KR102276353B1 - System and method for recommending movie based on embedding learning using fuzzy-ahp - Google Patents

Abstract

The present invention relates to a system and method for recommending a movie based on a word-to-vector (Word2Vec) learning using a fuzzy-AHP. In particular, it relates to a system and method for recommending a movie based on a word-to-vector learning using a fuzzy-AHP that use the fuzzy-AHP to classify user's favorite movies, obtain movie-related text data by scraping movie-related review data and news articles on Twitter or portal sites, performs word-to-vector learning using the movie-related text data, and generate recommended movies and preference data by matching user's favorite movie classification data and a word-to-vector learning result. An object of the present invention is to objectively and accurately recommend user's favorite movies.

Description

SYSTEM AND METHOD FOR RECOMMENDING MOVIE BASED ON EMBEDDING LEARNING USING FUZZY-AHP

The present invention relates to a word-to-vector (Word2Vec) learning-based movie recommendation system and method using a fuzzy hierarchical analysis process (FUZZY-AHP). Obtain movie-related text data by scraping movie-related review data and news articles on the site, learn word-to-vector using movie-related text data, and learn user's preferred movie classification data and word-to-vector learning results It relates to a word-to-vector learning-based movie recommendation system and method using a fuzzy hierarchical analysis process for generating recommended movies and preference data by matching .

Recently, with the onset of the 5G era, data is increasing exponentially. This results in the creation of huge amounts of data every minute of every day, whether it's content clicks or like button presses worldwide. Twitter has 470,000 retweets every minute, and YouTube has more than 4 million views per minute. In addition, on the portal site, various media companies are quickly reporting articles by field. If these various data can be effectively utilized, the performance of intelligent services applied in various fields will be improved.

The intelligent recommendation service that is being applied in the field of life and culture stands out in the fields of books, movies, and music.

However, in order for this intelligent recommendation service to show high performance, a large amount of training data is required, and there are problems that the inference technique needs to be more accurate. In addition, there is a problem in that it is difficult to provide an accurate recommendation service because the degree of preference for each user is subjective and uncertain.

F. T. S. Chan & N. Kumar. (2007). Global Supplier Development Considering Risk Factors using Fuzzy Extended AHP-based Approach. Original Research Article Omega, 35(4), 417-431. C. Lin & P. J. Hsieh. (2004). A Fuzzy Decision Support System for Strategic Portfolio Management. Decision Support Systems, 38 (2004), 383-398.

Accordingly, the present invention has been made to solve the above problems, and an object of the present invention is to objectively and accurately recommend a user's preferred movie, a word-to-vector learning-based movie recommendation system and method using a fuzzy layer analysis process is to provide

In order to achieve the above object, a word-to-vector learning-based movie recommendation system using a fuzzy hierarchical analysis process according to an embodiment of the present invention receives preference data of a movie preferred by a user and performs a fuzzy hierarchical analysis process (Fuzzy-AHP). ), a user information collecting unit configured to classify a user's favorite movie using the user information and to store user information and user's preferred movie classification data in the database 200; a learning data collection unit configured to obtain movie-related text data by scraping movie-related review data and news articles of a social network service (SNS) or portal site; an embedding learning unit configured to read the movie-related text data from the learning data collection unit, learn word-to-vector, and store the word-to-vector learning result in the database; and reading the user's preferred movie classification data and word-to-vector learning result data from the database, matching the two data to generate a recommended movie title and corresponding preference data, and then generating the recommended movie title and corresponding information and a movie recommendation unit configured to provide preference data to the user interface and to store the preference data in the database.

In the word-to-vector learning-based movie recommendation system using the fuzzy hierarchical analysis process according to the embodiment, the preference may be generated using the following [Equation 1].

[Equation 1]

는 선호도를 나타내고,

는 우선순위가 높은 종합 가중치를 나타내며,

는 우선순위가 높은 상대적 가중치를 나타내며,

는 세부 기준의 가중치를 보정하기 위한 실수를 나타냄][here

represents the preference,

represents the high-priority aggregate weight,

represents the relative weight with high priority,

represents a real number for correcting the weight of the detailed criteria]

In order to achieve the above object, a word-to-vector learning-based movie recommendation method using a fuzzy layer analysis process according to another embodiment of the present invention is a fuzzy layer analysis process in which a user information collecting unit receives preference data of a movie preferred by a user. classifying a user's favorite movie using (Fuzzy-AHP), and storing user information and user's favorite movie classification data in a database; acquiring movie-related text data by scraping, by a learning data collection unit, movie-related review data and news articles of a social network service (SNS) or portal site; The embedding learning unit reads the movie-related text data obtained previously, learns word-to-vector, and stores the word-to-vector learning result in the database; and a movie recommendation unit reads the user's preferred movie classification data and word-to-vector learning result data from the database, matches the two data to generate a recommended movie title and corresponding preference data, and then selects the recommended movie title and and providing the corresponding preference data to the user interface and storing the data in the database.

