendum', 'mps', 'way', 'brexit', ' You can configure a feature data set like campaign', 'say'].

Next, the server 120 constructs a feature matrix for the feature data set. The feature matrix may be composed of appearance frequencies of words included in the feature data set.

For example, suppose that the feature data set is [hundred, of, thousand, for, is] and the sentence is ('politics', 'Hundreds of thousands of people have marched for another E'). .

In this case, the feature matrix may be configured as [1, 2, 1, 1, 0].

The server 120 may classify a corresponding document according to a category using the feature matrix.

In step 230, the server 120 constructs a basic dictionary.

The basic dictionary may be configured by field (category), or may be configured regardless of category. In an embodiment of the present invention, a basic dictionary can be constructed by applying the formula of the n-gram model.

For the convenience of understanding and explanation, the n-gram will be briefly described.

P(next word|current word) represents the probability that the next word will appear after the current word. Also, P (next word|current word) may be a value obtained by dividing the frequency of occurrence of the word following the current word by the frequency of occurrence of the current word.

A method of constructing the basic dictionary will be described in more detail below.

For building a basic dictionary, digital content collected online can be used. In this case, the digital content may be collected by inputting a keyword into a search engine or may be collected based on information of an RSS feed.

The server 120 may extract text by converting digital content collected online. For example, since PDF has its own text conversion function, the server 120 can extract the entire text in the PDF using this. Also, in the case of an image file, text may be extracted through OCR conversion. Also, in the case of a subtitle file type such as SMI, text may be extracted using a regular expression, and in the case of an HTML type, text may be extracted using a regular expression. Also, in the case of a word file, text may be extracted using a conversion program.

For example, markup language-based data such as SMI or HTML can extract text excluding tags. For example, if the original source file is "<div class="lh s"><b>Reward quality writing.</b> When you spend time reading a story, a portion of your membership fee will go directly to its author. </div>" is assumed.

The server 120 may extract only the text after excluding the tags, such as "Reward quality writing. When you spend time reading a story, a portion of your membership fee will go directly to its author."

Then, the server 120 may extract the text by controlling unnecessary special characters, etc. in consideration of the characteristics of each language. For example, suppose English is spoken. In the case of English, since it consists of a to z or A to Z, the text may be extracted by removing all other characters.

After the texts for English are extracted, the server 120 removes proper nouns, comparatives, and auxiliary verbs in consideration of the characteristics of English, converts past tense into present tense, converts adverbs into adjectives, and converts verbs into prototypes, , the plural can be converted to singular, so that words can be extracted.

For example, in the case of English, proper nouns, comparatives, and auxiliary verbs are removed, the past tense is converted to the present tense, adverbs are converted into adjectives, verbs are prototypes, plurals are singular, etc. can be extracted.

For example, if the text is extracted as “Reward quality writing. When you spend time reading a story, a portion of your membership fee will go directly to its author.” You can extract words such as story, spend, membership, quality, to, of, a, reward, when, go, it, writing, reading, fee, portion, author, time, you".

The server 120 may arrange the extracted words in the order of word appearance frequency and number of appeared documents. (1) circumvallate, 335, 91, (2) circumvallate, 261, 91, etc., the frequency of occurrence of words and the number of documents can be arranged, respectively.

In addition, the server 120 may derive and add a ratio value for each word by using the frequency of occurrence of each word and the number of appeared documents. For example, the ratio value for each word may be calculated by dividing the word occurrence frequency by the number of appeared documents. (1) circumvallate, 335, 91, 3.6813186813 (2) circumvallate, 261, 91, 2.8681318681

Also, the server 120 may calculate a reference weight for each word and calculate a final weight using the calculated reference weight.

First, a method of calculating the reference weight will be described first.

For example, suppose the data set is as follows.

(1) circumvallate, 335, 91, 3.6813186813

(2) circumvallate, 261, 91, 2.8681318681

The server 120 may calculate the reference weight by dividing the target value by the frequency of appearance. Assuming the above two examples, when calculating the standard weight of word occurrence frequency, it is calculated as (335+261)/2, and the standard weight of the number of appeared documents is calculated as (91+91)/2, and The reference weight for the reference weight may be calculated as (3.6813186813+2.8681318681)/2. Accordingly, the reference weights for each of the word appearance frequency, the number of appeared documents, and the ratio value may be calculated as 298, 91, and 3.2747252747.

