WO2022086063A1

WO2022086063A1 - Method and device for calculating style information on social network service account

Info

Publication number: WO2022086063A1
Application number: PCT/KR2021/014314
Authority: WO
Inventors: 김태훈
Original assignee: 주식회사 파라스타
Priority date: 2020-10-21
Filing date: 2021-10-15
Publication date: 2022-04-28
Also published as: KR20220052726A

Abstract

According to one embodiment of the present disclosure, disclosed is a method performed by a computing device comprising at least one processor, for calculating style information including style representation on a social account. The method may comprise the steps of: calculating partial style information including a style representation for at least one piece of social data included in a social data set; and on the basis of feature information corresponding to the at least one piece of social data, calculating style information of the social account from the partial style information for the at least one piece of social data.

Description

Method and device for calculating style information of social network service accounts

The present invention relates to a method of calculating style information of social data, and more particularly, to a method of calculating style information of social data using a color matching emotion scale and an artificial neural network.

As the number of SNS service users increases explosively, techniques for classifying a large amount of data existing in social network services are emerging. Representatively, in the art, there is a technique for acquiring information from image data and classifying data using a convolutional neural network or a technique for acquiring information from text data and classifying text data using a natural language processing engine.

However, only by inputting images or texts into the artificial neural network to learn, it is not possible to classify the data existing on the SNS in a vast amount, and there is a problem in that the indicators related to the emotions actually felt by humans are not reflected. That is, there has been a demand in the art to classify social data based on human emotions.

Korean Patent Registration KR1768521, which is a prior art, discloses a technique for detecting an object and classifying the properties of an object by inputting a given image into a convolutional neural network, "Method and system for providing information data about an object included in an image". .

The present disclosure has been devised in response to the above-described background art, and aims to calculate style information of social data.

The technical problems of the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

Disclosed is a method of calculating style information including a style representation of a social account performed by a computing device including at least one processor according to an embodiment of the present disclosure for realizing the above-described task. do. The method includes: calculating partial style information including a style expression for at least one social data included in a social data set; and calculating style information of a social account from partial style information on the at least one social data based on attribute information corresponding to the at least one social data.

In an alternative embodiment, the calculating of partial style information including a style expression for the at least one social data may include calculating first sentiment information from the social data based on a sentiment word set and a color matching sentiment scale. step; calculating second sentiment information from the social data based on a detection result obtained by inputting the emotional word set and the image included in the social data into an object detection module; Alternatively, the method may include at least one of calculating the third emotional information from the social data based on a processing result obtained by inputting the emotional word set and the text included in the social data into a natural language processing module.

In an alternative embodiment, the sentiment word set may include one or more sentiment words included in the color matching sentiment scale.

In an alternative embodiment, the calculating of the first sentiment information from the social data may include: comparing distances between a plurality of color chips included in the color matching sentiment scale and a representative partial image of an image included in the social data to calculating a color chip corresponding to the representative partial image on the color matching sensitivity scale; and calculating the first emotion information based on the emotion word mapped to the color chip corresponding to the representative partial image.

In an alternative embodiment, the calculating of the first sentiment information from the social data includes: comparing distances between a plurality of color chips included in the color matching sentiment scale and a first partial image of the image included in the social data. calculating a color chip corresponding to the first partial image on the color matching sensitivity scale; calculating a color chip corresponding to a second partial image on the color matching emotional scale by comparing distances between a plurality of color chips included in the color matching emotional scale and a second partial image of the image included in the social data; determining to change the score of the first emotional word mapped to the color chip corresponding to the first partial image with respect to the plurality of emotional words included in the emotional word set; determining to change the score of a second sentiment word mapped to a color chip corresponding to the second partial image with respect to a plurality of sentiment words included in the sentiment word set; and calculating the first sentiment information based at least in part on the score of the first sentiment word and the score of the second sentiment word included in the sentiment word set.

In an alternative embodiment, the color chip corresponding to the partial image may be calculated based on a distance between an RGB value of at least one color included in the color chip and an RGB value of at least one pixel included in the partial image. can

In an alternative embodiment, calculating the second sentiment information from the social data may include: inputting an image included in the social data into an object detection module to obtain a detection result; obtaining a word vector of the at least one object by inputting the name of the at least one object included in the detection result into a word embedding model; calculating a similarity between a word vector of the at least one object and a word vector of each of a plurality of emotional words included in the emotional word set; determining to change the score of one or more sentiment words having a word vector similar to the word vector of the at least one object based on the similarity calculation result; and calculating second sentiment information based on scores of a plurality of sentiment words included in the sentiment word set.

In an alternative embodiment, the calculating of the third sentiment information from the social data may include: obtaining a processing result obtained by inputting text included in the social data into a natural language processing module; obtaining a word vector of the at least one keyword by inputting at least one keyword included in the processing result into a word embedding model; calculating a similarity between a word vector of the at least one keyword and a word vector of a plurality of sentiment words included in the sentiment word set; determining to change the score of one or more sentiment words having a word vector similar to a word vector of the at least one keyword based on the similarity calculation result; and calculating third emotional information based on scores of a plurality of emotional words included in the emotional word set.

In an alternative embodiment, the calculating of the partial style information including the style expression for the at least one social data may include, for each of the first emotional information, the second emotional information, and the third emotional information, according to the first emotional information. Comprising the step of calculating partial style information using a weight value, a weight according to the second emotion information, and a weight applied to the third emotion information, wherein the weight according to the third emotion information is applied to the third emotion information When the result value exceeds a predetermined upper limit value, the result value may be replaced with a predetermined upper limit value.

In an alternative embodiment, the step of calculating the style information of the social account from the partial style information of the at least one social data may include calculating a weight according to each social data based on attribute information corresponding to the at least one social data. generating; and calculating the style information of the social account by applying a weight according to the social data to the partial style information of the at least one social data.

In an alternative embodiment, the attribute information corresponding to the social data may include feedback information of other users on the social data or time information of the social data.

In an alternative embodiment, the style information of the social account may be calculated based on additionally calculated statistical values related to the style information of at least one other social account.

In an alternative embodiment, the calculating of partial style information including a style expression for the at least one social data may include: based on a sentiment word set and a color matching sentiment scale, a first sentiment from an image included in the social data. calculating information; calculating second sentiment information based on a detection result obtained by inputting the emotional word set and the image included in the social data into an object detection module; and calculating third sentiment information based on a processing result obtained by inputting the emotional word set and the text included in the social data into a natural language processing module.

A computer program stored in a computer-readable storage medium is disclosed according to an embodiment of the present disclosure for realizing the above-described problems. The computer program, when executed on one or more processors, causes the following operations to be performed for calculating style information including a style representation of a social account, the operations comprising: at least one included in a social data set; calculating partial style information including a style expression for social data of and calculating style information of a social account from partial style information on the at least one social data based on attribute information corresponding to the at least one social data.

Disclosed is an apparatus for calculating style information including a style representation of a social account according to an embodiment of the present disclosure for realizing the above-described task. The apparatus may include one or more processors; Memory; and a network unit, and wherein the one or more processors calculate partial style information including a style expression for at least one social data included in a social data set, and add to attribute information corresponding to the at least one social data. Based on the partial style information for the at least one piece of social data, style information of the social account may be calculated.

The technical solutions obtainable in the present disclosure are not limited to the above-mentioned solutions, and other solutions not mentioned are clearly to those of ordinary skill in the art to which the present disclosure belongs from the description below. can be understood

The present disclosure may effectively provide a method of calculating style information of a social account.

Effects obtainable in the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those of ordinary skill in the art to which the present disclosure belongs from the description below. .

Various aspects are now described with reference to the drawings, wherein like reference numbers are used to refer to like elements collectively. In the following example, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of one or more aspects. It will be apparent, however, that such aspect(s) may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to facilitate describing one or more aspects.

1 is a block diagram of a computing device for calculating style information of a social account according to an embodiment of the present disclosure.

2 is an exemplary diagram for expressing a color coordination emotional scale according to an embodiment of the present disclosure.

3 is an exemplary diagram exemplarily illustrating a color chip and a partial image according to an embodiment of the present disclosure.

4 is a schematic diagram illustrating a network function according to an embodiment of the present disclosure.

5 is an exemplary diagram illustrating a word vector in a two-dimensional coordinate space according to an embodiment of the present disclosure.

6 is a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Various embodiments are now described with reference to the drawings. In this specification, various descriptions are presented to provide an understanding of the present disclosure. However, it is apparent that these embodiments may be practiced without these specific descriptions.

The terms “component,” “module,” “system,” and the like, as used herein, refer to a computer-related entity, hardware, firmware, software, a combination of software and hardware, or execution of software. For example, a component can be, but is not limited to being, a process running on a processor, a processor, an object, a thread of execution, a program, and/or a computer. For example, both an application running on a computing device and the computing device may be a component. One or more components may reside within a processor and/or thread of execution. A component may be localized within one computer. A component may be distributed between two or more computers. In addition, these components can execute from various computer readable media having various data structures stored therein. Components may communicate via a network such as the Internet with another system, for example, via a signal having one or more data packets (eg, data and/or signals from one component interacting with another component in a local system, distributed system, etc.) may communicate via local and/or remote processes depending on the data being transmitted).

In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or.” That is, unless otherwise specified or clear from context, "X employs A or B" is intended to mean one of the natural implicit substitutions. That is, X employs A; X employs B; or when X employs both A and B, "X employs A or B" may apply to either of these cases. It should also be understood that the term “and/or” as used herein refers to and includes all possible combinations of one or more of the listed related items.

