Method and device for realizing creativity generation by aid of AI (Artificial Intelligence)

Abstract

The invention relates to the technical field of advertisement data processing, and discloses a method and device for achieving creativity generation by means of AI, and the method comprises the steps: obtaining user generation content from each social media platform in real time, and collecting hot topics of a current market; performing semantic analysis and sentiment analysis based on the hot topics to generate hot portraits; performing matching degree calculation on the hot portraits and target audience portraits of the product, and screening the hot portraits with the highest matching degree as the hot portraits of the target audience portraits; according to the hotspot portraits of the target audience portraits, performing two-dimensional screening in combination with the popularity of the hotspots to obtain target hotspot portraits; and inputting the target hotspot portrait and the associated product information into a preset deep learning model based on semantic matching, and outputting originality. The method has the advantages that the timeliness of the creativity is guaranteed, and the click rate and the conversion rate of the creativity are improved.

Description

Technical Field

[0001] The present application relates to the technical field of advertising data processing, and in particular to a method and device for realizing creative generation with the help of AI. Background Art

[0002] Currently, more and more companies and brands are using artificial intelligence (AI) to help generate ideas. Although AI can improve efficiency in generating ideas, it lags behind in capturing and utilizing hot content in the current market. Hot content is dynamic and changes rapidly, and relies on people's immediate attention and interests. AI models are usually based on past data and existing rules, which makes them inapplicable when dealing with the latest market trends and hot spots. As a result, AI-generated ideas often cannot be combined with hot content and lack timeliness.

[0003] With respect to the above-mentioned related technologies, the inventors found that the existing AI creative generation methods lack timeliness and affect click-through rates. Summary of the invention

[0004] In order to enable AI-generated creativity to be combined with hot content, ensure timeliness, and help improve the click-through rate of creativity, the present application provides a method and device for generating creativity with the help of AI.

[0005] In a first aspect, the present application provides a method for generating creative ideas with the help of AI.

[0006] This application is achieved through the following technical solutions:

[0007] A method for generating creative ideas with the help of AI, comprising the following steps:

[0008] Obtain user-generated content from various social media platforms in real time and collect hot topics in the current market;

[0009] Perform semantic analysis and sentiment analysis based on the hot topics to generate hot topic portraits;

[0010] Calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait;

[0011] According to the hot spot portrait of the target audience portrait, a dual-dimensional screening is performed in combination with the popularity of the hot spot to obtain a target hot spot portrait;

[0012] The target hot spot portrait and the associated product information are input into a preset deep learning model based on semantic matching to output creativity.

[0013] In a preferred example, the present application can be further configured as follows: based on the hot topic, semantic analysis and sentiment analysis are performed, and the steps of generating a hot topic portrait include:

[0014] Using natural language processing technology, perform word segmentation, part-of-speech tagging and named entity recognition on the hot topics to extract the core content of the hot topics;

[0015] Determine the theme and keywords based on the core content of the hot topic;

[0016] Counting user group attributes corresponding to the hot topics to identify target audiences and customize creative content;

[0017] Collect statistics on the dissemination channels and paths of the hot topics on different social media platforms to assist in optimizing content publishing platforms and dissemination strategies;

[0018] Track the entire life cycle of the hot topic from its emergence to its disappearance, determine the length attributes of the entire life cycle and the characteristics of each stage, so as to predict the duration of the hot topic and the best time to intervene;

[0019] Analyze the hot topics through time series to obtain the peak and trough periods of the hot topics for use in formulating marketing plans;

[0020] Evaluate the user's sentiment towards the hot topic, where the sentiment includes positive, negative and neutral, so as to adjust the brand communication strategy;

[0021] The core content, theme, keywords, user group attributes, communication channels, communication paths, length attributes of the entire life cycle and characteristics of each stage, popularity peaks, popularity troughs and emotional tendencies of the hot topics are summarized to form the hot topic portrait corresponding to the hot topic.

[0022] In a preferred example, the present application can be further configured as follows: the steps of calculating the matching degree between the hotspot portrait and the target audience portrait of the product, selecting the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait include:

[0023] Obtain the age, gender, interests, hobbies and consumption behavior of the target audience portrait;

[0024] Extracting age distribution characteristics, gender ratio characteristics, interest keyword characteristics and typical consumption behavior characteristics from each of the hot spot portraits and the target audience portraits;

[0025] Using a similarity algorithm, calculate the age distribution feature similarity, gender ratio feature similarity, interest keyword feature similarity, and typical consumption behavior feature similarity of each hotspot portrait and the target audience portrait respectively;

[0026] Based on the preset feature weights, weighted scores are performed on the age distribution feature similarity, the gender ratio feature similarity, the interest keyword feature similarity, and the typical consumer behavior feature similarity to obtain the matching degree of each hotspot portrait;

[0027] The matching degree of each hotspot portrait is compared with a preset matching degree threshold, and the hotspot portrait with the highest matching degree is determined as the hotspot portrait of the target audience portrait.