According to the word-to-vector learning-based movie recommendation system and method using the fuzzy hierarchical analysis process according to the embodiments of the present invention, the user's preferred movie is selected using the fuzzy hierarchical analysis process by receiving the user's preference data of the preferred movie. classify, and store user information and user's favorite movie classification data in a database; Scraping movie-related review data and news articles on SNS or portal sites to obtain movie-related text data; reading the previously obtained movie-related text data, learning word-to-vector, and storing the word-to-vector learning result in the database; After reading the user's preference movie classification data and word-to-vector learning result data from the database, and matching the two data to generate a recommended movie title and corresponding preference data, the recommended movie title and corresponding preference data is provided on the user interface and configured to be stored in the database, so that there is an excellent effect of objectively and accurately recommending a movie that the user prefers.

1 is a block diagram of a movie recommendation system based on word-to-vector learning using a fuzzy layer analysis process according to an embodiment of the present invention.
2 is a flowchart for explaining a word-to-vector learning-based movie recommendation method using a fuzzy hierarchical analysis process according to an embodiment of the present invention.
FIG. 3 is a diagram illustrating a hierarchical structure of the user information collecting unit of FIG. 1 .

In describing the embodiments of the present invention, if it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, the terms described below are terms defined in consideration of functions in the present invention, which may vary according to intentions or customs of users and operators. Therefore, the definition should be made based on the content throughout this specification. The terminology used in the detailed description is for the purpose of describing embodiments of the present invention only, and should in no way be construed as limiting. Unless explicitly used otherwise, expressions in the singular include the meaning of the plural. In this description, expressions such as “comprising” or “comprising” are intended to indicate certain features, numbers, steps, acts, elements, some or a combination thereof, one or more other than those described. It should not be construed as excluding the presence or possibility of other features, numbers, steps, acts, elements, or any part or combination thereof.

In each system shown in the figures, elements in some instances may each have the same reference number or a different reference number, suggesting that the represented element may be different or similar. However, elements may have different implementations and work with some or all of the systems shown or described herein. The various elements shown in the drawings may be the same or different. Which one is referred to as the first element and which is referred to as the second element is arbitrary.

In this specification, when any one component 'transmits', 'transfers' or 'provides' data or signal to another component, one component directly transmits data or signal to another component, as well as, and transmitting data or signals to the other component via at least one further component.

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

1 is a block diagram of a movie recommendation system based on word-to-vector learning using a fuzzy layer analysis process according to an embodiment of the present invention.

A movie recommendation system based on word-to-vector learning using a fuzzy layer analysis process according to an embodiment of the present invention, as shown in FIG. 1 , includes a user information collection unit 100 , a learning data collection unit 110 , and an embedding learning unit. 120 , and a movie recommendation unit 300 .

The user information collection unit 100 receives the user's preference data of the preferred movie through the user terminal, classifies the user's preferred movie using a fuzzy hierarchical analysis process (Fuzzy-AHP), and uses the user information and the user's preferred movie. It serves to store classification data in the database 200 .

The user information collection unit 100 may determine the importance of each level element through the hierarchical structure as shown in FIG. 3 , and may know the type of movie preferred by the user through normalization of the progressed result.

Fuzzy hierarchical analysis process (Fuzzy-AHP) is a methodology for making accurate decisions by mathematically ranking inaccuracies inherent in content or items of subjective and uncertain information. Through the fuzzy hierarchical analysis process, it is possible to objectively express information or preferences that are judged emotionally or subjectively evaluated. In the present invention, the user's preferred movies are configured in a hierarchical structure to make a comprehensive decision on the user's preferences.

The learning data collection unit 110 acquires movie-related text data by scraping movie-related review data and news articles from SNS (Social Network Service) or portal sites such as Facebook, Instagram, and Twitter. plays a role

Scraping refers to extracting necessary semi-structured data from a web page, and mainly uses a library capable of parsing from a web page. Since data on the web is usually in the form of a web page, necessary data can be obtained by scraping or crawling. In the present invention, it is possible to increase the movie recommendation performance by maximizing web data through scraping.

The embedding learning unit 120 reads the movie-related text data scraped by the learning data collection unit 110 , learns word-to-vector, and stores the word-to-vector learning result in the database 200 .

The word-to-vector learning technique is a learning-based natural language processing methodology that mathematically represents the similarity between words by learning words from movie-related text data and creating tokens. In the present invention, it is possible to accurately classify the scraped movie-related text data through the word-to-vector learning technique.

The movie recommendation unit 300 reads the user's preference movie classification data and word-to-vector learning result data from the database 200 and matches the two data to generate a recommended movie title and corresponding preference data, and then recommends the movie. It serves to provide the movie title and corresponding preference data to the user interface and also to store it in the database 200 .

The preference is generated using the following [Equation 1].