Also, the terms "comprises" and/or "comprising" should be understood to mean that the feature and/or element in question is present. However, it should be understood that the terms "comprises" and/or "comprising" do not exclude the presence or addition of one or more other features, elements and/or groups thereof. Also, unless otherwise specified or unless it is clear from context to refer to a singular form, the singular in the specification and claims should generally be construed to mean “one or more”.

And, the term “at least one of A or B” should be interpreted as meaning “includes only A”, “includes only B”, and “combined with the configuration of A and B”.

Those skilled in the art will further appreciate that the various illustrative logical blocks, configurations, modules, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein may be implemented in electronic hardware, computer software, or combinations of both. It should be recognized that they can be implemented with To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logics, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application. However, such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

Descriptions of the presented embodiments are provided to enable those skilled in the art to use or practice the present invention. Various modifications to these embodiments will be apparent to those skilled in the art of the present disclosure. The generic principles defined herein may be applied to other embodiments without departing from the scope of the present disclosure. Therefore, the present invention is not limited to the embodiments presented herein. This invention is to be interpreted in its widest scope consistent with the principles and novel features presented herein.

In the present disclosure, the term “social data” may be used as a term including social network service (SNS) or online postings, electronic documents, and the like in some embodiments. Social data may include various types of data such as images, videos, and texts. Social data may be used as a term to refer to an arbitrary data unit uploaded by an SNS user to their social account. The arbitrary data unit may include a data unit according to an online page, channel, or the like, or a data unit according to a posting date and time. Social data may refer to, for example, a set of text and images uploaded by a user at the same time.

In the present disclosure, the term “social data set” may be used as a term meaning a set including one or more social data. The social data set may refer to the entire social data set included in the social account on the SNS in some embodiments. The social data set may refer to one or more social data sets extracted according to an arbitrary criterion among all social data included in a social account on SNS in some other embodiments.

In the present disclosure, the term “emotional word” may be used as a meaning including a word expressing an emotion or atmosphere that a user feels from an image or text included in social data. For example, when the image included in the social data is an image about 'beach', the emotional word for the social data may correspond to "cool", "cool", and the like. As another example, when the word 'midsummer' is included in the text included in the social data, the emotional word for the social data may include "hot", "passionate", "severe heat", and the like. Emotional words include words for expressing images included in social data. Emotional words include words for expressing text included in social data. In the present disclosure, the term “emotional word set” may mean a word set including N (eg, 120) emotional words as described above.

In the present disclosure, the term “style information” may refer to information representing a style of social data. The style information may be information including one or more emotional words that may be calculated from at least one piece of social data. In the present disclosure, the term “style information” may be used as information expressing a style for the entire social account. In the present disclosure, the term "partial style information" means a subset of style information. As an example, the partial style information may be used as information representing a style for one piece of social data. For example, when the first social data includes an image about the sea and text such as "I came to the sea after a long summer vacation," the partial style information calculated for the first social data is "blue"; It may include emotional words such as "refreshing", "relaxing", "pleasant", and the like. Continuing, for example, when the second social data includes an image about a mountain and text such as "Hiking is always fun", the partial style information calculated for the second social data is "green" , "tough", "enjoyable", etc. may include emotional words. Thereafter, the style information calculated for the social account including the first social data and the second social data may be calculated including emotional words such as “dynamic”, “trend”, “vigorous”, and the like. Partial style information and a method of calculating style information will be described later in detail. The above-described partial style information and examples of style information are merely exemplary and do not limit the present disclosure.

The configuration of the computing device 100 shown in FIG. 1 is only a simplified example. In an embodiment of the present disclosure, the computing device 100 may include other components for performing the computing environment of the computing device 100 , and only some of the disclosed components may configure the computing device 100 .

The computing device 100 may include a processor 110 , a memory 130 , and a network unit 150 .

The processor 110 may include one or more cores, and a central processing unit (CPU) of a computing device, a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU). unit) and the like, and may include a processor for data analysis and deep learning. The processor 110 may read a computer program stored in the memory 130 to perform data processing for machine learning according to an embodiment of the present disclosure. According to an embodiment of the present disclosure, the processor 110 may perform an operation for learning the neural network. The processor 110 for learning of the neural network, such as processing input data for learning in deep learning (DL), extracting features from the input data, calculating an error, updating the weight of the neural network using backpropagation calculations can be performed. At least one of a CPU, a GPGPU, and a TPU of the processor 110 may process learning of a network function. For example, the CPU and the GPGPU can process learning of a network function and data classification using the network function. Also, in an embodiment of the present disclosure, learning of a network function and data classification using the network function may be processed by using the processors of a plurality of computing devices together. In addition, the computer program executed in the computing device according to an embodiment of the present disclosure may be a CPU, GPGPU or TPU executable program.

According to an embodiment of the present disclosure, the memory 130 may store any type of information generated or determined by the processor 110 and any type of information received by the network unit 150 .

According to an embodiment of the present disclosure, the memory 130 is a flash memory type, a hard disk type, a multimedia card micro type, or a card type memory (eg, SD or XD memory, etc.), Random Access Memory (RAM), Static Random Access Memory (SRAM), Read-Only Memory (ROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Programmable Read (PROM) -Only Memory), a magnetic memory, a magnetic disk, and an optical disk may include at least one type of storage medium. The computing device 100 may operate in relation to a web storage that performs a storage function of the memory 130 on the Internet. The description of the above-described memory is only an example, and the present disclosure is not limited thereto.

The network unit 150 according to an embodiment of the present disclosure may use any type of wired or wireless communication systems.

In the present disclosure, the network unit 150 may be configured regardless of its communication mode, such as wired and wireless, and may be configured with various communication networks such as a personal area network (PAN) and a wide area network (WAN). can In addition, the network may be a well-known World Wide Web (WWW), and may use a wireless transmission technology used for short-range communication, such as infrared (IrDA) or Bluetooth (Bluetooth).

The techniques described herein may be used in the networks mentioned above, as well as in other networks.

In order to calculate the style information including the style representation of the social account according to the present disclosure, the processor 110 receives partial style information including the style representation for at least one social data included in the social data set. can be calculated.

The processor 110 according to the present disclosure may calculate style information of a social account based on partial style information on at least one or more social data included in the social data set. The style information of the social account may include emotional words representing the SNS account. The style information of the social account may be calculated from two or more partial style information. The style information of the social account may be calculated from the first partial style information and the second partial style information. The first partial style information and the second partial style information may be respectively calculated from different first and second social data. For example, the processor 110 according to the present disclosure adds up partial style information for a plurality of social data to finally obtain style information of a social account including emotional words such as “cute”, “elegant”, “comfortable”, etc. can be calculated. The detailed description of the user's style information described above is only an example and does not limit the present disclosure.

According to the present disclosure, the step of the processor 110 calculating the partial style information including the style expression for the at least one social data is based on the emotional word set and the color matching emotion scale, the first emotional information from the social data calculating, based on a detection result obtained by inputting the emotional word set and the image included in the social data into an object detection module, calculating second emotional information from the social data or the emotional word set and the The method may include at least one of calculating third emotional information from the social data based on a processing result obtained by inputting text included in the social data into the natural language processing module.

2 is an exemplary diagram for expressing a color coordination emotional scale according to an embodiment of the present disclosure. The color matching emotion scale is based on the human psychology felt from the color chip, and the emotion is classified according to the color chip. The color chip may include a color scheme that is a combination of colors. The color matching emotion scale is based on the statistics of the emotion according to the color chip, and after dividing the emotion felt in the color chip into a scale, the emotion words according to the color chip are mapped for each different color chip. The color matching sensitivity scale may be expressed in a two-dimensional coordinate plane. The color matching emotion scale may be data obtained by mapping color chips according to a color combination for each emotional word according to each coordinate on a two-dimensional coordinate plane having two axes. The color matching sensibility scale may include a first axis and a second axis as two axes. The first axis may be, for example, an axis in which a psychological feeling (ie, emotion) of a color combination included in the color chip is expressed numerically according to the degree of dynamic. Both ends of the first axis may be dynamic-static, respectively. The second axis may be an axis in which a psychological feeling of a color combination included in the color chip is expressed numerically according to the degree of softness. Both ends of the second shaft may each be soft-hard. As shown in FIG. 2 , each color chip is mapped on two-dimensional coordinates and can be matched with emotional words such as “glamorous”, “elegant”, and “dynamic”. The color matching emotion scale may include data to which a color chip corresponding to each coordinate corresponding to one or more emotional words is mapped. For example, the (1, 1) point 211 of the color matching sensibility scale shown in FIG. 2 has the emotional word “natural” and may be mapped to a color chip indicated by reference numeral 213 . For another example, the (-1, -1) point 231 of the color matching sensibility scale shown in FIG. 2 has the emotional word “glamorous” and may be mapped to a color chip indicated by reference numeral 233 . It will be apparent to those skilled in the art that the above-described color matching emotion scale is only an example, and in implementing the present disclosure, a color matching emotion scale including various types of color chips and corresponding emotion words may be used.

In the detailed description and claims of the present disclosure, each emotion information expressed as "first emotion information", "second emotion information", and "third emotion information" is data of a plurality of emotion words included in the emotion word set. can be included as The first emotion information, the second emotion information, and the third emotion information may include one or more emotion words as emotion information that is a basis for calculating the partial style information. The sentiment information may include scores of each of a plurality of sentiment words included in the sentiment word set as data. The sentiment information may include sentiment words included in the sentiment word set as candidate sentiment words. For example, if the emotional words included in the emotional word set are ["dynamic", "glamorous", "calm"], the first emotional information is ["dynamic":5, "glamorous":6, "serious"] Like ":0], the score of each sentiment word may be included as data. The score of each sentiment word included in the sentiment word set may have a value of '0' as an initial value. Each emotional word included in the emotional word set may have a different pre-determined initial score for each emotional word.