The server 120 may calculate the final weight by using the reference weight.

For example, the weight of the word appearance frequency may be calculated as (word appearance frequency/reference weight of word appearance frequency + word constant). Also, the weight of the number of appeared documents may be calculated as (the number of appeared documents/the reference weight of the number of appeared documents + the document constant).

Finally, the weight can be calculated as ((word appearance frequency weight + number of appeared documents weight) x precision constant).

Here, it goes without saying that the word constant, the document constant, and the precision constant may change depending on implementation.

A process of calculating weights using a data set will be described.

(1) circumvallate, 335, 91, 3.6813186813

(2) circumvallate, 261, 91, 2.8681318681

The final weight for (1) = ((335 / 298 + 31) + (91 / 91 + 36)) * 0.001, and the final weight for (2) is ((261/298 + 31) + ( It can be calculated as 91 / 91 + 36)) * 0.001.

That is, (1) circumvallate, 335, 91, 3.6813186813, 0.06912416107

(2) circumvallate, 261, 91, 2.8681318681, 0.06887583893

A final weight may be reflected in each word as shown in FIG.

The server 120 may remove duplicate words using the final weight and sort the basic word dictionary according to priority.

The priority can be set in the order of the final weight, the ratio value, the number of documents appearing, and the word appearance frequency, and the basic dictionary can be constructed by arranging the words in the order of the highest priority.

For example, a basic dictionary can be constructed such as "the, of, and, to, a, in, is, for, was, with, ??".

In step 235, the server 120 builds a learning order dictionary (learning level dictionary) using the basic dictionary.

The learning order dictionary (learning level dictionary) may be constructed using the basic dictionary. The learning sequence dictionary (learning level dictionary) may be constructed for each field or may be constructed independently of the field.

After extracting the top n (15,000) words from the basic dictionary, the server 120 may classify the words according to the size of the word set for each level, and then assign the levels in order. For example, assuming that the size of the word set for each level is 150, the top n words may be separated by 150 to generate 100 word sets for each level.

In step 240, the server 120 determines the learning order (content level) of the content using the learning division dictionary.

This will be described in more detail.

It is possible to determine the learning order (content level) of the corresponding content by deriving a coverage ratio for each word included in the learning turn dictionary (learning level dictionary) to what extent they are included in the words extracted from each content.

For example, the first-order dictionary (level) is "reward, direct, story, spend, membership", the second-order dictionary (level) is "portion, author, time, you, is, the", and the third-order dictionary ( level) is "fee, nice, attack, stuck, computer".

It is assumed that the coverage ratio for each level of specific content is derived as 100% for home appliances (level) with the 1st priority, 80% for the dictionary with the 2nd priority (level), and 20% for the dictionary with the 3rd priority (level). In this case, the level of the corresponding content may be set as the first priority.

However, if the matching rate with the highest coverage ratio is 85%, the server 120 may set the level of the corresponding content as the second lower priority.

In step 245, the server 120 stores the content for which the learning sequence is set as the learning content.

4 is a flowchart illustrating a method of recommending learning content according to an embodiment of the present invention, FIG. 5 is a diagram illustrating a diagnostic test set according to an embodiment of the present invention, and FIG. 6 is an embodiment of the present invention It is a diagram illustrating recommended content for each category according to an example, FIG. 7 is a diagram illustrating recommended content for all documents according to an embodiment of the present invention, and FIG. 8 is a view showing weights according to an embodiment of the present invention It is a diagram showing the result of sorting the recommendation order. A method of recommending learning content for an initially registered user will be described.

In step 410, the server 120 configures a diagnostic test set.

The diagnostic test set is a data set for ascertaining a user's learning level in order to recommend learning content for an initially registered user, and may be configured by extracting words from a dictionary for each level. 5 shows an example of a diagnostic test set.

Accordingly, the diagnostic test set may include a plurality of words from the lowest level to the highest level.

