KR101931624B1 - Trend Analyzing Method for Fassion Field and Storage Medium Having the Same - Google Patents

Abstract

The trend analysis method of the fashion field according to the present invention is a trend analysis method of a fashion field performed through a management server installed with a program for analyzing trends of a fashion field stored in a storage medium, (B) extracting a plurality of emotional vocabularies included in the fashion field from the atypical text data, (c) analyzing the plurality of emotional vocabularies, and and (d) analyzing trends in the fashion field through the analysis result of step c).

Description

TECHNICAL FIELD [0001] The present invention relates to a trend analysis method for a fashion field,

The present invention relates to a trend analysis method in a fashion field and a storage medium including the same, and more particularly, to extract and analyze a plurality of emotional vocabularies included in a fashion field to accurately analyze and predict fashion trends And a storage medium including the same.

Recently, the Internet has been actively used, and various online services have been provided by the appearance of smart phones and the like. In addition, various related services are being researched and developed to provide users with such services.

In addition, service providers need to accurately determine the needs and demands of users according to the increasingly diverse needs of users. To do so, algorithms are being studied to determine the demand and demand of users more accurately.

However, in the case of the fashion / apparel field, it is very difficult to analyze the unstructured text existing on the online service because the majority of emotional and quantifiable data is formed.

For this reason, up to now, there has been little developed an analysis method capable of precisely predicting trends in the fashion / apparel field, and there is a problem that the accuracy of a few existing analysis systems is greatly reduced.

Therefore, a method for solving the above problems is required.

The trend analysis method of the fashion field according to the present invention has an object to precisely analyze and predict trends in the fashion field through effective analysis of emotional vocabulary.

The solution of the present invention is not limited to those mentioned above, and other solutions not mentioned can be clearly understood by those skilled in the art from the following description.

The trend analysis method of the fashion field according to the present invention is a trend analysis method of a fashion field performed through a management server installed with a program for analyzing trends of a fashion field stored in a storage medium, (B) extracting a plurality of emotional vocabularies included in the fashion field from the atypical text data, (c) analyzing the plurality of emotional vocabularies, and and (d) analyzing trends in the fashion field through the analysis result of step c).

In the step (b), an emotional vocabulary of a semantic unit including at least one word can be extracted.

The step (c) may further comprise: (c-1) extracting a standard term from among the plurality of emotional words extracted by the step (b) (C-2) comparing the vocabulary with the standard term to select a standard term having a similar meaning to the sensible vocabulary, and classifying the sensible vocabulary into the selected standard term.

And (c-1) extracting a standard term from a plurality of emotional vocabularies collected in a public channel of an online service, wherein the (c-2) The collected emotional vocabulary may be compared with the standard term extracted in the step (c-1).

Further, after the step (c-2), the step (c-3) may be performed to assign a weight to the classified standard terms.

In the step (c-3), at least any one of frequency, core degree, and polarity of the standard term may be applied as a weight.

In the step (c-3), the core diagram of the standard term is applied as a weight. In the step (c-3), a centroid index of the standard term is calculated (c-3-1) (C-3-2) comparing the centroid index with a preset reference, and (c-3-3) selecting the standard term as a keyword when the centroid index is greater than a predetermined reference .

The step (c-3-1) further comprises: (c-3-1-1) constructing a standard term tree using a plurality of standard terms, (C-3-1-2) calculating a shortest path number between a pair of arbitrary other nodes, calculating a shortest path number via the corresponding standard term among the shortest paths between the calculated pair of arbitrary nodes (C-3-1-3) calculating the individual centroid by dividing the number of the shortest paths between the pair of arbitrary nodes by the number of the shortest paths passing through the standard term (c-3-1-4) Step < / RTI >

The step (c-3-1) may further include repeating the steps (c-3-1-1) to (c-3-1-4) (c-3-1-6) calculating the centrality index by summing the plurality of individual centrality calculated in the step (c-3-1-4).

The trend analysis method and the storage medium including the same according to the present invention have the following effects.

First, there is an advantage of being able to analyze and predict trends of fashion / apparel field accurately and reasonably.

Second, there is an advantage that the emotional text can be analyzed quantitatively and the user can easily understand it.

Third, users have the advantage of being able to utilize them for active marketing through analysis results.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

FIG. 1 is a flowchart illustrating an entire process of a fashion trend analysis method according to an embodiment of the present invention.
FIG. 2 is a flowchart illustrating a detailed process of a sensory lexical analyzing step in a trend analysis method of a fashion field according to an embodiment of the present invention.
3 is a flowchart illustrating a detailed process of applying a weight according to a core diagram in a fashion trend analysis method according to an embodiment of the present invention.
FIG. 4 is a flowchart illustrating a process of calculating a centroid index in a fashion trend analysis method according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating a related word tree in a fashion trend analysis method according to an embodiment of the present invention. Referring to FIG.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. In describing the present embodiment, the same designations and the same reference numerals are used for the same components, and further description thereof will be omitted.