According to the present disclosure, the step of the processor 110 calculating the first emotion information from the social data may include comparing distances between a plurality of color chips included in the color matching emotion scale and a representative partial image of the image included in the social data. The method may include calculating a color chip corresponding to the representative partial image on the color matching emotion scale and calculating first emotional information based on the emotion word mapped to the color chip corresponding to the representative partial image.

The processor 110 may calculate a color chip corresponding to the representative partial image on the color matching emotional scale by comparing distances between the plurality of color chips included in the color matching emotional scale and the representative partial image of the image included in the social data. . The representative partial image may be generated based on an image included in the social data. The representative partial image may be one partial image including at least a part of the image included in the social data. The representative partial image may be an image itself included in the social data. The representative partial image may be separately generated from an image included in the social data through a series of generation processes.

In an embodiment according to the present disclosure, the representative partial image generation process may include lowering the resolution of an image included in the social data. For example, in the case of an image having a size of 9x9 pixels, the processor 110 may calculate an average value of the values of each RGB channel based on a window having a size of 3x3 pixels, and then replace it with 1x1 pixels. In this case, in the above-described example, the processor 110 may generate a representative partial image having a size of 3x3 pixels from an image of a size of 9x9 pixels through a window having a size of 3x3 pixels. The color chip corresponding to the representative partial image may be calculated based on a distance between an RGB value of at least one color included in the color chip and an RGB value of at least one pixel included in the representative partial image.

In another embodiment according to the present disclosure, the representative partial image generating process may include an operation of extracting one or more main colors from an image included in social data. The processor 110 may configure a representative partial image by extracting N main colors from an image included in the social data. The N main colors may be selected based on statistical calculations on colors of a plurality of pixels included in the image. The color data may be an ordered pair representing RGB values. The statistical operation on the color may include a filtering operation for reducing the total number of colors. The filtering operation may include an operation of substituting one RGB value for pixels having a difference in RGB values less than or equal to a threshold value. The representative partial image may include the top N main colors having the largest number of pixels in the image of the social data. The representative partial image may include not only the top N main colors having the largest number of pixels in the image of the social data, but also ratio weights between the top N main colors. For example, the processor 110 may extract four main colors from the image included in the social data in the following data format.

[ <(255,255,255), 0.56> <(198,197,193), 0.27> <(0,255,255), 0.10> <(0, 0, 0), 0.07> ]

In the above example, each tuple included in the data may include the RGB value of the main color and the ratio weight. In another embodiment of the present disclosure, the processor 110 may generate a representative partial image except for one or more main colors having the highest weight. For example, when extracting the top N main colors, the processor 110 may obtain N+M main colors, cut out the top M main colors, and extract the remaining N main colors as the main colors. M may be a natural number of 1 or more. When one or more main colors with high weight are excluded as described above, the processor 110 has the effect of extracting emotional information by focusing on the colors other than the background from the image photographed in the same background. In another embodiment, the processor 110 may perform an operation of extracting the top N main colors to be included in the representative partial image from the remaining image after cutting out the portion corresponding to the background.

Hereinafter, a specific method in which the processor 110 compares distances between a plurality of color chips and a representative partial image of an image included in the social data to calculate a color chip corresponding to the representative partial image on the color matching emotion scale will be described. . 3 is an exemplary diagram exemplarily illustrating a color chip and a partial image according to an embodiment of the present disclosure. As an embodiment according to the present disclosure, the processor 110 may calculate a distance between the representative partial image 310 composed of four pixels and each color chip 330 included in the plurality of color chips. The processor 110 may compare a distance between the plurality of color chips and the representative partial image 310 with each other.

In order to calculate the distance between the color chip 330 and the representative partial image 310 , the processor 110 may calculate the distance between the RGB value coordinates by matching each pixel and each color one-to-one. The distance function between the RGB value coordinates may include, for example, a Manhattan distance function (L1 distance function), a Euclidean distance function (L2 distance function), and the like.

When the distance function between the RGB value coordinates is the Manhattan distance function, the equation is the same as Equation 1.

is the Manhattan distance function.

Wow

represents different RGB value coordinates, respectively.

Wow

is each

Wow

of

Indicates the value of the second element.

When the distance function between the RGB value coordinates is the Euclidean distance function, Equation 2 is the same.

is the Euclidean distance function.

Wow

represents different RGB value coordinates, respectively.

Wow

is each

Wow

of

Indicates the value of the second element.

According to an embodiment of the present disclosure, the RGB value of the first pixel 311 included in the representative partial image 310 is [102,102,051], the RGB value of the second pixel 313 is [102,000,102], and the RGB value of the third pixel The value may be [255,153,255], and the RGB value of the fourth pixel 317 may be [255,204,255]. At this time, the color chip 330 for distance calculation includes a first color 331 having an RGB value of [000, 000, 102], a second color 333 having an RGB value of [051, 051, 102], It may be composed of a third color 335 having an RGB value of [051, 051, 153]. According to Equation 1, the distance between the first pixel 311 and the RGB value of the first color 331 is

can be Similarly, the distance between the first pixel 311 and the RGB values of the second color 333 may be calculated as 153 and the distance between the first pixel 311 and the RGB values of the third color 335 may be calculated as 204 . According to Equation 2, the distance between the first pixel 311 and the RGB value of the first color 331 is

can be Similarly, the distance between the RGB values of the first pixel 311 and the second color 333 may be 88.3, and the distance between the first pixel 311 and the RGB values of the third color 335 may be 124.9.

The processor 110 calculates the distances between the RGB values of the first pixel 311 included in the representative partial image 310 and the RGB values of the first, second, and third colors included in the color chip 330 . After each calculation, the distance between the first pixel 311 and the color chip 330 may be calculated by summing the distances between the RGB values. The summation may include, for example, a sum total, an average, a weighted sum, and the like. As an example, the distance between the first pixel 311 and the color chip 330 may be calculated as 255+153+204=612 as the sum of the distances between RGB values. As another example, the distance between the first pixel 311 and the color chip 330 may be calculated as 204, which is an average of 255, 153, and 204. The above-described summing method is merely an example and includes various summation methods for various values without limitation.

The processor 110 determines the distance between the first pixel 311 and the color chip 330 and the distance between the second pixel 313 and the color chip 330 for a plurality of pixels included in the representative partial image 310 , The distance between the third pixel 315 and the color chip 330 and the distance between the fourth pixel 317 and the color chip 330 may be calculated. Thereafter, the processor 110 may calculate the distance between the representative partial image 310 and the color chip 330 by summing the distances to the color chip 330 calculated for each pixel. The summation may include, for example, a sum total, an average, or a weighted sum.

As described above, the processor 110 calculates the distance between the representative partial image 310 and the color chip 330 for a plurality of color chips, and then performs each of the representative partial image 310 and each color chip included in the plurality of color chips. The distance between the color chips can be compared with each other. The processor 110 may calculate a color chip having the smallest distance among the distances between the representative partial image 310 and each color chip included in the plurality of color chips as a color chip corresponding to the representative partial image.

As an additional embodiment of the present disclosure, a method of calculating a color chip corresponding to the representative partial image when the processor 110 generates the representative partial image by extracting the top N main colors will be described. As described above, the representative partial image according to an embodiment of the present disclosure may include N main colors expressed by (A_1, A_2, A_3, ..., A_N). In addition, the ratio weights W_1, W_2, W_3, ..., W_N of each of the N main colors may be further included. Meanwhile, a first color chip among a plurality of color chips included in the color matching sensibility scale may include colors expressed by (B_1, B_2, B_3). In this case, the degree of similarity between the representative partial image and the first color chip may be calculated based on the similarity between the representative partial image and the color B_1, the similarity between the representative partial image and the color B_2, and the similarity between the representative partial image and the color B_3. Furthermore, the degree of similarity between the representative partial image and the B_1 color may be calculated by applying a ratio weight existing for each main color to the difference between the B_1 color and each main color included in the representative partial image. More specifically, the degree of similarity between the representative partial image and each color chip may be calculated based on Equation 3, for example.

In Equation 3, N denotes the total number of main colors included in the representative partial image, A_i denotes the RGB value of the i-th main color in the representative partial image, and B_j denotes the RGB value of the j-th color in the color chip. M is the number of colors included in one color chip, and may be, for example, three. The above-described calculation method and formula for the degree of similarity between the representative partial image and the color chip is merely an example for explaining the present disclosure, and the present disclosure is based on the color, ratio, and color included in the representative partial image and the color chip. Therefore, all methods that can calculate the similarity are included without limitation.

The processor 110 according to the present disclosure can calculate the degree of similarity between each color chip and the representative partial image as described above, and when it is performed for all color chips, the most similar predetermined number of color chips according to the degree of similarity can decide

According to an embodiment of the present disclosure, the processor 110 may calculate the first emotion information based on the emotion word mapped to the color chip corresponding to the representative partial image 310 . In an embodiment, the processor 110 may determine that the score of the emotional word of the color chip having the smallest distance from the representative partial image 310 among the plurality of emotional words is changed. For example, the alteration may include increasing the score of the sentiment word. In another embodiment, the processor 110 sets the color scheme with the emotional word of the color chip having the smallest distance from the representative partial image 310 among the plurality of emotional words and the emotional word of the color chip having the smallest distance from the representative partial image 310 . It is possible to increase the score of the top M (eg, 5) emotional words according to the distance on the sentiment scale. The processor 110 may calculate the first emotion information by increasing the score of one or more emotion words based on the emotion words mapped with the color chip corresponding to the representative partial image 310 . The first sentiment information may include a plurality of sentiment words and scores for each sentiment word as data.