[0028] In a preferred example, the present application can be further configured as follows: according to the hot spot portrait of the target audience portrait, a dual-dimensional screening is performed in combination with the heat of the hot spot, and the step of obtaining the target hot spot portrait includes:

[0029] Counting the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait;

[0030] Calculate the hot spot attention of each hot spot portrait based on the search volume index and the number of social media mentions;

[0031] Counting the current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait since their appearance, and determining the hot spot duration of each hot spot portrait;

[0032] Obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence;

[0033] The hot spot portrait with the largest hot spot heat is selected from the hot spot portraits of the target audience portrait as the target hot spot portrait.

[0034] In a preferred example, the present application can be further configured as follows: it also includes the following steps:

[0035] When calculating the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait, the following formula is used:

[0036] S = current social media search volume / maximum search volume × 100;

[0037] M = current number of social media mentions / maximum number of mentions × 100;

[0038] Among them, S is the search volume index, and M is the number of social media mentions;

[0039] Based on the search volume index and the number of social media mentions, the following formula is used to calculate the hot spot attention of each hot spot portrait:

[0040] C = α × S + β × M;

[0041] Among them, C is the hot spot attention, α is the weight coefficient of the search volume index, and β is the weight coefficient of the number of mentions on social media;

[0042] The current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait is counted, and the hot spot duration of each hot spot portrait is determined by using the following formula:

[0043] T = current duration / maximum duration × 100;

[0044] Where T is the hotspot duration, and the unit of the current duration is day;

[0045] When obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence, the following formula is used:

[0046] I = γ × C + δ × T;

[0047] Among them, I is the heat of the hot spot, γ is the weight coefficient of the hot spot attention, and δ is the weight coefficient of the hot spot persistence.

[0048] In a preferred example, the present application can be further configured as follows: the target hot spot portrait and the associated product information are input into a preset deep learning model based on semantic matching, and the step of outputting creativity includes:

[0049] Based on the deep learning model of semantic matching, the target hot spot portrait is associated with the selling point of the product and the market demand of the product to generate creative ideas in the form of advertising copy, visual design and video script.

[0050] In a preferred example, the present application may be further configured as follows: inputting the target hot spot portrait and the associated product information into a preset semantic matching-based deep learning model, and after the step of outputting the creative idea, further comprising:

[0051] Compare and score the ideas in multiple dimensions to determine the target ideas.

[0052] In a preferred example, the present application can be further configured as follows: the creativity is compared and scored in multiple dimensions, and the step of determining the target creativity includes:

[0053] Use the preset prediction model to compare and score each creative idea in multiple dimensions;

[0054] Wherein, the prediction model includes an input layer, a feature extraction layer, a multi-dimensional evaluation layer and an output layer;

[0055] The input layer obtains and outputs historical creative effect data, market feedback data, audience response data, and current market trends and competitive environment data;

[0056] Based on the preset evaluation feature, the feature extraction layer extracts the creative feature associated with the evaluation feature from the output data of the input layer;

[0057] The multi-dimensional evaluation layer scores the creative features output by the feature extraction layer and calculates a comprehensive score, and then matches the evaluation report containing pros and cons analysis and optimal recommendation of the target creative according to the comprehensive score;

[0058] The output layer outputs the comprehensive score and the evaluation report of the target creativity by the multi-dimensional evaluation layer.

[0059] In a preferred example, the present application can be further configured as follows: the training step of the prediction model includes:

[0060] Inputting large-scale data sets in vertical fields into the prediction model for training;

[0061] Optimizing the model parameters of the prediction model through cross-validation and parameter adjustment;

[0062] When the prediction accuracy of the prediction model reaches a preset accuracy threshold, the current prediction model is output.

[0063] In a preferred example, the present application may be further configured as follows: the pros and cons analysis of the evaluation report includes describing the performance of the target creativity in each evaluation dimension;

[0064] The preferential recommendation of the evaluation report includes, based on the comprehensive score, selecting the idea corresponding to the largest comprehensive score as the target idea.

[0065] In a second aspect, the present application provides a device for generating creative ideas with the help of AI.