The trend analysis method of the fashion field according to the present invention is performed through a management server installed with a program for trend analysis of a fashion field stored in a storage medium and can be installed in the management server and driven by the processor of the management server.

The trend analysis program of the fashion field driven thereby can be output through a video output device such as a display module, and can provide visual information to a user through a visualized graphical user interface.

In particular, the storage medium storing the trend analysis program of the fashion field may be installed in the management server using a mobile disk or a communication network, and the trend analysis program of the fashion field may be configured to allow the management server to be operated by various functional means . That is, in the present invention, information processing by software is realized through hardware.

Hereinafter, an algorithm of a trend analysis program in the fashion field of the present invention executed through the management server will be described.

FIG. 1 is a flowchart illustrating an entire process of a fashion trend analysis method according to an embodiment of the present invention.

As shown in FIG. 1, a fashion trend analysis method according to an embodiment of the present invention includes: (a) collecting atypical text data for an online service and storing the collected atypical text data in the management server; (B) extracting a plurality of emotional vocabularies included in the fashion field from the data, analyzing the plurality of emotional vocabularies, and analyzing the trends of the fashion field according to the analysis result of the (c) And (d) predicting.

In step (a), unregistered text data is extracted from various online services, collected, and stored in a management server.

Here, the online service may be various online service media using a communication network such as the web, SNS, etc., and is not limited to any online service. In the case of the above-mentioned unstructured text data, it refers to data of document type which is complicated in its structure and is not formatted.

That is, in this step, data of a document type is extracted for various online services, and then step (b) is performed.

In the step (b), a plurality of emotional vocabularies included in the fashion field are extracted from the atypical text data. In this step, a plurality of vocabularies related to the fashion field are extracted from the extracted atypical text data. At this time, representative vocabularies including representative vocabulary such as nouns and adverbs, Respectively.

In particular, in the case of the present embodiment, the step (b) may extract the emotional vocabulary of the semantic unit including at least one word. That is, not only the emotional vocabulary that is only one word, but also the emotional vocabulary that combines multiple words and express a meaning is also included in the object.

For example, the fashion related noun 'color, texture, touch, charm; As well as a list of individual vocabulary units such as 'urban, sensual, modern, active, attractive', verb and adjective 'sophisticated, sexy, elegant, unique, elegant, feminine' I would like to buy an emotional vocabulary that has a combination of multiple words such as "I like to see, I want to buy, I like".

The step (c) is a step of analyzing a plurality of emotional vocabularies extracted by the step (b). Various analysis techniques can be applied at this stage. In this embodiment, a method of classifying the extracted emotional vocabulary according to a standard term standard can be used.

The reason for doing this is that since the fashion field generally includes many foreign languages / foreign words, users can use rather distorted expressions instead of using only the correct expressions according to the standard language. Also, since users frequently use spoken language and internet terms on the online service, it is necessary to classify these emotional vocabularies uniformly according to certain criteria.

Specifically, in the present embodiment, the step (c) includes: (c-1) extracting standard terms among the plurality of emotional vocabulary extracted as shown in FIG. 2; (C-2) comparing the remaining emotional vocabularies with the standard term, selecting standard terms having a similar meaning to the emotional vocabulary, classifying the emotional vocabulary into the selected standard terms, and And (c-3) weighting the standard term that has been generated.

In the step (c-1), the emotional vocabulary similar to the standard word or the standard word among the plurality of extracted emotional vocabularies is separately extracted and defined as a standard term. In order to facilitate such a process, it is possible to extract standard terms from a plurality of emotional vocabularies collected from a public channel among online services.

The public channel refers to an online service field, such as an article, an advertisement, etc., which generally uses a standard language at a high frequency. That is, in the case of the public channel, the extracted emotional vocabulary can be classified into the standard terms with a high probability. Therefore, the emotional vocabulary collected in the public channel is extracted as a standard term as described above.

In the case of the step (c-2), the meaning of the remaining emotional vocabulary other than the standard term extracted by the step (c-1) is analyzed and included in each standard term selected.

That is, in this step, the emotional vocabulary collected on the online service of the remaining emotional vocabulary of the public channel and the private channel such as a blog or a shopping mall is compared with the standard term extracted in the step (c-1) .

Through the above process, the extracted emotional vocabulary can be classified into a uniform standard term form in a certain format.

Next, a step (c-3) of weighting the standard terms classified for the trend analysis of the fashion field is performed.