According to the present disclosure, the step of the processor 110 calculating the first emotion information from the social data may include comparing the distances between the plurality of color chips included in the color matching emotion scale and the first partial image of the image included in the social data. calculating a color chip corresponding to the first partial image on the color matching emotional scale by comparing the distances between the plurality of color chips included in the color matching emotional scale and the second partial image of the image included in the social data. Calculating a color chip corresponding to the second partial image on the color matching emotional scale, and the score of the first emotional word mapped with the color chip corresponding to the first partial image for a plurality of emotional words included in the emotional word set increasing the score of the second emotional word mapped to the color chip corresponding to the second partial image with respect to the plurality of emotional words included in the emotional word set, and the plurality of emotional words included in the emotional word set The method may include calculating the first emotional information based on the scores of the emotional words.

The first partial image and the second partial image according to the present disclosure may be partial images including at least a part of an image included in social data. At least a part of the image included in the social data may mean a set of one or more pixels existing in an arbitrary area of the image included in the social data. The first partial image and the second partial image may be partial images in which at least one pixel is not equal to each other. For example, the first partial image and the second partial image may be partial images having a size of 2x2 pixels as indicated by reference numeral 310 of FIG. 3 .

According to an embodiment of the present disclosure, the processor 110 compares distances between a plurality of color chips included in the color matching sensitivity scale and the first partial image to calculate a color chip corresponding to the first partial image on the color matching emotional scale The step of comparing the distances between a plurality of color chips included in the color matching sensitivity scale and the second partial image and calculating a color chip corresponding to the second partial image on the color matching emotional scale corresponds to the representative partial image described above. It can be clearly understood from the point of view of a person skilled in the art that it can be performed in the same way as the method of calculating the color chip, so it is omitted.

According to an embodiment of the present disclosure, the color chip corresponding to the first partial image is based on a distance between the RGB value of at least one color included in the color chip and the RGB value of at least one pixel included in the first partial image. can be calculated by Also, the color chip corresponding to the second partial image may be calculated based on a distance between an RGB value of at least one color included in the color chip and an RGB value of at least one pixel included in the second partial image. The method of calculating the distance between the RGB values may be understood in the same manner with reference to FIG. 3 .

After calculating the color chip corresponding to the first partial image as described above, the processor 110 refers to the color matching emotion scale for a plurality of emotional words included in the emotional word set to the color chip corresponding to the first partial image. It may be determined that the score of the mapped first sentiment word is changed. For example, the alteration may include increasing the score of the first sentimental word. In the same manner, the processor 110 calculates a color chip corresponding to the second partial image, and then refers to the color matching emotion scale for a plurality of emotional words included in the emotional word set, and the color chip corresponding to the second partial image. It may be determined to change the score of the second sentiment word mapped to . For example, the alteration may include increasing the score of the second sentiment word. After the processor 110 performs score adjustment on the plurality of emotional words included in the emotional word set by each partial image, the first emotional information based on the scores of the plurality of emotional words included in the emotional word set can be calculated. Concretely, for example, suppose that the emotional word set contains [“dynamic”, “cynic”, “cute”]. In this case, the first emotional word mapped to the color chip corresponding to the first partial image may be “dynamic”, and the processor 110 may increase the score of the first emotional word. Thereafter, the second emotional word mapped to the color chip corresponding to the second partial image may also be “dynamic”, and thus the processor 110 may increase the score of the second emotional word. Finally, the scores of the plurality of emotional words included in the emotional word set may be [“dynamic”:2, “cynic”:0, “cute”:0]. The processor 110 may calculate this score as the first emotion information. The above-described embodiment is only one embodiment, and the present disclosure includes various embodiments for calculating the first sentiment information from a sentiment word set including a plurality of sentiment words without limitation.

According to an embodiment of the present disclosure, the step of the processor 110 calculating the second emotional information from the social data may include: inputting an image included in the social data into an object detection module to obtain a detection result; obtaining a word vector of the at least one object by inputting the name of the included at least one object into a word embedding model, each of the word vector of the at least one object and a plurality of emotional words included in the emotional word set calculating the similarity between word vectors of , increasing the score of one or more sentiment words having a word vector similar to the word vector of the at least one object based on the similarity calculation result, and included in the sentiment word set The method may include calculating second sentiment information based on the scores of the plurality of sentiment words.

Throughout this specification, computational models, neural networks, network functions, and neural networks may be used interchangeably. For example, an object detection module, a word embedding model, or a natural language processing module in the present disclosure may operate based on a neural network, which will be specifically described below.

A neural network may be composed of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network is configured to include at least one or more nodes. Nodes (or neurons) constituting the neural networks may be interconnected by one or more links.

In the neural network, one or more nodes connected through a link may relatively form a relationship between an input node and an output node. The concepts of an input node and an output node are relative, and any node in an output node relationship with respect to one node may be in an input node relationship in a relationship with another node, and vice versa. As described above, an input node-to-output node relationship may be created around a link. One or more output nodes may be connected to one input node through a link, and vice versa.

In the relationship between the input node and the output node connected through one link, the value of the data of the output node may be determined based on data input to the input node. Here, a link interconnecting the input node and the output node may have a weight. The weight may be variable, and may be changed by a user or an algorithm in order for the neural network to perform a desired function. For example, when one or more input nodes are interconnected to one output node by respective links, the output node sets values input to input nodes connected to the output node and links corresponding to the respective input nodes. An output node value may be determined based on the weight.

As described above, in a neural network, one or more nodes are interconnected through one or more links to form an input node and an output node relationship in the neural network. The characteristics of the neural network may be determined according to the number of nodes and links in the neural network, the correlation between the nodes and the links, and the value of a weight assigned to each of the links. For example, when the same number of nodes and links exist and there are two neural networks having different weight values of the links, the two neural networks may be recognized as different from each other.

A neural network may consist of a set of one or more nodes. A subset of nodes constituting the neural network may constitute a layer. Some of the nodes constituting the neural network may configure one layer based on distances from the initial input node. For example, a set of nodes having a distance of n from the initial input node may constitute n layers. The distance from the initial input node may be defined by the minimum number of links that must be passed to reach the corresponding node from the initial input node. However, the definition of such a layer is arbitrary for description, and the order of the layer in the neural network may be defined in a different way from the above. For example, a layer of nodes may be defined by a distance from the final output node.

The initial input node may mean one or more nodes to which data is directly input without going through a link in a relationship with other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes among nodes in the neural network. In addition, the hidden node may mean nodes constituting the neural network other than the first input node and the last output node.

The neural network according to an embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as progresses from the input layer to the hidden layer. can Also, in the neural network according to another embodiment of the present disclosure, the number of nodes in the input layer may be less than the number of nodes in the output layer, and the number of nodes may be reduced as the number of nodes progresses from the input layer to the hidden layer. there is. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as the number of nodes progresses from the input layer to the hidden layer. can The neural network according to another embodiment of the present disclosure may be a neural network in a combined form of the aforementioned neural networks.

A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can be used to identify the latent structures of data. In other words, it can identify the potential structure of photos, texts, videos, voices, and music (e.g., what objects are in the photos, what the text and emotions are, what the texts and emotions are, etc.) . Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted boltzmann machines (RBMs). machine), a deep belief network (DBN), a Q network, a U network, a Siamese network, and a Generative Adversarial Network (GAN). The description of the deep neural network described above is only an example, and the present disclosure is not limited thereto.

In an embodiment of the present disclosure, the network function may include an autoencoder. The auto-encoder may be a kind of artificial neural network for outputting output data similar to input data. The auto encoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between the input/output layers. The number of nodes in each layer may be reduced from the number of nodes of the input layer to an intermediate layer called the bottleneck layer (encoding), and then expanded symmetrically with reduction from the bottleneck layer to the output layer (symmetrically with the input layer). The auto-encoder can perform non-linear dimensionality reduction. The number of input layers and output layers may correspond to a dimension after preprocessing the input data. In the auto-encoder structure, the number of nodes of the hidden layer included in the encoder may have a structure that decreases as the distance from the input layer increases. If the number of nodes in the bottleneck layer (the layer with the fewest nodes located between the encoder and decoder) is too small, a sufficient amount of information may not be conveyed, so a certain number or more (e.g., more than half of the input layer, etc.) ) may be maintained.

The neural network may be trained using at least one of supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Learning of the neural network may be a process of applying knowledge for the neural network to perform a specific operation to the neural network.

A neural network can be trained in a way that minimizes output errors. In the training of a neural network, iteratively input the training data into the neural network, calculate the output of the neural network and the target error for the training data, and calculate the error of the neural network from the output layer of the neural network to the input layer in the direction to reduce the error. It is a process of updating the weight of each node in the neural network by backpropagation in the direction. In the case of teacher learning, learning data in which the correct answer is labeled in each learning data is used (ie, labeled learning data), and in the case of comparative learning, the correct answer may not be labeled in each learning data. That is, for example, learning data in the case of teacher learning related to data classification may be data in which categories are labeled in each of the learning data. Labeled training data is input to the neural network, and an error can be calculated by comparing the output (category) of the neural network with the label of the training data. As another example, in the case of comparison learning related to data classification, an error may be calculated by comparing the input training data with the neural network output. The calculated error is back propagated in the reverse direction (ie, from the output layer to the input layer) in the neural network, and the connection weight of each node of each layer of the neural network may be updated according to the back propagation. The change amount of the connection weight of each node to be updated may be determined according to a learning rate. The computation of the neural network on the input data and the backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of repetitions of the learning cycle of the neural network. For example, in the early stage of learning of a neural network, a high learning rate can be used to enable the neural network to quickly obtain a certain level of performance, thereby increasing efficiency, and using a low learning rate at a later stage of learning can increase accuracy.