Also, in configuring the diagnostic test set, a diagnostic test set for a specific category may be configured by reflecting user information obtained from the user in advance.

Here, the user information may be, for example, a field in which the user is interested, a keyword, a purpose of learning, an identity, a gender, and the like.

In step 415 , the server 120 provides a diagnostic quiz to the user terminal 110 using the configured diagnostic test set.

Next, in step 420 , the server 120 obtains a diagnostic quiz result from the user terminal 110 .

In step 425 , the server 120 analyzes the diagnostic quiz result and recommends learning content to the user terminal 110 .

For example, the server 120 may analyze the diagnostic quiz result and determine the lowest level among the quizzes in which the user has presented an incorrect answer as the user learning level. The server 120 may recommend learning content according to the determined learning level to the user terminal 110 .

According to an embodiment of the present invention, in recommending learning content, feature values of documents are derived using Doc2Vec, then embedded in a vector space, and similar content can be derived and recommended.

In addition to Doc2Vec, TF/IDF, Cosine Similarity, and Euclidean Methods may be used.

According to an embodiment of the present invention, core data may be extracted from each content based on a topic modeling technique. For example, the topic modeling technique may be MG-LDA (Multi Grain Latent Dirichlet Allocation).

The MG-LDA can extract the nouns from each sentence after classifying the document from which the topic is to be extracted for each sentence. In this case, the MG-LDA can estimate the topic by estimating the probability that a word exists in a specific topic and the probability that the specific topic exists in the document as a combined probability by simultaneously extracting a global topic and a local topic.

Accordingly, in an embodiment of the present invention, it should be understood that core data (keyword) for each content is extracted and mapped, even if there is no separate explanation.

The server 120 may recommend content for each category by reflecting user information in recommending content suitable for the determined learning level, or may recommend content for all documents.

In addition, the server 120 may recommend the contents to the user terminal 110 by collecting the recommended contents, calculating the weights, and arranging the contents in the order of the weights.

The weight of the collected content may be calculated using the following equation.

Here, r represents the number of users who viewed the read_cnt document, l represents the number of users who liked the like_cnt document, s represents the number of users who shared the share_cnt document with others, and rt represents the total read_cnt , lt represents the total like_cnt, st represents the entire share_cnt, and similarity represents the document similarity. Document similarity may be derived using, for example, TF/IDF, Cosine Similarity, or Euclidean Method.

Contents extracted by category are exemplified in FIG. 6 , and contents extracted from the entire document are exemplified in FIG. 7 . If weights are reflected for the extracted contents as shown in FIGS. 6 and 7 , they may be sorted as shown in FIG. 8 . A final order of contents recommended to a user among the collected contents using the calculated weight may be determined as shown in FIG. 8 .

9 is a flowchart illustrating a content recommendation method according to another embodiment of the present invention, FIG. 10 is a diagram illustrating data for selecting similar learners according to an embodiment of the present invention, and FIG. 11 is an embodiment of the present invention It is a diagram illustrating a result of selecting similar learners according to an embodiment, FIG. 12 is a diagram illustrating a list of contents learned by the selected learners according to an embodiment of the present invention, and FIG. 13 is an embodiment of the present invention It is a diagram exemplifying the results of sorting by reflecting the weights according to .

In step 910 , the server 120 obtains learning state information from the user terminal 110 and aggregates learning state statistics using the obtained learning state information. user terminal 110

Here, the learning statistics may be aggregated into weekly growth statistics, monthly learning statistics, weekly word learning statistics, daily learning statistics, and the like. The server 120 may set a word that has passed all the tests for each review session among the learning contents provided to the user terminal 110 as a known word. In addition, the server 120 may start the review and pass the test even in the first stage, or set the wrong word even once during learning as the learning word. Also, the server 120 may set a word that has not started learning as an unknown word. The server 120 may receive a learning result for the learning content provided for each level from the user terminal 110, and use it to aggregate weekly growth statistics, monthly learning statistics, weekly word learning statistics, and daily learning statistics, respectively. .

In step 915, the server 120 may extract similar learners by reflecting the learning state statistics aggregation result.

For example, the server 120 derives the feature values of learners with Doc2Vec using ID, preferred category, preferred keyword, learning purpose, occupation, gender, learned content list, learning content list, and learned word list. After that, it can be embedded in the word space.