[0066] This application is achieved through the following technical solutions:

[0067] A device for realizing creative generation by means of AI, comprising:

[0068] The data module is used to obtain user-generated content from various social media platforms in real time and collect hot topics in the current market;

[0069] A hot topic portrait construction module is used to perform semantic analysis and sentiment analysis based on the hot topic to generate a hot topic portrait;

[0070] A target audience hotspot portrait module is used to calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait;

[0071] A target hotspot screening module is used to perform dual-dimensional screening based on the hotspot portrait of the target audience portrait and the heat of the hotspot to obtain a target hotspot portrait;

[0072] The creative generation module is used to input the target hot spot portrait and the associated product information into a preset deep learning model based on semantic matching to output creativity.

[0073] In a preferred example, the present application can be further configured as follows: further comprising:

[0074] The creative prediction module is used to compare and score the creative ideas in multiple dimensions to determine the target creative ideas.

[0075] In a third aspect, the present application provides a computer device.

[0076] This application is achieved through the following technical solutions:

[0077] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of any one of the above-mentioned methods for generating creativity with the help of AI are implemented.

[0078] In a fourth aspect, the present application provides a computer-readable storage medium.

[0079] This application is achieved through the following technical solutions:

[0080] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of any one of the above-mentioned methods for generating creativity with the help of AI.

[0081] In a fifth aspect, the present application provides a computer program product.

[0082] This application is achieved through the following technical solutions:

[0083] A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the steps of any one of the above methods for generating creativity with the help of AI are implemented.

[0084] In summary, compared with the prior art, the technical solution provided by this application has at least the following beneficial effects:

[0085] Acquire user-generated content from various social media platforms in real time, collect hot topics in the current market, and obtain hot content in the market in real time, which is conducive to timely capturing and utilizing the latest market trends and ensuring the timeliness of generating creativity; perform semantic analysis and sentiment analysis based on hot topics to generate hot profiles to provide a data basis for subsequent creative generation, customize creative content, adjust communication strategies, improve marketing effectiveness, and enhance brand image and user satisfaction; calculate the matching degree between the hot profile and the target audience portrait of the product, select the hot profile with the highest matching degree as the hot profile of the target audience portrait, so as to obtain hot spots that highly overlap with the product audience characteristics, which is conducive to the generated creativity to resonate with the target audience; perform two-dimensional screening based on the hot profile of the target audience portrait and the popularity of the hot spots to obtain the target Hot spot portraits are created to make the hot spots match the product audience and have high popularity in the current market, which is conducive to improving the click-through rate of creativity. Finally, the target hot spot portrait and the related product information are input into the preset deep learning model based on semantic matching, and creativity is output, so that the creative content can be associated with the product information, which can effectively convey the core selling points and brand value of the product, thereby enhancing the brand's market competitiveness and the audience's willingness to buy. The generated creativity can be combined with the current hot content and keep up with market trends. At the same time, hot spots can attract target audiences, enhance the attractiveness of creativity, and the high correlation between creative content and products, which not only improves the quality of creativity, but also enhances the interaction between the brand and the audience, and ultimately improves the market performance and competitiveness of the product, thereby ensuring the timeliness of creativity and helping to improve the click-through rate and conversion rate of creativity. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 An overall flow chart of a method for generating creative ideas with the help of AI is provided as an exemplary embodiment of the present application.

[0087] Figure 2 A flowchart of an execution of a method for generating creative ideas with the help of AI is provided as another exemplary embodiment of the present application.

[0088] Figure 3 A structural block diagram of a device for realizing creative generation with the help of AI is provided as another exemplary embodiment of the present application. DETAILED DESCRIPTION

[0089] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make modifications to the present embodiment without any creative contribution as needed, but such modifications are protected by the patent law as long as they are within the scope of the claims of the present application.

[0090] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0091] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0092] Existing AI tools for generating ideas often have the following problems:

[0093] 1. Lack of innovation and appeal. AI lags behind in capturing and utilizing hot content in the current market. Hot content is dynamic and changes rapidly, relying on people's immediate attention and interests. However, AI models are usually based on past data and existing rules, which makes them not responsive enough in responding to the latest market trends and hot spots. As a result, the generated creative content often lacks innovation and appeal, lacks flexibility and creativity, cannot be combined with hot content, and the creativity looks bland.