In the training of neural networks, in general, the training data may be a subset of real data (that is, data to be processed using the trained neural network), and thus, the error on the training data is reduced, but the error on the real data is reduced. There may be increasing learning cycles. Overfitting is a phenomenon in which errors on actual data increase by over-learning on training data as described above. For example, a phenomenon in which a neural network that has learned a cat by seeing a yellow cat does not recognize that it is a cat when it sees a cat other than yellow may be a type of overfitting. Overfitting can act as a cause of increasing errors in machine learning algorithms. In order to prevent such overfitting, various optimization methods can be used. In order to prevent overfitting, methods such as increasing the training data, regularization, and dropout that deactivate some of the nodes of the network in the process of learning, and the use of a batch normalization layer are applied. can

In the detailed description and claims of the present invention, a convolutional neural network may include a sequence of neural network layers. The neural network layer may include a convolutional layer, a pooling layer, a fully connected layer, and the like. The initial input to the convolutional neural network may be received by the lowest first layer in the sequence. A convolutional neural network may sequentially input an initial input to layers in a sequence to generate a final output from the initial input. The initial input may be, for example, an image and the final output may be a set of scores for each category in a set of categories comprising one or more categories.

In the description and claims of the present invention, each neural network layer in a convolutional neural network includes a set of nodes. Each neural network layer may receive an initial input to the convolutional neural network or an output of the previous neural network layer as an input. For example, in a sequence including a plurality of neural network layers, the N-th neural network layer may receive the output of the N-1 th neural network layer as an input. Each neural network layer within a convolutional neural network generates an output from its input. If the layer is the highest layer in the sequence, it is treated as the final output of the convolutional neural network.

According to an embodiment of the present disclosure, when the processor 110 inputs an image included in social data to the object detection module, the processor 110 may obtain a detection result from the object detection module. The object detection module may include at least one node including a learnable parameter. The object detection module may include at least one convolutional neural network layer for object detection. The object detection module may be a previously learned module to detect various types of objects present in the input image. The object detection module may output at least one of a position, size, name, or type of one or more objects present in the input image. The detection result obtained by the processor 110 from the object detection module includes the names of one or more objects present in the image included in the social data. For example, suppose an image contained in social data includes a cup and a clock. In this case, the processor 110 may obtain the name of the object such as “cup” and “watch” as a detection result from the object detection module.

According to an embodiment of the present disclosure, the processor 110 may obtain a word vector of the at least one object by inputting the name of at least one object included in the detection result obtained from the object detection module into the word embedding model. Word 　 embedding is a term referring to a method for expressing a word as a vector, and mainly means converting from a sparse representation of a word vector to a dense representation of a word vector. For example, in the One Hot Vector generated through the One Hot Encoding method, the index value of the word to be expressed may be 1, and the remaining index values may be expressed as 0. As such, in the sparse expression, most of the values of elements of a vector or matrix are expressed as 0. On the other hand, in the dense representation, the values of elements of a vector or matrix may have real values, and as a result, the dimension of a word vector may be reduced compared to a one-hot vector. Word 　 embedding may mean an operation of expressing a word as a dense 　 expression vector. A vector generated through a word embedding method may be referred to as an embedding vector or a word embedding vector throughout this specification. The word embedding method may include word2vec, FastText, Glove, and the like.

In an embodiment of the present disclosure, the word embedding model may include one or more neural network layers. The word embedding model may include at least one hidden layer. The layer may include one or more nodes. The word embedding model may obtain an input value of the hidden layer by using a first matrix for a sparse vector of a word input to the input layer, and transmit an output value to the output layer using the input value of the hidden layer and the second matrix. The output value may be a dense vector of input words. The input word may be one of a word related to an object that may exist in an image included in the social data or an emotional word included in the emotional word set. The neural network structure of the above-described word embedding model is merely an example, and the present disclosure is not limited thereto. The connection weight and bias value of each node included in the word embedding model may be pre-trained by a corpus including a plurality of words. The corpus for dictionary learning may include emotional words included in the emotional word set according to an embodiment of the present disclosure. As a result, the learned word embedding model may convert a word vector of a sparse expression into a word vector of a dense expression for a plurality of emotional words included in the emotional word set.

According to an embodiment of the present disclosure, the processor 110 may calculate a similarity between a word vector of at least one object included in the image of the social data and a word vector of a plurality of sentiment words included in the emotional word set.

In an embodiment, the similarity may be calculated based on a distance of two word vectors in a vector space. The distance in the vector space may be defined as, for example, a Manhattan distance or a Euclidean distance. Word vectors generated through the word embedding model according to the present disclosure may form a cluster according to the degree of similarity. For example, when the word A and the word B are similar words, the distance between the word vector for word A and the word vector for word B in the vector space may be close. In this case, the word A and the word B will be located close to each other in the vector space, and the distance value is small, so that the word A and the word B may have high similarity.

In another embodiment, the similarity may be defined as cosine similarity. The cosine similarity may be a value for determining the similarity by obtaining a cosine angle between two vectors. The cosine similarity may have a value between -1 and 1 depending on the degree of agreement between two vectors. The cosine similarity value may have a value of 1 when the two vectors completely match. The description of the embodiment of the degree of similarity is merely an example, and the present disclosure is not limited thereto.

The processor 110 according to the present disclosure may determine to change the score of one or more sentiment words having a word vector similar to the word vector of at least one object based on the similarity calculation result. An example of a change may include increasing the score of a sentiment word. Hereinafter, a method of determining one or more emotional words having a word vector similar to a word vector of an object will be described with reference to FIG. 5 .

According to an embodiment of the present disclosure, the processor 110 may determine one sentiment word having the highest similarity to the word vector of at least one object included in the image of the social data. Specifically, in FIG. 5, the word vector coordinates 511 of the first emotional word, the word vector coordinates 513 of the second emotional word, the word vector coordinates 515 of the third emotional word, and the fourth emotional word are exemplarily shown in FIG. Word vector coordinates 517 are shown. In this case, the word vector of the sentiment word having the highest similarity to the coordinates 531 of the word vector of the first object included in the image may be determined as the word vector coordinates 511 of the first sentiment word. In this case, the similarity may be calculated based on the Euclidean distance between word vector coordinates. The processor 110 may determine that the word vector coordinates 531 of the first object are closest to the word vector coordinates 531 of the first emotional word, and increase the score of the first emotional word.

According to another embodiment of the present disclosure, the processor 110 may determine top N sentiment words having a word vector having a high similarity to a word vector of at least one object included in the image of the social data. Specifically, if N is 2, and the processor 110 seeks to find the upper two sentiment word word vectors similar to the word vector of the first object, the processor 110 is ordered close to the word vector coordinates 531 of the first object. As shown, it is possible to calculate the word vector coordinates 511 of the first emotional word, which are the word vector coordinates of the upper two emotional words, and the word vector coordinates 513 of the second emotional word. As a result, the processor 110 may determine the upper two word vectors having a high similarity to the word vector of the first object as the word vector of the first sentiment word and the word vector of the second sentiment word. Subsequently, the processor 110 may determine to change the scores of the first sentiment word and the second sentiment word based on the similarity calculation result. An example of such a change may include increasing the scores of the first sentiment word and the second sentiment word.

In a further embodiment of the present disclosure, when determining one or more sentiment words having a word vector similar to a word vector of at least one object included in the image of the social data, the processor 110 is additionally based on the detection result of the object detection module can do. The detection result of the object detection module may include at least one of a name of the detected object and a detection accuracy (confidence score). Specifically, for example, the processor 110 may obtain 'lemon' and 'flower' as detection results after inputting an image of social data into the object detection module. The processor 110 may additionally acquire the detection accuracy (eg, ['lemon', 0.95] or ['flower', 0.88], etc.) for each detection result from the object detection module. When a plurality of objects are detected from the image of the social data by the object detection module, the processor 110 may determine one or more sentiment words for the image of the social data based additionally on detection accuracy. For example, it is assumed that the processor 110 according to the present disclosure extracts top N sentiment words having a word vector having a high similarity with respect to an individual object detected from an image of social data. In this case, the top N emotional word set A (for example, 'fresh', 'yellow', 'cool', etc.) will be extracted for the first object (eg 'lemon') extracted from the image of social data can Also, for the second object (for example, 'flower') extracted from the image of the same social data, the top N emotional word set B (for example, 'beautiful', 'soft', 'lively', etc.) will be extracted. can Furthermore, the processor 110 sets the top N sentiment words having a word vector with high similarity to the word vector of the first object and the second object in consideration of the detection accuracy for each object included in the detection result obtained from the object detection module. C can be determined. In the process of the processor 110 determining the sentiment word set C, the result of multiplying the detection accuracy (eg, 0.95) of the first object by the sentiment word set A as a weight and the detection accuracy of the second object (eg 0.88) ) as a weight, a calculation process of summing the result of multiplying the emotional word set B may be included. The processor 110 according to the present disclosure may change the score for at least one sentiment word based on the determined score of the sentiment word set C. FIG. It will be apparent to those skilled in the art that the above-described examples are merely exemplary descriptions for carrying out the present disclosure, and do not limit the present disclosure, and that operations on sentiment word sets may be performed in various aspects. The method of calculating the style information of the social account according to the present disclosure has the effect of more accurately extracting emotional words by reflecting the object detection result of the object detection module and the reliability of the object detection module in the style information calculation process as described above. For example, the higher the detection accuracy of the object, the more reliable the detection result is, and the style information calculation method according to the present disclosure assigns a high weight to a powerful detection object having high reliability, so that the object detection module is present in the image. It has the advantage that reliable results can be derived even if some objects are not recognized properly due to noise. That is, the present disclosure discloses a method of calculating style information having a characteristic robust to noise.