Next, the server 120 may select similar learners using TF/IDF, Cosine Similarity, Euclidean Method, or the like.

10 illustrates data for selecting similar learners. As shown in FIG. 10, after deriving the characteristic values of learners using ID, preferred category, preferred keyword, learning purpose, occupation, gender, learned content list, learning content list, and learned word list, using this Similar learners can be selected.

In step 920, the server 120 extracts the learning content learned by the selected similar learners. Here, the learning content may be content for each category or may be extracted for the entire document.

In step 925, the server 120 calculates a weight for the extracted content. Since the weight calculation is the same as described with reference to Equation 2 in FIG. 4 , a redundant description will be omitted.

In step 930 , the server 120 sequentially recommends the contents sorted by priority according to the weight to the user terminal 110 .

In this way, content learned by similar learners may be provided to the user terminal 110 .

11 exemplifies a result of selecting similar learners, and FIG. 12 exemplifies a list of contents learned by the selected learners. As shown in FIG. 12 , after extracting contents learned by similar learners, weights are calculated and sorted according to the weights, and as shown in FIG. 13 , a recommendation order can be determined and provided to the user terminal 110 .

14 is a flowchart illustrating a content recommendation method according to another embodiment of the present invention, FIG. 15 is a diagram illustrating a similar content recommendation result according to another embodiment of the present invention, and FIG. 16 is another embodiment of the present invention It is a diagram illustrating a result of classifying easy learning content reflecting coverage according to an embodiment, and FIG. 17 is a diagram illustrating a result of classifying difficult learning content reflecting coverage according to another embodiment of the present invention.

In step 1410, the server 120 extracts content similar to the learning content that the user is currently learning.

As described above, similar content can be extracted using TF/IDF, Cosine Similarity, and Euclidean Method after deriving feature values of documents using Doc2Vec and embedding them in a vector space.

In step 1415, the server 120 calculates coverage for the extracted similar content, respectively.

For example, the coverage may be calculated using Equation (3).

Here, m represents the number of matches between the words the learner knows and the content words, and t represents the total number of words in the content.

That is, the learning coverage refers to the number of words the user knows compared to the total number of words in the content.

In operation 1420, the server 120 classifies the easy content and the difficult content, respectively, by using the coverage for the extracted similar content.

In classifying content that is easy for a user and content that is difficult for a user, respectively, using coverage for similar content, user coverage may be used.

The user coverage is a result of deriving the coverage of words that the user knows from the test content. Accordingly, if the coverage of the extracted similar content is smaller than the user coverage, it is classified as easy content, and if the user coverage is smaller than the coverage of the extracted similar content, the similar content may be classified as difficult content.

That is, if user coverage < coverage of similar content, the corresponding similar content may be classified as easy content, and if user coverage > coverage of similar content, the corresponding similar content may be classified as difficult content.

In operation 1425 , the server 120 may recommend easy content or difficult content to the user terminal 110 using the classified result.

In this case, when the server 120 recommends a plurality of easy and difficult contents to the user terminal 110 , as described in FIG. 4 , the server 120 calculates a weight with reference to Equation 2 and then aligns the contents using the calculation. The recommendation order may be determined and recommendations may be made sequentially.

18 is a flowchart illustrating a content recommendation method according to an embodiment of the present invention.

In step 1810, the server 120 builds a learning content database.

Since this is the same as that described with reference to FIG. 2 , the overlapping description will be omitted.

In step 1815 , the server 120 provides the diagnostic quiz to the user terminal 110 .

In step 1820 , the user terminal 110 learns the diagnostic quiz provided from the server 120 , and then transmits the learning result to the server 120 .

In step 1825 , the server 120 recommends learning content in consideration of the learning result and provides it to the user terminal 110 .

Since this is the same as that described with reference to FIG. 4 , the overlapping description will be omitted.

In step 1830 , the user terminal 110 learns the recommended learning content from the server 120 .

After selecting the learning content, the user terminal 110 may learn the learning content by reading, reviewing, testing, repeating, or completing the original text.

In step 1835 , the user terminal 110 transmits learning activity information to the server 120 .