[0094] 2. Lack of connection with product information. AI-generated ideas often appear unnatural or stiff in combination with specific product information. Although AI can generate a variety of ideas, these ideas often cannot naturally lead to product information and lack a deep connection with the brand or product. In the process of generating ideas, if the core value and unique selling point of the product are ignored, the result may be that the idea is disconnected from the product and it is difficult to effectively convey the brand information. For example, a good idea should not only attract the audience's attention, but also cleverly integrate the product features so that the audience can naturally accept the product information while enjoying the idea.

[0095] 3. Lack of prediction of the effect of creativity. It is often impossible to effectively predict in advance whether the creativity generated by AI can achieve the expected effect in actual application. Traditional AI models lack a dynamic evaluation mechanism for the effect of creativity, and it is difficult to make real-time adjustments based on market feedback. For mass-produced creativity, AI lacks a selection mechanism and cannot screen out the most promising and effective creativity.

[0096] Therefore, although the existing AI has shown certain potential in creative generation, it is still far from the stage where it can be directly used for execution. Therefore, in order to enable AI to have more advanced semantic understanding and creative generation capabilities, timely capture and utilize the latest market trends, real-time insight into market hotspots to enhance the attractiveness of creativity, and closely integrate creativity with product information to deliver product value, as well as deeply understand the product's positioning, audience and market demand, and organically integrate these elements into creativity, rather than simply embedding product information rigidly into it, this application proposes a two-dimensional screening based on real-time insight into hot spots and popularity, and deeply associates product information based on a deep learning model based on semantic matching to generate creativity, so as to ensure the timeliness of creativity and help improve the click-through rate and conversion rate of creativity.

[0097] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0098] Reference Figure 1 , an embodiment of the present application provides a method for generating creative ideas with the help of AI, and the main steps of the method are described as follows.

[0099] S1: Obtain user-generated content from various social media platforms in real time and collect hot topics in the current market;

[0100] S2: Perform semantic analysis and sentiment analysis based on the hot topics to generate hot topic portraits;

[0101] S3: Calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait;

[0102] S4: performing dual-dimensional screening based on the hotspot portrait of the target audience portrait and the popularity of the hotspot to obtain a target hotspot portrait;

[0103] S5: Input the target hot spot portrait and the associated product information into a preset deep learning model based on semantic matching to output creativity.

[0104] In one embodiment, after inputting the target hot spot portrait and the associated product information into a preset deep learning model based on semantic matching and outputting the creative idea, the process further includes:

[0105] S6: Compare and score the ideas in multiple dimensions to determine the target ideas.

[0106] Reference Figure 2Specifically, through data collection and big data analysis technology, we use web crawler engines and API interfaces to obtain user-generated content (hot spots) from major social media platforms in real time, monitor and analyze various social media platforms in real time, and collect hot topics in the current market. Among them, hot topics can include breaking news, popular events, hot topics, etc.

[0107] Optionally, data cleaning is performed on the collected hot topics to remove noise and duplicate data to retain high-quality data related to the hot topics, which is beneficial for subsequent model learning.

[0108] Conduct semantic analysis and sentiment analysis on hot topics to extract the core content and sentiment tendency of the hot topics, and form a hot topic portrait by integrating the results of semantic analysis and sentiment analysis.

[0109] The matching degree between the hot profile and the target audience (TA) profile of the product is calculated, and the hot profile with the highest matching degree is selected as the hot profile of the target audience profile.

[0110] And based on the hot spot portrait of the target audience portrait, two-dimensional screening is performed in combination with the popularity of the hot spots, and the hot spots are adapted to obtain the target hot spot portrait.

[0111] Next, the target hot spot portrait and the associated product information (product selling point) are input into a preset deep learning model based on semantic matching to output the creativity. In this embodiment, the deep learning model based on semantic matching can be any one of the Transformer model, the BERT model or the ESIM model.

[0112] Finally, we use the big model to conduct multi-dimensional comparison and scoring of the creative content generated by AI to obtain a batch of high-quality and popular ideas.

[0113] In one embodiment, the steps of performing semantic analysis and sentiment analysis based on the hot topics to generate hot topic portraits include:

[0114] Using natural language processing (NLP) technology, the hot topics are segmented, POS-tagged, and named entity recognized to extract the core content of the hot topics;

[0115] Based on the core content of the hot topic, determine the theme and keywords, determine the theme and main information of the hot topic, and conduct topic content analysis;

[0116] Collect statistics on user group attributes corresponding to the hot topics, including gender, age, region, interests, etc. of the user groups discussing the hot topics, so as to identify and analyze the target audience, understand the characteristics and needs of the target audience, and customize creative content;

[0117] Count the dissemination channels and paths of the hot topics on different social media platforms to conduct dissemination channel analysis, understand the dissemination of hot topics on different social media platforms, and assist in optimizing content publishing and dissemination strategies;

