KR101910424B1 - Method for movie ratings prediction using sentiment analysis of movie tags, recording medium and device for performing the method - Google Patents

Abstract

A method for predicting a movie score using emotion analysis of a tag comprises: constructing a neighboring group to use a user's rating for a movie to be evaluated as data to be predicted; Extracting tag information and rating information for the movie to be evaluated from the neighboring group; Performing emotional analysis using the tag information; And predicting a rating for the movie to be evaluated by the user using the emotional analysis result and the rating information. Accordingly, the emotional analysis using the tag information and the rating information can predict the movie rating that the user has not yet evaluated.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for predicting movie scores using emotion analysis of a tag,

[0001] The present invention relates to a movie rating prediction method using tag emotion analysis, a recording medium and an apparatus for performing the same, and more particularly, to a movie rating prediction method using tag emotion analysis through a collaborative filtering technique, And a recording medium and an apparatus for performing the same.

In today's information-rich world, users have difficulty in judging and selecting a lot of information by themselves. Accordingly, the recommendation system is evaluated as an important technique for presenting information necessary for the user in the modern information which is pouring various information.

The recommendation system allows users to get less information and to get the information they need, and in the case of a company that provides a recommendation system, it makes a profit by selling the product.

Recommendation system can be roughly divided into two. There are Collaborative Filtering and Content-based Filtering.

In the movie recommendation system such as MovieLens, there are not many cases where the tag information inputted by the user is used as the new information. Even if the tag information is used, it is used only for the relevance measurement, not the application of the tag information.

Non-Patent Document 1 of the prior art document uses a collaborative filtering technique and suggests the best weighting results through a genetic algorithm while applying different weight values. The non-patent document 2 of the prior art document includes the tag information newly added in the MovieLens to calculate the inter-user association.

However, in the case of this conventional technology, the tag information is used as it is for judging similarity among another users.

 A genetic algorithm solution to the collaborative filtering problem-Yilmaz Ar, Erkan Bostanci Ankara  Improving Recommendation Accuracy Based on Item-Specific Tag Preferences-FATIH GEDIKLI and DIETMAR JANNACH

SUMMARY OF THE INVENTION Accordingly, the present invention has been made in view of the above problems, and it is an object of the present invention to provide a method for predicting a movie score using emotion analysis of a tag.

Another object of the present invention is to provide a recording medium on which a computer program for performing a movie rating prediction method using emotion analysis of the tag is recorded.

It is still another object of the present invention to provide an apparatus for performing a method for predicting a movie score using emotion analysis of the tag.

According to an embodiment of the present invention, there is provided a method for predicting a movie rating using tag emotional analysis, the method comprising: constructing a neighborhood group for utilizing a user's rating for a movie to be evaluated as data to be predicted; Extracting tag information and rating information for the movie to be evaluated from the neighboring group; Performing emotional analysis using the tag information; And predicting a rating for the movie to be evaluated by the user using the emotional analysis result and the rating information.

In the embodiment of the present invention, the step of constructing a neighboring group for use as data for predicting a user's rating on the movie to be evaluated may be performed by k-means clustering, You can configure a collection of other users.

In the embodiment of the present invention, the step of performing emotional analysis using the tag information may use an emotional analyzer.

In an embodiment of the present invention, the result of emotional analysis through the emotional analyzer may be output as each emotion value of positive, negative, and objective.

In an embodiment of the invention, the sum of the positive, negative and objective emotion values may be one.

In the embodiment of the present invention, the step of predicting the evaluation of the movie to be evaluated by the user using the result of the emotional analysis and the rating information may include: calculating a value of 5 * (positive emotion value - negative emotion value) 1: 1 ratio with the rating information.

According to another embodiment of the present invention, a computer-readable storage medium stores a computer program for performing a movie rating prediction method using emotion analysis of a tag.

According to another aspect of the present invention, there is provided an apparatus for predicting a movie score using a sensibility analysis of a tag, the apparatus comprising: Group organization; An information extraction unit for extracting tag information and rating information for the movie to be evaluated from the neighboring group; An emotional analysis unit for performing emotional analysis using the tag information; And a rating predicting unit for predicting a rating of a movie to be evaluated by the user using the result of the sensitivity analysis and the rating information.

In an embodiment of the present invention, the emotion analyzer may output emotion values of positive, negative, and objective with a sum of 1 using an emotion analyzer.