According to another embodiment of the present disclosure, the processor 110 may determine one or more sentiment words having a word vector within a predetermined threshold similarity to a word vector of at least one object included in the image of the social data. In detail, reference numeral 535 of FIG. 5 denotes a circle indicating an area separated by a predetermined distance from the word vector coordinate 533 of the second object. When the similarity is calculated based on, for example, the Euclidean distance between the vectors, the circle 535 representing the area separated by a certain distance from the word vector coordinate 533 of the second object is the area representing the predetermined threshold similarity range. can be Accordingly, the processor 110 considers the word vector coordinates 533 of the second object and a predetermined threshold similarity, and exists inside the circle 535 indicating the area separated by a predetermined distance from the word vector coordinates 533 of the second object. The word vector coordinates 515 of the third emotional word and the word vector coordinates 517 of the fourth emotional word may be calculated. As a result, the processor 110 may determine one or more word vectors within a predetermined threshold similarity with the word vector of the second object as the word vector of the third sentiment word and the word vector of the fourth sentiment word. Subsequently, the processor 110 may determine to change the scores of the third sentiment word and the fourth sentiment word based on the similarity calculation result. An example of such a change may include increasing the scores of the third sentiment word and the fourth sentiment word.

As described above, the processor 110 calculates a word vector of one or more sentiment words similar to the word vector of an object through a word vector by a word embedding model, and increases the score of the sentiment word based on the similarity calculation result, The second emotion information may be calculated based on the scores of the plurality of emotional words included in the emotional word set.

The step of the processor 110 according to the present disclosure calculating the third emotional information from the social data includes: obtaining a processing result obtained by inputting text included in the social data into a natural language processing module; at least included in the processing result inputting one keyword into a word embedding model to obtain a word vector of the at least one keyword; calculating the similarity between the word vector of the at least one keyword and the word vector of a plurality of sentiment words included in the sentiment word set increasing the score of one or more sentiment words having a word vector similar to the word vector of the at least one keyword based on the similarity calculation result, and based on the scores of a plurality of sentiment words included in the sentiment word set It may include calculating the third emotional information.

The natural language processing module according to the present disclosure may receive text included in social data and perform a natural language processing (NLP) operation. The natural language processing operation performed by the natural language processing module may include a morpheme analysis operation for the input text, a keyword extraction operation, and the like.

The natural language processing module according to the present disclosure may analyze the input text by dividing it into morpheme units through a morpheme analysis operation. For example, the natural language processing module can analyze the input text by dividing it into a real morpheme and a formal morpheme. For another example, the natural language processing module may analyze the input text by dividing it into lexical morphemes and grammatical morphemes. For example, if the natural language processing module performs a stemming operation on the text “I went home”, the text is converted to morphemes can be analyzed. For another example, when a morphological analysis operation is performed on the text “mountain is blue”, morpheme analysis may be performed on the text as “mountain/noun + li/subordinate + blue/adjective + da/endum”.

The natural language processing module according to the present disclosure may perform an operation of extracting a keyword after the morpheme analysis operation. The natural language processing module may calculate one or more extracted keywords as a processing result. The processing result may include an importance level corresponding to the keyword extracted from the text. A keyword may be extracted in the form of a root or prototype of an actual morpheme or a lexical morpheme in the morpheme analysis result. For example, when two morphemes “home” and “a” are analyzed as actual morphemes, the natural language processing module may extract “home” and “go” as keywords and then calculate the two keywords as processing results. The above-described example regarding the operation of the natural language processing module is merely an example for explanation, and the present disclosure includes without limitation a natural language processing module for extracting keywords from input text.

According to an embodiment of the present disclosure, after obtaining the processing result from the natural language processing module, the processor 110 inputs at least one keyword included in the processing result into the word embedding model to obtain a word vector of the at least one keyword. there is. Thereafter, the processor 110 may calculate a similarity between a word vector of at least one keyword and a word vector of a plurality of sentiment words included in the sentiment word set. Calculation of similarity between word vectors may be performed based on, for example, distance or cosine similarity in vector space, as described above. The processor 110 may determine to change the score of one or more sentiment words having a word vector similar to a word vector of at least one keyword based on the similarity calculation result. An example of a change may include increasing the score of a sentiment word. One or more word vectors similar to the word vectors of the keyword may be determined as described above with reference to FIG. 5 .

In a further embodiment of the present disclosure, the processor 110 may additionally be based on a processing result of the natural language processing module when determining one or more emotional words to be subjected to a score change from text included in the social data. The processing result of the natural language processing module may include at least one of the extracted keyword and the importance level corresponding to the keyword. Specifically, for example, the processor 110 inputs the text of the social data (eg, “night snack is also chicken”) to the natural language processing module, and then obtains processing results including 'night snack' and 'chicken'. can The processor 110 may additionally obtain a processing result including a degree of importance (eg, ['night snack', 0.12] or ['chicken', 0.33], etc.) corresponding to each keyword from the natural language processing module. When a plurality of keywords are extracted from the text of the social data by the natural language processing module, the processor 110 may determine one or more sentiment words for the text of the social data based on the importance. For example, it is assumed that the processor 110 according to the present disclosure extracts top N sentiment words having a word vector with high similarity for each keyword extracted from text of social data. In this case, the top N emotional word set D (for example, 'pleasant', 'guilty', 'depressed', etc.) can In addition, for the second keyword (eg, 'chicken') extracted from the text of the same social data, the top N emotional word set E (eg, 'crispy', 'delicious', 'happy', etc.) can be extracted. there is. Furthermore, the processor 110 considers the importance of each keyword included in the processing result obtained from the natural language processing module, and the top N sentiment word set F having a word vector with high similarity to the word vector of the first keyword and the second keyword. can be decided In the process of the processor 110 determining the emotional word set F, the result of multiplying the emotional word set D by the importance (for example, 0.12) of the first keyword as a weight and the importance (for example, 0.33) of the second keyword A calculation process of summing the multiplication result of the emotional word set E as a weight may be included. The processor 110 according to the present disclosure may change the score for at least one sentiment word based on the determined score of the sentiment word set F. It will be apparent to those skilled in the art that the above-described examples are merely exemplary descriptions for the implementation of the present disclosure, and do not limit the present disclosure, and that operations on a plurality of keywords and emotional word sets extracted in various aspects can be performed. will be.

An example in which the processor 110 of the present disclosure calculates third sentiment information from text included in social data will be described. It is assumed that the processor 110 inputs the text “It is summer, but I am stuck at home because of the epidemic” into the natural language processing module. The processor 110 may obtain a processing result including keywords 'summer', 'epidemic', 'home', and 'stuck' from the natural language processing module. Thereafter, the processor 110 may obtain a word vector for each of 'summer', 'epidemic', 'home', and 'stuck' by inputting keywords included in the processing result into the word embedding model. The processor 110 may determine the most similar one or more emotional words by calculating the similarity between the word vector of each keyword and the word vector of a plurality of emotional words included in the emotional word set. Additionally, for example, suppose that the emotional word set contains the emotional words . At this time, the processor 110 may determine the emotional word having the word vector most similar to the word vector for the keyword 'summer' as 'active', and then increase the score of the emotional word 'active'. For example, increasing the score of a sentiment word in the present disclosure may include giving the sentiment word a relatively high weight or giving the sentiment word a relatively high priority over other sentiment words. . The processor 110 determines the emotional word having the word vector most similar to the word vector for the keyword 'epidemic' as 'depressed' and 'desperate', and then determines the emotional word score of 'depressed' and 'desperate' Each can be increased. Finally, the score distribution of the plurality of emotional words included in the emotional word set is ['gloomy':2, 'cozy':1, 'happy':1, 'desperate':1, 'bright':1, 'dark' ':1, 'lonely':3, 'active':1]. The processor 110 may calculate the third sentiment information based on the score distribution of the plurality of sentiment words included in the sentiment word set. The above-described example is merely an example and does not limit the present disclosure.

In an embodiment of the present disclosure, the step of the processor 110 calculating partial style information including a style expression for at least one social data may include: first emotional information, second emotional information, and third emotional information, respectively. The method may include calculating partial style information using values to which a weight according to the first emotion information, a weight according to the second emotion information, and a weight according to the third emotion information are applied.

In the present disclosure, the first emotion information may be treated as information about the emotion that a person receives from the color of the image included in the social data. The second emotion information may be treated as information about emotion that a person receives from an object included in the social data. The third emotion information may be treated as information about emotion that a person receives from the meaning of the text included in the social data. Based on the difference between the types of each emotion information as described above, the user may set different weights according to the first emotion information, the weights according to the second emotion information, and the weights according to the third emotion information.

For example, the first emotional information calculated from the social data is ['gloomy':0, 'cozy':4, 'happy':1, 'desperate':0, 'bright':1, 'dark': 0, 'lonely':2, 'active':0]. The second emotional information calculated from the social data is ['depressed':0, 'cozy':0, 'happy':3, 'desperate':0, 'bright':1, 'dark':2, 'lonely' ':0, 'active':1]. The third emotional information calculated from social data is ['gloomy':0, 'cozy':4, 'happy':1, 'desperate':0, 'bright':1, 'dark':0, 'loneliness' It can contain data such as ':2, 'active':0]. In this case, the weight according to the first emotion information, the weight according to the second emotion information, and the weight according to the third emotion information may be set to 3, 2, 1. The processor 110 may calculate partial style information of social data by multiplying and summing a weight according to the type of each emotional information for each emotional information. When additionally described based on the example above, partial style information is ['gloomy':0, 'cozy':16, 'happy':10, 'desperate':0, 'bright':5, 'dark': 4, 'lonely':8, 'active':2]. A weight according to each emotion information may be set as a ratio of an arbitrary value.

