KR-20260063669-A - Training method of neural network model capable of evaluating online content and content evaluation method using the same

KR20260063669AKR 20260063669 AKR20260063669 AKR 20260063669AKR-20260063669-A

Abstract

The present invention relates to a method for training a neural network model to evaluate content based on various attributes of online content, and a method for evaluating content and suggesting content modifications using the same. A method for training a neural network model for content evaluation according to one embodiment of the present invention is characterized by comprising the steps of: extracting multivariate features from multiple contents collected from the web; calculating an evaluation score for each content based on search results for each tag included in the multiple contents; and training a neural network model using a training dataset composed of the multivariate features and the evaluation score.

Inventors

장민서

Assignees

장민서

Dates

Publication Date: 20260507
Application Date: 20241030

Claims (14)

A step of extracting multivariate features from multiple contents collected from the web; A step of calculating an evaluation score for each content based on search results for each tag included in the above multiple contents; and A step comprising training a neural network model using a training dataset composed of the above multivariate features and the above evaluation scores. Training method for a neural network model for content evaluation.
In paragraph 1, The above multivariate features are determined according to at least one of the following: the publication period of the content, keywords in the title and body, sentences, vocabulary, average sentence length, consistency, sentiment, number of images, field, and number of tags. Training method for a neural network model for content evaluation.
In paragraph 1, The above extraction step includes a step of extracting title features based on whether a keyword included in the title of the content appears among multiple keywords stored in a database. Training method for a neural network model for content evaluation.
In paragraph 1, The above extraction step includes a step of extracting text features based on the frequency with which keywords included in the text of each content appear in all texts of the multiple contents. Training method for a neural network model for content evaluation.
In paragraph 4, The above extraction step includes the step of extracting the TF-IDF (Term Frequency - Inverse Document Frequency) values of keywords included in the body of the content as body features. Training method for a neural network model for content evaluation.
In paragraph 1, The above extraction step includes a step of extracting vocabulary features based on the difficulty of keywords included in the body of the content. Training method for a neural network model for content evaluation.
In paragraph 1, The above extraction step includes a step of extracting consistency features based on the average similarity between adjacent sentences included in the body of the content. Training method for a neural network model for content evaluation.
In paragraph 1, The above extraction step is, A step of identifying the sentiment class for each sentence by inputting each sentence included in the body of the above content into a Natural Language Model (NLP) that performs a sentiment classification task; and A step comprising extracting the average value of the sentiment class for each of the above sentences as a sentiment feature Training method for a neural network model for content evaluation.
In paragraph 1, The step of calculating the above evaluation score is, A step of searching for tags included in the above content on a platform where the above content is posted; and A step comprising calculating the evaluation score based on the degree to which the content is exposed in the search results. Training method for a neural network model for content evaluation.
In paragraph 1, The step of calculating the above evaluation score includes the step of calculating the above evaluation score according to the following [mathematical formula]. [Mathematical Formula] (Here, S c is the evaluation score, T is the total number of tags where the search result exists, k is the tag index, r k is the exposure rank of the content within the search result, and m is the number of search results within a single webpage of the platform where the content is published, and is a hyperparameter) Training method for a neural network model for content evaluation.
In paragraph 1, The above neural network model performs a regression task that receives the above multivariate features as input and outputs a predicted evaluation score. Training method for a neural network model for content evaluation.
In paragraph 1, The above training step includes the step of setting the multivariate features as input data for the neural network model and setting the evaluation score as output data for the neural network model to perform supervised learning on the neural network model. Training method for a neural network model for content evaluation.
A step of extracting multivariate features from target content; and A step comprising deriving an evaluation score for the target content by inputting the multivariate features into a neural network model trained according to any one of claims 1 to 12. Content evaluation method using a neural network model.
In Paragraph 13, A step of setting a control feature by changing at least one of the individual features constituting the multivariate feature within a preset range; A step of inputting the above control features into the neural network model to derive an expected score for the target content; and If the above predicted score is higher than the above evaluation score, the method includes the step of transmitting a message regarding the above modified individual feature to a user terminal. Content evaluation method using a neural network model.

Description

Training method of neural network model capable of evaluating online content and content evaluation method using the same The present invention relates to a method for training a neural network model to evaluate content based on various attributes of online content, and a method for evaluating content and suggesting content modifications using the same. With the rapid growth of the SNS (Social Network Service) market, posting content on blogs on portal sites or SNS platforms has become a daily routine, and a large amount of content for advertising or online product sales is also being posted on the web. In particular, since exposure to the public is very important for advertising or promotional content, various factors are considered to evaluate it; however, there are limitations in that the objectivity and accuracy of the evaluation cannot be guaranteed because existing methods mostly evaluate content based only on easily identifiable quantitative factors such as views, recommendations, comments, and the influence of the poster. FIG. 1 is a drawing illustrating an online content evaluation system according to an embodiment of the present invention. FIG. 2 is a flowchart illustrating a learning method for a neural network model for content evaluation according to an embodiment of the present invention. FIG. 3 is a diagram illustrating the process of extracting multivariate features from content. Figure 4 is a diagram illustrating the process of calculating evaluation scores by content. Figure 5 is a diagram illustrating the learning process of a neural network model. FIG. 6 is a flowchart illustrating a content evaluation method using a neural network model according to an embodiment of the present invention. FIG. 7 is a flowchart illustrating a method for proposing content modification according to an embodiment of the present invention. The aforementioned objectives, features, and advantages are described in detail below with reference to the attached drawings, thereby enabling those skilled in the art to easily implement the technical concept of the present invention. In describing the present invention, detailed descriptions of known technologies related to the present invention are omitted if it is determined that such descriptions would unnecessarily obscure the essence of the invention. Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings. In the drawings, the same reference numerals are used to indicate the same or similar components. In this specification, terms such as "first," "second," etc. are used to describe various components, but these components are not limited by these terms. These terms are used merely to distinguish one component from another, and unless specifically stated otherwise, the first component may be the second component. Additionally, in this specification, the statement that any configuration is disposed on the "upper (or lower)" or "upper (or lower)" of a component may mean not only that any configuration is disposed in contact with the upper (or lower) surface of said component, but also that another configuration may be interposed between said component and any configuration disposed on (or below) said component. Furthermore, where it is stated in this specification that one component is "connected," "coupled," or "connected" to another component, it should be understood that while the components may be directly connected or connected to each other, another component may be "interposed" between each component, or each component may be "connected," "coupled," or "connected" through another component. Additionally, singular expressions used in this specification include plural expressions unless the context clearly indicates otherwise. In this application, terms such as "composed of" or "comprising" should not be interpreted as necessarily including all of the various components or steps described in the specification, and should be interpreted as meaning that some of the components or steps may not be included, or that additional components or steps may be included. Additionally, in this specification, "A and/or B" means A, B, or A and B unless specifically stated otherwise, and "C to D" means C or more and D or less, unless specifically stated otherwise. The present invention relates to a method for training a neural network model to evaluate content based on various attributes of online content, and a method for evaluating content and suggesting content modifications using the same. FIG. 1 is a drawing illustrating an online content evaluation system according to one embodiment of the present invention. Referring to FIG. 1, a content evaluation system (1) according to one embodiment of the present invention may include a server (10) that collects, analyzes, and evaluates a plurality of contents (30) from the web, and a user terminal (20) that uploads target content to the server (10) or