[0118] Track the entire life cycle of the hot topic from its emergence to its disappearance, determine the length attributes of the entire life cycle and the characteristics of each stage, so as to predict the duration of the hot topic and the best time to intervene through life cycle analysis;

[0119] By analyzing the hot topics through time series, the peak and trough periods of the hot topics are obtained to analyze the hot trends, which is conducive to formulating accurate marketing plans based on the results;

[0120] Evaluate the user's emotional inclination towards the hot topic, which includes positive, negative and neutral, so as to adjust the brand communication strategy and enhance the emotional resonance with the audience by analyzing the changing trend of the user's emotional response;

[0121] Summarize the core content, theme, keywords, user group attributes, communication channels, communication paths, length attributes of the entire life cycle and characteristics of each stage, popularity peaks, popularity troughs and emotional tendencies of the hot topics, and form the hot topic portrait corresponding to the hot topic, so as to form a hot topic portrait through in-depth analysis of the semantics and emotions of the hot topics, and provide data support for subsequent marketing strategies.

[0122] The analysis results of hot spot portraits also have important application value in subsequent creative generation. For example, by understanding the characteristics and needs of the target audience, creative content can be customized to improve marketing effectiveness; by identifying the main communication channels of hot spots, content publishing platforms can be optimized and the scope of communication can be expanded; through emotional response analysis, communication strategies can be adjusted to enhance brand image and user satisfaction.

[0123] In one embodiment, the steps of calculating the matching degree between the hotspot portrait and the target audience portrait of the product, and selecting the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait include:

[0124] Obtain the age, gender, hobbies and consumption behavior of the target audience portrait to obtain product TA portrait data;

[0125] Extract age distribution characteristics, gender ratio characteristics, interest keyword characteristics and typical consumption behavior characteristics from each of the hot spot portraits and the target audience portraits to achieve key feature extraction;

[0126] Using similarity algorithms, such as cosine similarity and Pearson correlation coefficient, the similarity of age distribution characteristics, gender ratio characteristics, interest keyword characteristics, and typical consumption behavior characteristics between each hot profile and the target audience profile is calculated, and the similarity between the hot profile and the product TA profile in various characteristics is determined. For example, the similarity between the hot audience and the product audience in age distribution and gender ratio is calculated;

[0127] Based on the preset feature weights, the similarity of the age distribution feature, the similarity of the gender ratio feature, the similarity of the interest keyword feature, and the similarity of the typical consumption behavior feature are weighted and scored. According to the importance of each feature, a weight is manually assigned to each feature, and then the similarity of each feature is weighted and comprehensively scored to obtain the matching degree of each hotspot portrait;

[0128] Compare the matching degree of each hotspot portrait with the preset matching degree threshold, and determine the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait. For example, set the matching degree threshold to 80%, and the hotspots with a comprehensive score exceeding this threshold are those that overlap with the product audience. Then select the hotspot portrait with the largest difference from the matching degree threshold, that is, the one that highly overlaps with the product audience characteristics, to obtain the hotspot portrait of the target audience portrait, thereby ensuring that the generated creativity can resonate with the target audience. For example, if a hotspot mainly attracts young female audiences, and the target audience of the product is also young women, then this hotspot will be given priority.

[0129] In one embodiment, the step of performing dual-dimensional screening based on the hot spot portrait of the target audience portrait and the popularity of the hot spot to obtain the target hot spot portrait includes:

[0130] Counting the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait;

[0131] Calculate the hot spot attention of each hot spot portrait based on the search volume index and the number of social media mentions;

[0132] Counting the current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait since their appearance, and determining the hot spot duration of each hot spot portrait;

[0133] Obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence;

[0134] The hot spot portrait with the largest hot spot heat is selected from the hot spot portraits of the target audience portrait as the target hot spot portrait.

[0135] The popularity of a hot spot takes into account the current level of attention and duration of the hot spot, and evaluates the influence of the hot spot through quantitative indicators, so as to screen out hot spots that are highly matched with the product audience and have high popularity in the current market, as the input basis for creative generation.