In the embodiment of the present invention, the evaluation predicting unit may derive a value of 5 * (positive emotion value - negative emotion value) in a 1: 1 ratio with rating information.

According to the method for predicting the movie score using the emotional analysis of the tag, the emotional analysis of tag information, which is a word related to the movie created by the users, and the emotional analysis of the movie, Can be predicted. Accordingly, in the present invention, it is possible to predict a more accurate score through emotional analysis in consideration of a case in which the movie has similar appeal but the used word is different.

In addition, in the information age, users have difficulty in judging the information by themselves. The recommendation system allows users to get less information and to get the information they need, and in the case of a company that provides a recommendation system, it makes a profit by selling the product. The movie recommendation system according to the present invention can provide a benefit to both the user and the enterprise by recommending a movie satisfactory to the user.

1 is a conceptual diagram of movie rating prediction using emotion analysis of a tag according to the present invention.
2 is a block diagram of an apparatus for predicting a movie score using emotion analysis of a tag according to an embodiment of the present invention.
3 is a block diagram of an apparatus for predicting a movie score using emotion analysis of a tag according to another embodiment of the present invention.
FIG. 4 is a flowchart illustrating a method of predicting a movie score using emotion analysis of a tag according to an exemplary embodiment of the present invention.

The following detailed description of the invention refers to the accompanying drawings, which illustrate, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that the various embodiments of the present invention are different, but need not be mutually exclusive. For example, certain features, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the invention in connection with an embodiment. It is also to be understood that the position or arrangement of the individual components within each disclosed embodiment may be varied without departing from the spirit and scope of the invention. The following detailed description is, therefore, not to be taken in a limiting sense, and the scope of the present invention is to be limited only by the appended claims, along with the full scope of equivalents to which such claims are entitled, if properly explained. In the drawings, like reference numerals refer to the same or similar functions throughout the several views.

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

1 is a conceptual diagram of movie rating prediction using emotion analysis of a tag according to the present invention.

The present invention uses the data of a movie recommendation service as data of a recommendation system and does not simply use tag information directly input by a newly added user recently to measure relevance, As a rating information. For example, the movie recommendation service for extracting the data may be a MovieLens.

Therefore, it is possible to predict the rating of a movie that has not yet been evaluated by the user by combining the conventional rating and the emotion analysis result using the tag information. In the present invention, a tag is a word associated with a movie created by a user directly on a movie.

1 and 2, a movie rating prediction apparatus 10 (hereinafter, referred to as a device) using emotion analysis of a tag according to an embodiment of the present invention includes a neighboring group constructing unit 110, an information extracting unit 130, An analysis unit 150 and a rating prediction unit 170.

The device 10 of the present invention may be implemented with software (application) for performing movie rating prediction using emotion analysis of a tag, and the neighbor group constructing unit 110, the information extracting unit 130, The configuration of the emotional analysis unit 150 and the evaluation unit 170 may be controlled by software for performing movie rating prediction using emotion analysis of the tag executed in the device 10. [

The device 10 may be a separate terminal or some module of the terminal. The configuration of the neighboring group constructing unit 110, the information extracting unit 130, the emotional analyzing unit 150, and the predicting unit 170 may be an integrated module or may be composed of one or more modules. have. However, conversely, each configuration may be a separate module.

The device 10 may be mobile or stationary. The device 10 may be in the form of a server or an engine and may be a device, an apparatus, a terminal, a user equipment (UE), a mobile station (MS) a wireless device, a handheld device, and the like.

The device 10 may execute or produce various software based on an operating system (OS), i.e., a system. The operating system is a system program for allowing software to use the hardware of a device. The operating system includes a mobile computer operating system such as Android OS, iOS, Windows Mobile OS, Sea OS, Symbian OS, Blackberry OS, MAC, AIX, and HP-UX.

The neighboring group constructing unit 110 constitutes a neighboring group for utilizing the user's rating of the movie to be evaluated as data to be predicted. The neighboring group can utilize data such as text, SNS text, and the like displayed on the movie related sites of users in order to constitute a group having a similar tendency to the user who predicts the rating.

For example, the neighbors can construct a set of other users with a similar tendency to the user through k-means clustering.

Non-hierarchical clustering (k-means clustering) examines all possible ways to divide n objects into g clusters to form optimized clusters.