In the step (c-3), at least one of the frequency, the core degree and the polarity of the standard term may be applied as a weight. Here, the frequency refers to the frequency of use of the standard term, and the polarity represents the degree of the positive / negative degree of the standard term. In the case of the polarity, a methodology such as a 5-step / 7-step Ricard scale can be applied for analysis.

On the other hand, in the case of the core diagram, the centroid index of each emotional vocabulary is used.

3 is a flowchart illustrating a detailed process of applying a weight according to a core diagram in a trend analysis method of a fashion field according to an embodiment of the present invention.

3, the step (c-3) includes: (c-3-1) calculating a centroid index of a predetermined standard term among a plurality of standard terms; and (C-3-2) comparing the centroid index with a predetermined standard, and (c-3-3) selecting the standard term as a key word when the centroid index is equal to or greater than a predetermined reference.

The step (c-3-1) is a step of selecting a standard term among a plurality of standard terms classified by the step (c-2) and calculating a centroid index therefor. In other words, in this step, the centroid index for each extracted standard term is calculated, so that each standard term is given a centroid index. A method of calculating the centrality index will be described later.

The step (c-3-2) is a step of comparing the centroid index of each standard term with a predetermined criterion. The setting criterion is a minimum criterion for the selected standard term to be selected as a keyword, and it is compared whether the centroid index of each standard term calculated in the step (c-3-1) meets the setting criterion.

The step (c-3-3) is a step of selecting a target word as a key word when the centroid index of the selected standard term is greater than a predetermined reference as a result of the comparison in the step (c-3-2). That is, when the centroid index of the standard term is equal to or exceeds the predetermined standard, it is determined that the keyword satisfies the requirements of the keyword, and the keyword is selected as the keyword.

In this embodiment, key words for trend analysis in the fashion field can be extracted in the same manner as described above. Hereinafter, the process of calculating the centrality index in the step (c-3-1) will be described in detail.

FIG. 4 is a flowchart illustrating a detailed process of calculating a centrality index in a fashion trend analysis method according to an exemplary embodiment of the present invention. FIG. 4 is a flowchart illustrating a trend analysis of a fashion field according to an exemplary embodiment of the present invention , And a related word tree.

4, the step (c-3-1) includes the steps of (c-3-1-1) constructing a standard term tree using a plurality of classified standard terms, (C-3-1-2) calculating, from the tree, a shortest path number between a pair of arbitrary nodes other than the node in which the standard term is located; calculating a shortest path (C-3-1-3) calculating a shortest path number via the standard term among the plurality of standard nodes, and dividing the number of the shortest paths between the pair of arbitrary nodes by the number of the shortest paths passing through the standard term (C-3-1-4) repeating the steps (c-3-1-1) to (c-3-1-4) ), And calculating (c-3-1-6) a centroid index by summing the plurality of individual centrality calculated in the (c-3-1-4) step.

In the case of the step (c-3-1-1), a standard term tree is constructed using a plurality of standard terms classified in the step (c-2). In this process, a connection network of all standard terms is formed in consideration of various factors such as the relation among the standard terms. FIG. 5 shows an example of such a standard term tree. As shown in FIG. 4, each of the standard terms W1 to W8 forms one node and forms a connection network connected to each other in a net form.

In the step (c-3-1-2), the number of the shortest paths between a pair of arbitrary nodes other than the node in which the selected standard term is located is calculated from the standard term tree.

For example, referring to FIG. 5, if the selected standard term is W2, the shortest path number between any pair of all other nodes except W2 is calculated. W1-W3-W6, W1-W2-W7-W6, and W1-W3-W6 when the pair of nodes arbitrarily selected are W1 and W6 and the shortest distance among the paths connecting W1 and W6, W5-W6, and W1-W2-W5-W6.

In the step (c-3-1-3), the shortest path number among the shortest paths among the calculated pairs of arbitrary nodes is calculated.

For example, the shortest path including the selected standard term W2 among the four shortest paths calculated in step (c-3-1-2) described above with reference to FIG. 5 is W1-W2-W4-W6, W1-W2-W7-W6, and W1-W2-W5-W6.

In the step (c-3-1-4), the individual centrality is calculated by dividing the number of the shortest paths between a pair of arbitrary nodes by the number of the shortest paths passing through the standard term. That is, in the example of FIG. 5, since the shortest paths including four total shortest paths and W2 including the double selected standard terms are three, the individual centrality of W2 will be 3/4.

The step (c-3-1-5) repeats the steps (c-3-1-1) to (c-3-1-4) described above. That is, since the shortest path between W1 and W6 is calculated in the step (c-3-1-2) of FIG. 5, in this step, the selection of one of the nodes W1 to W8 (C-3-1-1) to (c-3-1-4).