[0136] Specifically, when counting the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait, the following formula is used:

[0137] S = current social media search volume / maximum search volume × 100;

[0138] M = current number of social media mentions / maximum number of mentions × 100;

[0139] Among them, S is the search volume index, and M is the number of social media mentions;

[0140] Based on the search volume index and the number of social media mentions, the following formula is used to calculate the hot spot attention of each hot spot portrait:

[0141] C = α × S + β × M;

[0142] Among them, C is the hot spot attention, α is the weight coefficient of the search volume index, and β is the weight coefficient of the number of social media mentions; α and β can be adjusted according to specific needs;

[0143] The current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait is counted, and the hot spot duration of each hot spot portrait is determined by using the following formula:

[0144] T = current duration / maximum duration × 100;

[0145] Where T is the hotspot duration, and the unit of the current duration is day;

[0146] When obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence, the following formula is used:

[0147] I = γ × C + δ × T;

[0148] Among them, I is the heat of the hot spot, which is used to evaluate the influence of the heat, γ is the weight coefficient of the hot spot attention, and δ is the weight coefficient of the hot spot persistence. γ and δ can be adjusted according to specific needs.

[0149] On the basis of matching the hot image and the product audience image, we then conduct dual-dimensional screening in combination with the popularity of the hot spot, quantify the influence of the hot spot, and select the hot image with the highest popularity from the hot image of the target audience image as the target hot image. This not only ensures that the creativity can resonate with the target audience, but also can use the high-heat hot spots in the current market to generate creative ideas that are both innovative and attractive and closely integrated with product information, thereby improving marketing effectiveness. This not only improves the market responsiveness of creativity, but also enhances the interactive effect between the brand and the audience, which is conducive to enterprises occupying a favorable position in the fierce market competition.

[0150] In one embodiment, the target hot spot portrait and the associated product information are input into a preset deep learning model based on semantic matching, and the step of outputting creativity includes:

[0151] Based on the deep learning model of semantic matching, the target hot spot portrait is associated with the selling point of the product and the market demand of the product to generate creative ideas in the form of advertising copy, visual design and video script.

[0152] In order to generate highly compatible and attractive creative ideas, and organically integrate creative content with product selling points, we use big data analysis technology and NLP technology to deeply analyze the core selling points, target audience portraits, brand tone and market positioning of products, and conduct comprehensive evaluations based on product sales data and user behavior data to extract key features and unique values ​​of products, ensuring that the generated creative ideas are in line with current market trends; and use deep learning models based on semantic matching to enable AI to understand the potential connection between hot content and product selling points, and generate highly relevant creative content that can naturally convey product value based on the degree of matching, deeply associate market hot spots with specific selling points and market demand of products, such as advertising copy, visual design, video scripts and other multimedia creative forms, which is conducive to the seamless integration of creative content with product information and sales policies. For example, "Graduation proposal is very romantic, but how can you get married right after graduation without a car? xx car, down payment only xxxx yuan, monthly payment as low as xxx, drive home quickly", so that it can not only combine the current hot spots of "graduation proposal", but also smoothly bring out product information. Creative content can effectively convey the core selling points and brand value of a product, thereby enhancing the brand's market competitiveness and the audience's willingness to purchase, ensuring the high relevance of creative content to the product, not only improving the quality of creative content, but also enhancing the interaction between the brand and the audience, ultimately improving the product's market performance and competitiveness.

[0153] In one embodiment, the steps of comparing and scoring the ideas in multiple dimensions and determining the target idea include:

[0154] Use the preset prediction model to compare and score each creative idea in multiple dimensions;

[0155] Wherein, the prediction model includes an input layer, a feature extraction layer, a multi-dimensional evaluation layer and an output layer;

[0156] The input layer obtains and outputs historical creative effect data, market feedback data, audience response data, and current market trends and competitive environment data; the historical creative effect data includes key performance indicators (KPIs) such as click-through rate, conversion rate, and user feedback of previous creatives; the market feedback data comes from the actual market response, such as social media comments, customer satisfaction surveys, etc.; the audience response data includes the behavior data and preference data of the target audience;

[0157] Based on the preset evaluation features, the feature extraction layer extracts creative features associated with the evaluation features from the output data of the input layer, such as features related to the innovativeness of the creative, audience resonance points, market adaptability, etc. The current market trend and competition environment data are the latest market trend and competition environment data, which are used to capture the changes and competition dynamics of the current market;

[0158] The multi-dimensional evaluation layer scores the creative features output by the feature extraction layer to achieve multi-dimensional evaluation, including the innovation, attractiveness, relevance to product information, market adaptability, etc. of the creative idea, and calculates a comprehensive score. Then, based on the comprehensive score, an evaluation report containing pros and cons analysis and recommendation of the target creative idea is matched; the pros and cons analysis content and recommendation reasons are manually processed and matched based on historical creative data and historical market feedback data;

[0159] The output layer outputs the comprehensive score and the evaluation report of the target creativity by the multi-dimensional evaluation layer.