This is a useful algorithm for finding similar clusters when there are data points, and arbitrarily assigns k centroids when there are data points to calculate each data point and centroid distance. Then assign the points to the nearest centroid, and average centroids assigned to the same cluster to move the centroid.

Then again calculate the distance between all the points and the centroids and assign the points to the nearest centroid again. This process is repeated to stop the algorithm when there are no more changes in the points assigned to centroid.

Non-hierarchical cluster analysis has the advantage that it can find a meaningful data structure and can be applied to various types of data without prior knowledge about the internal structure of given data.

Through this, a neighboring group is formed to utilize the user rating of the movie to be evaluated as data to be predicted. Accordingly, since the data of the neighboring group composed of users having a tendency similar to the user is utilized, the accuracy of the user's rating prediction can be improved.

The information extracting unit 130 extracts tag information and rating information for the movie to be evaluated from the neighboring group.

A tag is a word or a keyword that is given for use in searching for information, and is also called a tag. The necessity of tag is increased according to the appearance of multimedia information such as a photograph or a video among Internet information. Unlike the information written in text, these multimedia information can not be searched with keywords. Therefore, the multimedia information can be searched by hierarchical classification (Taxonomy) for information search.

However, for example, photographs taken while traveling can be classified as 'travel', but they can be classified according to the city where they are taken, or classified according to the event type of a photograph, so that a single piece of information belongs to a plurality of fields. In this case, hierarchical classification becomes impossible, and the classification system becomes enormous and it takes much time to search. So we use it to search by assigning tags made up of several texts.

That is, in a reality where it is practically impossible to maintain a standardized classification system for all users, information is searched and classified into a set of tags by arranging all words (tags) suitable for specific information. As a central feature of Web 2.0, it is called Folksonomy because it is classified as not centralized by users' participation.

Social networks such as Twitter, Facebook, and Instagrams use tags to search for articles of interest, and tags are used as keywords that represent a kind of common interest. Tags are a kind of community role, and the use of tags to upload and share photos on social networks is increasing. In addition, Naver's Pholar, launched in 2015, uses a system that follows tags that correspond to hobbies, rather than following friends.

In the present invention, the tag information is related to a movie directly written by the users belonging to the neighbor group on the movie. For example, tag information can be extracted using data of a movie lens (MovieLens), which is one of movie recommendation services.

Also, the information extracting unit 130 extracts rating information on the movie to be evaluated from the neighboring group. The rating information may have a range from 0 to 5, for example.

The emotional analysis unit 150 performs emotional analysis using the tag information. It is meaningful that the present invention utilizes the tag information directly input by the user to predict the result of using the emotional analysis rather than simply use it to measure the relevance.

There are not many cases where the user has directly used the tag information that is input by the user as the new information on the movie recommendation site such as MovieLens. Even if the tag information is used, it is used only for the correlation measurement, not the application of the tag information.

As described above, in the conventional case, the tag information is used to determine the similarity among other users. However, in the present invention, emotional analysis is performed by using the tag information. It is possible to make a predictive rating through.

The emotional analysis unit 150 may perform emotional analysis using the tag information using an emotional analyzer. For example, the emotion analyzer can use SentiWordNet.

SentiWordNet recently used emotional dictionary, which is one of the representative methods of emotional analysis using data created by users on social networks such as Twitter, as interest in opinion mining is rapidly increasing.

When emotion analysis is performed using SentiWordNet, it is output as three emotion values: positive, negative, and objective. The sum of the three emotion values is 1, and the range of each emotion value is 0 to 1.

The rating estimating unit 170 estimates a rating for the movie to be evaluated by the user using the emotion analysis result of the emotion analyzing unit 150 and the rating information output from the information extracting unit 130 do.

For example, the rating estimator 170 may derive a value of 5 * (positive emotion value - negative emotion value) by integrating the rating information with the rating information at a ratio of 1: 1. In the present invention, when the data of MovieLens is used, the rating is in the range of 0 to 5 or less, and each maximum value of positive and negative sensibility values is 1, so that it can be normalized by multiplying by 5.

In another embodiment, a weight may be assigned to the rating formula, and an EJC (Extended Jaccard Coefficient) may be used in the case of the weight. In addition, the rating formula may be changed depending on the data used and the characteristics of the emotion analyzer.

Accordingly, emotional analysis is performed on tag information, which is a word related to a movie created by users, and ratings of a movie that the user has not yet evaluated through emotional analysis and rating information can be predicted.