In step 1840 , the server 120 aggregates learning state statistics using the learning activity information received from the user terminal 110 .

Next, in step 1845 , the server 120 reflects the learning state statistics aggregation result to recommend learning content and provides it to the user terminal 110 . Since this is the same as that described with reference to FIGS. 9 and 14 , a redundant description will be omitted.

In step 1850 , the user terminal 110 learns the recommended learning content from the server 120 . That is, after selecting the recommended learning content, the user terminal 110 may learn the learning content by reading, reviewing, testing, repeating, or completing the original text.

In step 1855, the server 120 manages the learning schedule including the forgetting state update.

For example, the server 120 may demote the review session when the test result for the learning content is not reflected within a specified period of the user's next review session.

For example, if the second session on January 4 is 3 days after completing the first study on January 1, and if the test result is not reflected within January 5, the review session may be demoted to the first session. have. Even if the server 120 has not completed the test, if it is attempted, the server 120 may decide whether to demote again after granting a spare date.

In the case of being demoted according to the forgetting status update, the word learning status of the corresponding round may be rolled back. However, the server 120 may exclude words whose learning status is updated through other content when rolling back.

In step 1860 , the user terminal 110 transmits a learning statistics check and review request to the server 120 .

In step 1865 , the server 120 provides learning statistics-related information and review-related contents to the user terminal 110 according to the learning statistics confirmation and review request of the user terminal 110 . Accordingly, the user terminal 110 may check and review the learning statistics.

The process of checking the learning statistics may be frequently performed in the process of learning learning content, and may be performed in parallel by the user terminal 110 separately from the operation of the server 120 after step 1820 .

In addition, the user terminal 110 may separately perform the process of reviewing the learning content, and if the review is not performed within the review schedule specified by the server 120, it may be demoted when the forgetting state is updated.

Although described above with reference to the embodiments of the present invention, those of ordinary skill in the art can variously modify the present invention within the scope not departing from the spirit and scope of the present invention described in the claims below. and can be changed.

Claims (6)

collecting digital content online;
generating a category list by applying the collected digital content to a classification model, and classifying each digital content according to the generated category list;
After analyzing the digital content and extracting each word, a basic word dictionary is generated, a dictionary for each learning level is generated using the basic word dictionary, and a coverage ratio of the dictionary for each learning level of the digital content is derived, and the digital content determining a learning sequence of ; and
A method of recommending learning content, comprising the step of recommending learning content in consideration of a user learning result and a learning sequence of the digital content.
According to claim 1,
In generating the basic word dictionary,
analyzing the digital content and deriving a frequency of occurrence, a number of document appearances, and a ratio value for each extracted word; the ratio value is calculated by dividing the frequency of occurrence by the number of occurrences of the document;
calculating a weight of each word using the frequency of occurrence, the number of appearances in the document, and the ratio value; and
and generating the basic word dictionary by arranging the words in the order of the highest weight and extracting the top n words.
According to claim 1,
obtaining learning activity information of the recommended learning content from a user terminal;
After collecting learning status statistics using the learning activity information, selecting similar learners using the aggregation result - The similar learners have ID, preferred category, preferred keyword, learning purpose, occupation, gender, list of learned contents, After deriving the characteristic values of learners including the list of contents being learned and the list of words learned, embedding them in a word space to select similar learners; and
The method of recommending learning content further comprising the step of extracting and recommending the learning content learned by the similar learner.
According to claim 1,
The step of recommending the content,
calculating weights for each of the extracted learning contents;
sorting the learning content using the weight; and
and recommending the learning content sequentially according to the sorted order.
According to claim 1,
selecting content similar to the learning content being studied by the user;
calculating coverage for the similar content, respectively;
classifying difficult learning content and easy learning content using the coverage;
Further comprising the step of providing the classified difficult learning content and easy learning content to the user terminal in order,
The learning content recommendation method, characterized in that the coverage is calculated in consideration of whether a word known by the user matches a word included in the similar content.
According to claim 1,
Updating the forgetting state, but further comprising the step of demoting the review cycle or rolling back the learning state of the level when the user does not test the recommended learning content within a specified review period.

Images