[0160] In one embodiment, the pros and cons analysis of the evaluation report includes describing the performance of the target creativity in each evaluation dimension;

[0161] The preferential recommendation of the evaluation report includes, based on the comprehensive score, selecting the idea corresponding to the largest comprehensive score as the target idea.

[0162] In order to optimize the generated batch of creative ideas and predict the creative effects, a prediction model is used to compare and score the large number of generated ideas in multiple dimensions to quantify the overall pros and cons of the ideas. The prediction model is a multi-layer neural network structure, and the performance of each idea in each evaluation dimension is described in detail through pros and cons analysis to help decision makers understand the potential and market adaptability of the ideas; then, the optimal recommendation template is matched according to market demand, and the more promising ideas are recommended for actual delivery, ensuring that the final selected ideas have higher market potential and effects.

[0163] In one embodiment, the step of training the prediction model includes:

[0164] Inputting large-scale data sets in vertical fields into the prediction model for training;

[0165] Optimizing the model parameters of the prediction model through cross-validation and parameter adjustment;

[0166] When the prediction accuracy of the prediction model reaches a preset accuracy threshold, the current prediction model is output.

[0167] Among them, large-scale data sets refer to data sets with a parameter scale of more than 1 billion. Using large-scale creative data and market feedback data, large-scale data sets in vertical fields are input into the prediction model for training. The model is dynamically iterated through cross-validation and parameter adjustment to achieve continuous updating and optimization of the model, so that it can identify and learn more successful creative patterns. It can not only efficiently generate and evaluate a large number of creative ideas, but also ensure the adaptability of creativity in different product and market environments. The model can better understand and evaluate the generated creativity, so that the final selected creativity has higher market potential and actual effect, thereby improving the quality and relevance of creative content, greatly improving the efficiency of creative generation and delivery, achieving the goal of brand promotion, and providing strong support for enterprises in the fierce market competition.

[0168] In summary, a method for achieving creative generation with the help of AI obtains user-generated content from various social media platforms in real time, collects hot topics in the current market, and obtains hot content in the market in real time, which is conducive to timely capturing and utilizing the latest market trends and ensuring the timeliness of generating creative ideas; semantic analysis and sentiment analysis are performed based on hot topics to generate hot profiles to provide a data basis for subsequent creative generation, which can customize creative content, adjust communication strategies, improve marketing effects, and enhance brand image and user satisfaction; the matching degree between the hot profile and the target audience portrait of the product is calculated, and the hot profile with the highest matching degree is selected as the hot profile of the target audience portrait to obtain hot spots that highly overlap with the product audience characteristics, which is conducive to the generated creativity to resonate with the target audience; according to the hot profile of the target audience portrait, combined with the heat of the hot spot, Perform two-dimensional screening to obtain the target hot spot portrait, so that the hot spot can match the product audience and have a high popularity in the current market, which is conducive to improving the creative click-through rate; finally, the target hot spot portrait and the related product information are input into the preset semantic matching-based deep learning model, and the creativity is output, so that the creative content can be associated with the product information, and the core selling point and brand value of the product can be effectively conveyed, thereby enhancing the brand's market competitiveness and the audience's willingness to buy; the generated creativity can be combined with the current hot content and keep up with market trends. At the same time, the hot spot can attract the target audience, enhance the attractiveness of the creativity, and the high correlation between the creative content and the product, which not only improves the quality of the creativity, but also enhances the interaction between the brand and the audience, and ultimately improves the market performance and competitiveness of the product, thereby ensuring the timeliness of the creativity and helping to improve the click-through rate and conversion rate of the creativity.

[0169] A method of generating ideas with the help of AI can provide real-time insights into and use of market hotspots, capture the latest market dynamics, and match them with product audiences, thereby generating ideas that both keep up with market trends and resonate with target audiences. This is conducive to generating highly attractive creative content, which not only improves the innovation and attractiveness of creative content, but also enhances the interaction between brands and audiences, thereby improving the market performance of products.

[0170] A method of using AI to achieve creative generation integrates product information into the AI ​​generation model, so that the model can deeply understand the core value and unique selling points of the product, and organically integrate these elements into the creativity, ensuring that the generated creativity can naturally convey the brand value. The generated creativity can not only attract the audience's attention, but also effectively convey product information, enhance brand awareness and market influence. By deeply associating product selling points, creativity can be more naturally integrated into product characteristics, allowing the audience to naturally accept product information while enjoying the creativity.