Referring to FIG. 3, the movie rating predicting apparatus 30 using emotion analysis of a tag according to another embodiment of the present invention may have the same structure as that of the device 10 of FIG. 2 except that the information extracting unit is separated. Can proceed. Therefore, the same components as those of the device 10 of FIG. 2 are denoted by the same reference numerals, and repeated descriptions are omitted.

The apparatus 30 includes a tag information extracting unit 330 for extracting tag information about a movie to be evaluated from the neighboring group and rating information about ratings of the movie to be evaluated from the neighboring group, And an extracting unit 340 separately.

The tag information extracted from the tag information extracting unit 330 is transmitted to the rating predicting unit 170 through the emotional analysis unit 150. That is, in the present invention, the tag information is used as a kind of second rating information through emotion analysis rather than being used as it is.

On the other hand, the rating information extracted from the rating information extracting unit 340 is transmitted to the rating predicting unit 170 and used as information of rating prediction.

Accordingly, emotional analysis is performed on tag information, which is a word related to a movie created by users, and ratings of a movie that the user has not yet evaluated through emotional analysis and rating information can be predicted.

In addition, the recommendation system allows users to get less information and to obtain the necessary information, and in the case of a company that provides a recommendation system, it makes profit through the sale of goods. The movie recommendation system according to the present invention can provide a benefit to both the user and the enterprise by recommending a movie satisfactory to the user.

FIG. 4 is a flowchart illustrating a method of predicting a movie score using emotion analysis of a tag according to an exemplary embodiment of the present invention.

The movie rating prediction method using the sensitivity analysis of the tag according to the present embodiment can be performed in substantially the same configuration as the device 10 of FIG. 2 or the device 30 of FIG. Therefore, the same components as those of the device 10 of Fig. 2 or the device 30 of Fig. 3 are denoted by the same reference numerals, and repeated descriptions are omitted.

Further, it can be executed by software (application) for performing movie rating prediction using emotion analysis of the tag in this embodiment.

Referring to FIG. 4, the movie rating prediction method using the emotion analysis of the tag according to the embodiment of the present invention constructs a neighboring group for use as data for predicting a user's rating on a movie to be evaluated (step S10).

The neighboring group can utilize data such as text, SNS text, and the like displayed on the movie related sites of users in order to constitute a group having a similar tendency to the user who predicts the rating.

For example, the neighbors can construct a set of other users with a similar tendency to the user through k-means clustering.

Through this, a neighboring group is formed to utilize the user rating of the movie to be evaluated as data to be predicted. Accordingly, the accuracy of the user's rating prediction can be improved by utilizing data of a neighboring group composed of users having a similar tendency to the user.

If the neighboring group is constructed, tag information and rating information for the movie to be evaluated are extracted from the neighboring group (step S30).

In the present invention, the tag information is related to a movie directly written by the users belonging to the neighbor group on the movie. For example, tag information can be extracted using data of a movie lens (MovieLens), which is one of movie recommendation services. The rating information may have a range from 0 to 5, for example.

When the tag information is extracted, emotional analysis is performed using the tag information (step S50). In step S50, emotional analysis can be performed using the tag information using the emotion analyzer. For example, the emotion analyzer can use SentiWordNet.

It is meaningful that the present invention utilizes the tag information directly input by the user to predict the result of using the emotional analysis rather than merely to measure the relevance.

There are not many cases where the user has directly used the tag information that is input by the user as the new information on the movie recommendation site such as MovieLens. Even if the tag information is used, it is used only for the correlation measurement, not the application of the tag information.

In the present invention, tag information is used to determine similarity among other users. However, in the present invention, a similar feeling is felt in a movie by performing emotional analysis of a tag, can do.

When the result of the emotional analysis is derived, the score of the movie to be evaluated by the user is predicted using the emotional analysis result and the rating information (step S70).

For example, a value of 5 * (positive emotion value - negative emotion value) can be derived by integrating the rating information with a rating ratio of 1: 1. In the present invention, when the data of MovieLens is used, the rating is in the range of 0 to 5 or less, and each maximum value of positive and negative sensibility values is 1, so that it can be normalized by multiplying by 5.

In another embodiment, a weight may be assigned to the rating formula, and an EJC (Extended Jaccard Coefficient) may be used for the weight. In addition, the rating formula may be changed depending on the data used and the characteristics of the emotion analyzer.