Wherein the step (c-3-1-6) further comprises the step of summing the plurality of individual centrality calculated in the step (c-3-1-4) by the step (c-3-1-5) . In other words, in this step, all the individual centrality calculated for the number of all cases is summed up, which becomes the final centrality index.

Therefore, in the step (c-3-2), the final centroid index of the corresponding standard term is compared with predetermined criteria. If the standard criterion is not less than the predetermined standard, the standard term is used as a keyword in the step (c-3-3) Respectively.

Step (c) has been described in detail, and then step (d) for predicting trends in the fashion field is performed according to the analysis result of step (c).

In this step, a process of analyzing the trend of the fashion field is performed through the emotional vocabulary to which the analysis result of the step (c), that is, at least one of the frequency, the core degree and the polarity is given as the weight.

In this step, the selection of high-frequency emotional vocabulary in unstructured data, the selection of emotional vocabulary that is statistically significant for each channel / subject, the frequency and use field, and the selection of basic vocabulary for emotion in the field of fashion This allows precise analysis and prediction of fashion trends.

As described above, the present invention is advantageous in that it quantitatively analyzes sensible text rationally and realistically, and accurately analyzes and predicts trends in the fashion / apparel field.

Then, the analysis result can be visualized and provided to the user from the management server, and the user can utilize the analysis result displayed through the user terminal for aggressive marketing.

The embodiments and the accompanying drawings described in the present specification are merely illustrative of some of the technical ideas included in the present invention. Accordingly, the embodiments disclosed herein are for the purpose of describing rather than limiting the technical spirit of the present invention, and it is apparent that the scope of the technical idea of the present invention is not limited by these embodiments. It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (10)

1. A trend analysis method for a fashion field performed through a management server having a program for analyzing trends of a fashion field stored in a storage medium,
(A) collecting unstructured text data on an online service and storing the collected unstructured text data in the management server;
(B) extracting a plurality of emotional vocabularies included in the fashion field from the atypical text data;
(C) analyzing the plurality of emotional vocabularies; And
Analyzing trends in the fashion field through the analysis result of step (c);
/ RTI >
The step (c)
(C-1) extracting a standard term from a plurality of emotional vocabulary extracted by the step (b); And
Comparing the remaining emotional vocabulary from the standard term extracted in the step (c-1) with the standard term, selecting a standard term having a similar meaning to the emotional vocabulary, classifying the emotional vocabulary into the selected standard term (C-2);
A trend analysis method in the fashion field. The method according to claim 1,
The step (b)
A method for analyzing a trend in a fashion field in which an emotional vocabulary of a semantic unit including at least one word is extracted. delete The method according to claim 1,
The step (c-1)
It is assumed that standard terms are extracted from a plurality of emotional vocabularies collected from a public channel among online services,
The step (c-2)
And comparing the remaining emotional vocabulary of the public channel and the emotional vocabulary collected in the private channel with the standard term extracted in the step (c-1). The method according to claim 1,
After the step (c-2)
Wherein the step (c-3) of weighting the classified standard terms is further performed. 6. The method of claim 5,
The step (c-3)
Wherein at least one of the frequency, the core degree and the polarity of the standard term is applied as a weight. The method according to claim 6,
In the step (c-3), the core diagram of the standard term is applied as a weight,
The step (c-3) includes: (c-3-1) calculating a centroid index of the standard term;
(C-3-2) comparing the centroid index with a preset reference; And
(C-3-3) selecting the standard term as a key word when the centroid index is equal to or greater than a preset reference;
A trend analysis method in the fashion field. 8. The method of claim 7,
The step (c-3-1)
(C-3-1-1) constructing a standard term tree using a plurality of standard terms;
(C-3-1-2) calculating, from the standard term tree, a shortest path number between a pair of arbitrary nodes other than the node where the standard term is located;
(C-3-1-3) calculating the shortest path number via the corresponding standard term among the shortest paths among the calculated pairs of arbitrary nodes; And
(C-3-1-4) calculating individual centrality by dividing the number of the shortest paths between the pair of arbitrary nodes by the number of the shortest paths passing through the standard term;
A trend analysis method in the fashion field. 9. The method of claim 8,
The step (c-3-1)
(C-3-1-1) repeating the steps (c-3-1-1) to (c-3-1-4); And
(C-3-1-6) calculating a centrality index by summing the plurality of individual centrality calculated in the step (c-3-1-4);
A trend analysis method of a fashion field. A computer-readable storage medium storing a program for causing a computer to execute the trend analysis method of any one of claims 1, 2, 4 to 9.

Images