[Equation 1]

는 선호도를 나타내고,

는 우선순위가 높은 종합 가중치를 나타내며,

는 우선순위가 높은 상대적 가중치를 나타내며,

는 세부 기준의 가중치를 보정하기 위한 실수를 나타냄][here

represents the preference,

represents the high-priority aggregate weight,

represents the relative weight with high priority,

represents a real number for correcting the weight of the detailed criteria]

A movie recommendation method using a word-to-vector learning-based movie recommendation system using a fuzzy hierarchical analysis process according to an embodiment of the present invention configured as described above will be described with reference to the drawings.

2 is a flowchart for explaining a method for recommending a movie based on embedding learning using a fuzzy layer analysis process according to an embodiment of the present invention, where S means a step.

First, the user information collection unit 100 receives the user's preferred movie preference data through the user terminal and classifies the user's preferred movie using a fuzzy hierarchical analysis process (S10), and the user information and the user's preferred movie The classification data is stored in the database 200 (S20).

Then, the learning data collection unit 110 scrapes the review data and news articles related to the movie of the SNS or portal site to obtain the movie-related text data (S30).

Next, the embedding learning unit 120 reads the movie-related text data obtained from the learning data collection unit 110 to learn word-to-vector, and stores the word-to-vector learning result in the database 200 ( S40 ).

Next, the movie recommendation unit 300 reads the user's favorite movie classification data and word-to-vector learning result data from the database 200 and matches the two data to generate a recommended movie title and corresponding preference data, The recommended movie title and corresponding preference data are provided to the user interface and stored in the database 200 (S50).

According to the word-to-vector learning-based movie recommendation system and method using a fuzzy hierarchical analysis process according to an embodiment of the present invention, the user's preferred movie is classified using the fuzzy hierarchical analysis process by receiving the user's preference data of the preferred movie. and store user information and user's favorite movie classification data in a database; Scraping movie-related review data and news articles on SNS or portal sites to obtain movie-related text data; reading the previously obtained movie-related text data, learning word-to-vector, and storing the word-to-vector learning result in the database; After reading the user's preference movie classification data and word-to-vector learning result data from the database, and matching the two data to generate a recommended movie title and corresponding preference data, the recommended movie title and corresponding preference data is provided on the user interface and configured to be stored in the database, so that a user's preferred movie can be objectively and accurately recommended.

In the drawings and specification, the best embodiment is disclosed, and specific terms are used, but these are used only for the purpose of describing the embodiments of the present invention, and are used to limit the meaning or limit the scope of the present invention described in the claims it didn't happen Therefore, it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

100: user information collection unit
110: learning data collection unit
120: embedding learning unit
200: database
300: movie recommendation book

Claims (4)

The user's preference data of the preferred movie is received through the user terminal, the user's preferred movie is classified using a fuzzy hierarchical analysis process (Fuzzy-AHP), and the user information and the user's preferred movie classification data are stored in the database 200. a user information collection unit 100 configured to store;
a learning data collection unit 110 configured to obtain movie-related text data by scraping movie-related review data and news articles of a social network service (SNS) or portal site;
The movie-related text data scraped by the learning data collection unit is read and word-to-vector learns to classify the movie-related text data, and the classified movie-related text data is stored in the database as a word-to-vector learning result. an embedding learning unit 120 configured to store; and
After reading the user's preference movie classification data and word-to-vector learning result data from the database, and matching the two data to generate the recommended movie title and corresponding preference data, the recommended movie title and the corresponding preference and a movie recommendation unit 300 capable of objectively recommending a movie preferred by a user by providing data to the user interface and storing the data in the database;
The word-to-vector learning is a learning-based natural language processing methodology that mathematically represents the similarity between words by learning the words of movie-related text data and creating tokens,
The preference is generated using the following [Equation 1], a word-to-vector learning-based movie recommendation system using a fuzzy hierarchical analysis process.

[Equation 1]

[here

represents the preference,

represents the aggregate weight with high priority

represents the relative weight with high priority,

represents a real number for correcting the weight of the detailed criteria]
delete A movie recommendation method using the word-to-vector learning-based movie recommendation system using the fuzzy hierarchical analysis process according to claim 1,
The user information collection unit 100 receives the user's preference data of the preferred movie, classifies the user's preferred movie using a fuzzy hierarchical analysis process (Fuzzy-AHP), and stores user information and user's preferred movie classification data in a database. storing in (200);
obtaining, by the learning data collection unit 110, movie-related text data by scraping movie-related review data and news articles of a social network service (SNS) or portal site;
the embedding learning unit 120 reading the movie-related text data from the learning data collection unit, learning word-to-vector, and storing the word-to-vector learning result in the database; and
After the movie recommendation unit 300 reads the user's favorite movie classification data and word-to-vector learning result data from the database, matches the two data to generate a recommended movie title and corresponding preference data, the recommended movie A movie recommendation method based on word-to-vector learning using a fuzzy hierarchical analysis process, comprising: providing a title and preference data corresponding thereto to a user interface and storing the title in the database.
delete

Images