In an embodiment of the present disclosure, a result value obtained by applying a weight according to the third emotion information to the third emotion information may be compared with a predetermined upper limit value. When it is determined as a result of the comparison that the result value exceeds the upper limit value, the result value may be substituted with the upper limit value. For example, when a result value obtained by multiplying the third emotion information by a weight according to the third emotion information exceeds a predetermined upper limit value, the result value may be replaced with a predetermined upper limit value. In an embodiment, the predetermined upper limit value may be set based on a score of an emotional word. For example, a result value obtained by multiplying the third emotion information by a weight according to the third emotion information may be replaced with 10 when it exceeds 10 so that a maximum score of 10 or more cannot be obtained for one emotion word.

In another embodiment of the present disclosure, the weight according to the third emotion information may be adjusted by the processor 110 so that a result of multiplying the third emotion information does not exceed a predetermined upper limit value. For example, when the predetermined upper limit for each emotional word is 10, and the score of the highest emotional word in the entire emotional word set is 5, the processor 110 may set the weight according to the third emotional information to 2 or less. there is. The above-described example is merely an example and does not limit the present disclosure.

In the step of the processor 110 according to the present disclosure, calculating the style information of the social account from the partial style information of the at least one social data, a weight according to each social data based on attribute information corresponding to the at least one social data and calculating style information of a social account by applying a weight according to the social data to partial style information for at least one piece of social data. The attribute information corresponding to the social data may include feedback information of other users on the social data or time information of the social data. Specifically, the feedback information may include, for example, the number of 'likes', the number of comments, the number of views, and the number of clicks of other users on social data. The time information may include, for example, a creation time, a modification time, and the like.

In an embodiment of the present disclosure, the processor 110 may generate a weight according to the social data based on the feedback information for each social data in the plurality of social data. The feedback information may include, for example, the number of 'likes', the number of plays, and the number of comments, and the processor 110 generates a weight according to social data using at least one or more data among the number of 'likes', the number of plays, and the number of comments. can do.

For example, if the number of 'likes' for the first social data: 100, the number of 'likes' for the second social data: 50, and the number of 'likes' for the third social data: 50, the processor 110 may sum the partial style information for the first social data, the partial style information for the second social data, and the partial style information for the third social data in a ratio of 2:1:1.

For example, the number of 'likes' for the first social data: 100 - the number of comments: 0, the number of 'likes' for the second social data: 50 - the number of comments: 20, ' for the third social data Number of Likes: 50 - Number of Comments: 0 If the number of comments is 0, the processor 110 assigns a higher weight to the number of comments than the number of 'Likes' because the act of writing a comment corresponds to more active feedback information than the act of displaying 'Like'. It is possible to create weights according to social data. That is, when the proportion of 'likes' and comments is 1: 9 in generating weights according to social data, partial style information for the first social data, partial style information for the second social data, and the third social data are The partial style information for the stylus can be summed at a ratio of 100*1+0*9:50*1+30*9:50*1+0*9 (=10:32:5).

In another embodiment of the present disclosure, the processor 110 may generate a weight according to the social data based on time information of each social data in the plurality of social data. The processor 110 may generate a weight based on a difference value between the generation time of the social data and the current time. For example, if the difference between the generation time of the first social data and the current time is 1 day, and the difference between the generation time of the second social data and the current time is 7 days, the processor 110 determines the partial style for the first social data Information and partial style information for the second social data may be weighted and summed in a ratio reflecting 7:1 or an arbitrary length of time.

In an embodiment of the present disclosure, when the processor 110 calculates the style information of the social account from the partial style information of the at least one social data, the style information of the social account is the previously calculated style information of the at least one other social account. It may be additionally calculated based on statistical values related to style information. The statistical value may be a value calculated from the previously calculated style information of a plurality of other social accounts. The statistical value may include, for example, an average value, quartile value, variance, standard deviation, and the like for style information of a plurality of different social accounts.

For example, it is assumed that the calculated average values of style information of a plurality of different social accounts are equal to ['A':1, 'B':5, 'C':10]. In this case, A, B, and C mean emotional words, respectively. And when the style information of the first social account, which is the target for calculating the style information, is ['A':5, 'B':5, 'C':10], the processor 110 is The style information may be corrected in consideration of a difference from the average value of the style information of other social accounts. For example, the processor 110 may proportionally multiply the score for each emotional word included in the style information of the first social account by a certain constant C by the difference between the score and the average score, and then add it. That is, when C is 2, the style information of the first social account corrected in the above example may be expressed as ['A':13, 'B':5, 'C':10]. When the statistical value of the style information of a plurality of other social accounts calculated in this way is used in the process of calculating the style information of the social account that is the subject of the style information calculation, the unique individuality contrary to the overall trend can be further emphasized. there is

Although the present disclosure has been described above as being generally capable of being implemented by a computing device, those skilled in the art will appreciate that the present disclosure may be implemented in hardware and software in combination with computer-executable instructions and/or other program modules and/or in combination with computer-executable instructions and/or other program modules that may be executed on one or more computers. It will be appreciated that it can be implemented as a combination.

Generally, program modules include routines, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, those skilled in the art will appreciate that the methods of the present disclosure can be applied to single-processor or multiprocessor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, and the like. It will be appreciated that each of these may be implemented in other computer system configurations, including those capable of operating in connection with one or more associated devices.

The described embodiments of the present disclosure may also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Computers typically include a variety of computer-readable media. Any medium accessible by a computer can be a computer-readable medium, and such computer-readable media includes volatile and nonvolatile media, transitory and non-transitory media, removable and non-transitory media. including removable media. By way of example, and not limitation, computer-readable media may include computer-readable storage media and computer-readable transmission media. Computer readable storage media includes volatile and nonvolatile media, temporary and non-transitory media, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. includes media. A computer-readable storage medium may be RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device, or other magnetic storage device. device, or any other medium that can be accessed by a computer and used to store the desired information.

Computer readable transmission media typically embodies computer readable instructions, data structures, program modules or other data, etc. in a modulated data signal such as a carrier wave or other transport mechanism, and Includes all information delivery media. The term modulated data signal means a signal in which one or more of the characteristics of the signal is set or changed so as to encode information in the signal. By way of example, and not limitation, computer-readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer-readable transmission media.

An example environment 1100 implementing various aspects of the disclosure is shown including a computer 1102 , the computer 1102 including a processing unit 1104 , a system memory 1106 , and a system bus 1108 . do. A system bus 1108 couples system components, including but not limited to system memory 1106 , to the processing device 1104 . The processing device 1104 may be any of a variety of commercially available processors. Dual processor and other multiprocessor architectures may also be used as processing unit 1104 .

The system bus 1108 may be any of several types of bus structures that may further interconnect a memory bus, a peripheral bus, and a local bus using any of a variety of commercial bus architectures. System memory 1106 includes read only memory (ROM) 1110 and random access memory (RAM) 1112 . A basic input/output system (BIOS) is stored in non-volatile memory 1110, such as ROM, EPROM, EEPROM, etc., the BIOS is the basic input/output system (BIOS) that helps transfer information between components within computer 1102, such as during startup. contains routines. RAM 1112 may also include high-speed RAM, such as static RAM, for caching data.

The computer 1102 may also include an internal hard disk drive (HDD) 1114 (eg, EIDE, SATA) - this internal hard disk drive 1114 may also be configured for external use within a suitable chassis (not shown). Yes—a magnetic floppy disk drive (FDD) 1116 (eg, for reading from or writing to removable diskette 1118), and an optical disk drive 1120 (eg, a CD-ROM) for reading from, or writing to, disk 1122, or other high capacity optical media, such as DVD. The hard disk drive 1114 , the magnetic disk drive 1116 , and the optical disk drive 1120 are connected to the system bus 1108 by the hard disk drive interface 1124 , the magnetic disk drive interface 1126 , and the optical drive interface 1128 , respectively. ) can be connected to The interface 1124 for implementing an external drive includes at least one or both of Universal Serial Bus (USB) and IEEE 1394 interface technologies.

These drives and their associated computer-readable media provide non-volatile storage of data, data structures, computer-executable instructions, and the like. In the case of computer 1102, drives and media correspond to storing any data in a suitable digital format. Although the description of computer readable media above refers to HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs, those skilled in the art will use zip drives, magnetic cassettes, flash memory cards, cartridges, etc. It will be appreciated that other tangible computer-readable media such as etc. may also be used in the exemplary operating environment and any such media may include computer-executable instructions for performing the methods of the present disclosure.

A number of program modules may be stored in the drive and RAM 1112 , including an operating system 1130 , one or more application programs 1132 , other program modules 1134 , and program data 1136 . All or portions of the operating system, applications, modules, and/or data may also be cached in RAM 1112 . It will be appreciated that the present disclosure may be implemented in various commercially available operating systems or combinations of operating systems.

A user may enter commands and information into the computer 1102 via one or more wired/wireless input devices, for example, a pointing device such as a keyboard 1138 and a mouse 1140 . Other input devices (not shown) may include a microphone, IR remote control, joystick, game pad, stylus pen, touch screen, and the like. Although these and other input devices are connected to the processing unit 1104 through an input device interface 1142 that is often connected to the system bus 1108, parallel ports, IEEE 1394 serial ports, game ports, USB ports, IR interfaces, It may be connected by other interfaces, etc.

A monitor 1144 or other type of display device is also coupled to the system bus 1108 via an interface, such as a video adapter 1146 . In addition to the monitor 1144, the computer typically includes other peripheral output devices (not shown), such as speakers, printers, and the like.