Accordingly, emotional analysis is performed on tag information, which is a word related to a movie created by users, and ratings of a movie that the user has not yet evaluated through emotional analysis and rating information can be predicted.

In addition, the recommendation system allows users to get less information and to obtain the necessary information, and in the case of a company that provides a recommendation system, it makes profit through the sale of goods. The movie recommendation system according to the present invention can provide a benefit to both the user and the enterprise by recommending a movie satisfactory to the user.

The movie rating prediction method using the emotion analysis of the tag may be implemented in an application or in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program commands, data files, data structures, and the like, alone or in combination.

The program instructions recorded on the computer-readable recording medium may be ones that are specially designed and configured for the present invention and are known and available to those skilled in the art of computer software.

Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like.

Examples of program instructions include machine language code such as those generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules for performing the processing according to the present invention, and vice versa.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the present invention as defined by the following claims. You will understand.

In today's world, where information is pouring out, the recommendation system is an important technology that can provide users with necessary information. The recommendation system allows users to get less information and to get the information they need, and in the case of a company that provides a recommendation system, it makes a profit by selling the product.

The Movie Recommendation System can benefit both users and businesses by recommending a movie that the user is satisfied with. It is expected that the present invention can be applied to a recommendation system of a movie or other personalization field by a movie recommendation system technique.

10, 30: Movie rating system using emotion analysis of tag
110: Neighborhood group component
130:
330: Tag information extracting unit
350: rating information extracting unit
150: Emotional Analysis Department
170:

Claims (10)

A method for predicting a movie score using emotion analysis of a tag, the method comprising:
Constructing a neighboring group for use as data for predicting a user's rating on a movie to be evaluated by the movie rating prediction device;
Extracting tag information and rating information for the movie to be evaluated from the neighboring group by the movie rating prediction device;
Performing emotional analysis using the tag information by the movie rating prediction device; And
And estimating a rating for the movie to be evaluated by the user using the result of the emotional analysis and the rating information by the movie rating prediction device,
Wherein the rating information is a constant value within a predetermined constant range,
Wherein the step of performing emotional analysis using the tag information by the movie rating prediction apparatus comprises:
Outputting emotional analysis results as positive, negative, and objective emotional values through an emotional analyzer using an emotional analyzer using data created by users on a social network,
The sum of the positive, negative and objective emotional values is 1,
Wherein the step of predicting an evaluation of the movie by the user using the result of the emotional analysis and the rating information by the movie rating apparatus comprises:
Calculating a difference value between the positive emotion value and the negative emotion value output as a result of the emotion analysis and multiplying the difference value between the positive emotion value and the negative emotion value by the maximum value of the predetermined constant range to calculate a normalization value , And integrating the normalization value with the rating information in a 1: 1 ratio,
Wherein a different weight is assigned according to the rating information when the normalization value and the rating information are combined.
The method according to claim 1, wherein the step of constructing a neighboring group for use as data for predicting a user's rating on the movie to be evaluated by the movie rating prediction apparatus comprises:
A method for predicting movie rating using tag emotion analysis, comprising constructing a collection of other users with similar tendencies to the user through k-means clustering.
delete delete delete delete A computer program stored on a medium for performing a movie rating prediction method using emotional analysis of a tag according to any one of claims 1 to 2.
A neighbors group constituting a neighbors group to utilize a user 's rating for a movie to be evaluated as data to be predicted;
An information extraction unit for extracting tag information and rating information for the movie to be evaluated from the neighboring group;
An emotional analysis unit for performing emotional analysis using the tag information; And
And a rating prediction unit for predicting a rating of a movie to be evaluated by the user using the result of the sensitivity analysis and the rating information,
Wherein the rating information is a constant value within a predetermined constant range,
Performing emotional analysis using the tag information may include:
The result of emotional analysis is output as positive, negative, and objective emotional values through an emotional analyzer using data created by users on a social network,
The sum of the positive, negative and objective emotional values is 1,
Predicting a rating for a movie to be evaluated by the user using the emotional analysis result and the rating information,
Calculating a difference value between the positive emotion value and the negative emotion value output as a result of the emotion analysis and multiplying the difference value between the positive emotion value and the negative emotion value by the maximum value of the predetermined constant range to calculate a normalization value , Integrates the normalization value with the rating information at a 1: 1 ratio,
Wherein when the normalization value and the rating information are combined, a different weight is assigned according to the rating information, and the predicted score is predicted as an integrated rating.
delete delete

Images