Computer 1102 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1148 via wired and/or wireless communications. Remote computer(s) 1148 may be workstations, computing device computers, routers, personal computers, portable computers, microprocessor-based entertainment devices, peer devices, or other common network nodes, and are typically connected to computer 1102 . Although it includes many or all of the components described for it, only memory storage device 1150 is shown for simplicity. The logical connections shown include wired/wireless connections to a local area network (LAN) 1152 and/or a larger network, eg, a wide area network (WAN) 1154 . Such LAN and WAN networking environments are common in offices and companies, and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to a worldwide computer network, for example, the Internet.

When used in a LAN networking environment, the computer 1102 is coupled to the local network 1152 through a wired and/or wireless communication network interface or adapter 1156 . Adapter 1156 may facilitate wired or wireless communication to LAN 1152 , which LAN 1152 also includes a wireless access point installed therein for communicating with wireless adapter 1156 . When used in a WAN networking environment, the computer 1102 may include a modem 1158, be connected to a communication computing device on the WAN 1154, or establish communications over the WAN 1154, such as over the Internet. have other means. A modem 1158 , which may be internal or external and a wired or wireless device, is coupled to the system bus 1108 via a serial port interface 1142 . In a networked environment, program modules described for computer 1102 , or portions thereof, may be stored in remote memory/storage device 1150 . It will be appreciated that the network connections shown are exemplary and other means of establishing a communication link between the computers may be used.

The computer 1102 may be associated with any wireless device or object that is deployed and operates in wireless communication, for example, a printer, scanner, desktop and/or portable computer, portable data assistant (PDA), communication satellite, wireless detectable tag. It operates to communicate with any device or place, and phone. This includes at least Wi-Fi and Bluetooth wireless technologies. Accordingly, the communication may be a predefined structure as in a conventional network or may simply be an ad hoc communication between at least two devices.

Wi-Fi (Wireless Fidelity) makes it possible to connect to the Internet, etc. without a wired connection. Wi-Fi is a wireless technology such as cell phones that allows these devices, eg, computers, to transmit and receive data indoors and outdoors, ie anywhere within range of a base station. Wi-Fi networks use a radio technology called IEEE 802.11 (a, b, g, etc) to provide secure, reliable, and high-speed wireless connections. Wi-Fi can be used to connect computers to each other, to the Internet, and to wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks may operate in unlicensed 2.4 and 5 GHz radio bands, for example at 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual band). .

One of ordinary skill in the art of this disclosure will understand that information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, instructions, information, signals, bits, symbols, and chips that may be referenced in the above description are voltages, currents, electromagnetic waves, magnetic fields or particles, optical field particles or particles, or any combination thereof.

Those of ordinary skill in the art of the present disclosure will recognize that the various illustrative logical blocks, modules, processors, means, circuits, and algorithm steps described in connection with the embodiments disclosed herein include electronic hardware, (convenience For this purpose, it will be understood that it may be implemented by various forms of program or design code (referred to herein as software) or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. A person skilled in the art of the present disclosure may implement the described functionality in various ways for each specific application, but such implementation decisions should not be interpreted as a departure from the scope of the present disclosure.

The various embodiments presented herein may be implemented as methods, apparatus, or articles of manufacture using standard programming and/or engineering techniques. The term article of manufacture includes a computer program, carrier, or media accessible from any computer-readable storage device. For example, computer-readable storage media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash drives. memory devices (eg, EEPROMs, cards, sticks, key drives, etc.). Also, various storage media presented herein include one or more devices and/or other machine-readable media for storing information.

It is to be understood that the specific order or hierarchy of steps in the presented processes is an example of exemplary approaches. Based on design priorities, it is to be understood that the specific order or hierarchy of steps in the processes may be rearranged within the scope of the present disclosure. The appended method claims present elements of the various steps in a sample order, but are not meant to be limited to the specific order or hierarchy presented.

The description of the presented embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the scope of the present disclosure. Thus, the present disclosure is not intended to be limited to the embodiments presented herein, but is to be construed in the widest scope consistent with the principles and novel features presented herein.

As described above, the relevant contents have been described in the best mode for carrying out the invention.

The present invention may be used in a computing device that calculates style information of social data.

Claims

A method of calculating style information comprising a style representation of a social account performed by a computing device comprising at least one processor, the method comprising:

calculating partial style information including a style expression for at least one social data included in the social data set; and

calculating style information of a social account from partial style information on the at least one social data based on attribute information corresponding to the at least one social data;

containing,

How to calculate style information for social accounts.
The method of claim 1,

Calculating partial style information including a style expression for the at least one social data may include:

calculating first sentiment information from the social data based on a sentiment word set and a color matching sentiment scale;

calculating second sentiment information from the social data based on a detection result obtained by inputting the emotional word set and the image included in the social data into an object detection module; or

calculating third emotional information from the social data based on a processing result obtained by inputting the emotional word set and the text included in the social data into a natural language processing module;

comprising at least one of

How to calculate style information for social accounts.
3. The method of claim 2,

The emotional word set is,

Containing one or more emotional words included in the color matching emotional scale,

How to calculate style information for social accounts.
3. The method of claim 2,

Calculating the first emotional information from the social data includes:

calculating a color chip corresponding to the representative partial image on the color matching emotional scale by comparing distances between a plurality of color chips included in the color matching emotional scale and a representative partial image of the image included in the social data; and

calculating first emotion information based on the emotion word mapped to the color chip corresponding to the representative partial image;

containing,

How to calculate style information for social accounts.
3. The method of claim 2,

Calculating the first emotional information from the social data includes:

calculating a color chip corresponding to the first partial image on the color matching emotional scale by comparing distances between a plurality of color chips included in the color matching emotion scale and a first partial image of the image included in the social data;

calculating a color chip corresponding to a second partial image on the color matching emotional scale by comparing distances between a plurality of color chips included in the color matching emotional scale and a second partial image of the image included in the social data;

determining to change the score of the first emotional word mapped to the color chip corresponding to the first partial image with respect to the plurality of emotional words included in the emotional word set;

determining to change the score of a second sentiment word mapped to a color chip corresponding to the second partial image with respect to a plurality of sentiment words included in the sentiment word set; and

calculating first sentiment information based at least in part on the score of the first sentiment word and the score of the second sentiment word included in the sentiment word set;

containing,

How to calculate style information for social accounts.
6. The method according to claim 4 or 5,

The color chip corresponding to the partial image,

calculated based on a distance between an RGB value of at least one color included in the color chip and an RGB value of at least one pixel included in the partial image,

How to calculate style information for social accounts.
3. The method of claim 2,

Calculating the second emotional information from the social data includes:

obtaining a detection result by inputting an image included in the social data into an object detection module;

obtaining a word vector of the at least one object by inputting the name of the at least one object included in the detection result into a word embedding model;

calculating a similarity between a word vector of the at least one object and a word vector of each of a plurality of emotional words included in the emotional word set;

determining to change the score of one or more sentiment words having a word vector similar to the word vector of the at least one object based on the similarity calculation result; and

calculating second emotional information based on scores of a plurality of emotional words included in the emotional word set;

containing,

How to calculate style information for social accounts.
3. The method of claim 2,

Calculating the third emotional information from the social data includes:

obtaining a processing result obtained by inputting text included in the social data into a natural language processing module;

obtaining a word vector of the at least one keyword by inputting at least one keyword included in the processing result into a word embedding model;

calculating a similarity between a word vector of the at least one keyword and a word vector of a plurality of sentiment words included in the sentiment word set;

determining to change the score of one or more sentiment words having a word vector similar to a word vector of the at least one keyword based on the similarity calculation result; and

calculating third emotional information based on scores of a plurality of emotional words included in the emotional word set;

containing,

How to calculate style information for social accounts.
The method of claim 1,

Calculating partial style information including a style expression for the at least one social data includes:

Partial style information is obtained by applying a weight according to the first emotion information, a weight according to the second emotion information, and a weight according to the third emotion information to each of the first emotion information, the second emotion information, and the third emotion information. calculating;

including,

When the result value to which the weight according to the third emotion information is applied to the third emotion information exceeds a predetermined upper limit value, the result value is replaced with a predetermined upper limit value,

How to calculate style information for social accounts.
The method of claim 1,

Calculating the style information of the social account from the partial style information of the at least one social data includes:

generating a weight according to each social data based on attribute information corresponding to the at least one social data; and

calculating style information of a social account by applying a weight according to the social data to partial style information of the at least one social data;

containing,

How to calculate style information for social accounts.
11. The method of claim 10,

The attribute information corresponding to the social data is,

Including feedback information of other users on the social data or time information of the social data,

How to calculate style information for social accounts.
The method of claim 1,

The style information of the social account,

calculated based on additionally calculated statistical values related to style information of at least one other social account,

How to calculate the user's style information.
The method of claim 1,

Calculating partial style information including a style expression for the at least one social data includes:

calculating first emotion information from the image included in the social data based on the emotion word set and the color matching emotion scale;

calculating second sentiment information based on a detection result obtained by inputting the emotional word set and the image included in the social data into an object detection module; and

calculating third sentiment information based on a processing result obtained by inputting the emotional word set and the text included in the social data into a natural language processing module;

containing,

How to calculate style information for social accounts.
A computer program stored in a computer readable storage medium, wherein the computer program, when executed on one or more processors, performs the following operations for calculating style information including a style representation of a social account, the operation heard:

calculating partial style information including a style expression for at least one social data included in the social data set; and

calculating style information of a social account from partial style information on the at least one social data based on attribute information corresponding to the at least one social data;

containing,

A computer program stored in a computer-readable storage medium.
An apparatus for calculating style information including a style representation of a social account, comprising:

one or more processors;

Memory; and

network department;

including, and

The one or more processors,

Calculating partial style information including a style expression for at least one social data included in the social data set,

calculating style information of a social account from partial style information on the at least one social data based on attribute information corresponding to the at least one social data;

A device for calculating style information for social accounts.