[0171] A method of generating ideas with the help of AI, which compares and scores ideas in multiple dimensions, selects the ideas with the greatest potential and effectiveness for key deployment, and dynamically provides feedback on whether the ideas have achieved the expected results in actual applications, significantly improving the market performance and conversion rate of the ideas. It can efficiently generate and evaluate a large number of ideas, predict the effects of ideas, and select high-quality ideas. By introducing the effect pre-test function, it predicts the execution quality of ideas, ensuring that the ideas finally selected have the highest market potential and actual effects, providing strong support for enterprises in the fierce market competition.

[0172] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0173] Reference Figure 3 The embodiment of the present application also provides a device for realizing creative generation by means of AI, and the device for realizing creative generation by means of AI corresponds one to one with a method for realizing creative generation by means of AI in the above embodiment. The device for realizing creative generation by means of AI includes:

[0174] The data module is used to obtain user-generated content from various social media platforms in real time and collect hot topics in the current market;

[0175] A hot topic portrait construction module is used to perform semantic analysis and sentiment analysis based on the hot topic to generate a hot topic portrait;

[0176] A target audience hotspot portrait module is used to calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait;

[0177] A target hotspot screening module is used to perform dual-dimensional screening based on the hotspot portrait of the target audience portrait and the heat of the hotspot to obtain a target hotspot portrait;

[0178] The creative generation module is used to input the target hot spot portrait and the associated product information into a preset deep learning model based on semantic matching to output creativity.

[0179] A device for generating creative ideas by means of AI also includes:

[0180] The creative prediction module is used to compare and score the creative ideas in multiple dimensions to determine the target creative ideas.

[0181] For the specific definition of a device for realizing creative generation with the help of AI, please refer to the definition of a method for realizing creative generation with the help of AI above, which will not be repeated here.

[0182] Each module in the above-mentioned device for realizing creative generation by means of AI can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0183] In one embodiment, a computer device is provided, which may be a server. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, any one of the above-mentioned methods for generating creativity with the help of AI is implemented.

[0184] In one embodiment, a computer-readable storage medium is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, any one of the above-mentioned methods for generating creativity with the help of AI is implemented.

[0185] In one embodiment, a computer program product is provided, which includes a computer program, and when the computer program is executed by a processor, any one of the above methods for generating creativity with the help of AI is implemented.

[0186] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. When the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0187] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

Claims

1. A method for generating creative ideas with the help of AI, characterized in that: The following steps are included: Obtain user-generated content from various social media platforms in real time and collect hot topics in the current market; Perform semantic analysis and sentiment analysis based on the hot topics to generate hot topic portraits; Calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait; According to the hot spot portrait of the target audience portrait, a dual-dimensional screening is performed in combination with the popularity of the hot spot to obtain a target hot spot portrait; The target hot spot portrait and the associated product information are input into a preset deep learning model based on semantic matching to output creativity.

2. The method for generating creative ideas with the help of AI according to claim 1, characterized in that: Based on the hot topic, semantic analysis and sentiment analysis are performed to generate a hot topic portrait, including: Using natural language processing technology, perform word segmentation, part-of-speech tagging and named entity recognition on the hot topics to extract the core content of the hot topics; Determine the theme and keywords based on the core content of the hot topic; Counting user group attributes corresponding to the hot topics to identify target audiences and customize creative content; Collect statistics on the dissemination channels and paths of the hot topics on different social media platforms to assist in optimizing content publishing platforms and dissemination strategies; Track the entire life cycle of the hot topic from its emergence to its disappearance, determine the length attributes of the entire life cycle and the characteristics of each stage, so as to predict the duration of the hot topic and the best time to intervene; Analyze the hot topics through time series to obtain the peak and trough periods of the hot topics for use in formulating marketing plans; Evaluate the user's sentiment towards the hot topic, where the sentiment includes positive, negative and neutral, so as to adjust the brand communication strategy; The core content, theme, keywords, user group attributes, communication channels, communication paths, length attributes of the entire life cycle and characteristics of each stage, popularity peaks, popularity troughs and emotional tendencies of the hot topics are summarized to form the hot topic portrait corresponding to the hot topic.

3. The method for generating creative ideas with the help of AI according to claim 1, characterized in that: The steps of calculating the matching degree between the hotspot portrait and the target audience portrait of the product, and selecting the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait include: Obtain the age, gender, interests, hobbies and consumption behavior of the target audience portrait; Extracting age distribution characteristics, gender ratio characteristics, interest keyword characteristics and typical consumption behavior characteristics from each of the hot spot portraits and the target audience portraits; Using a similarity algorithm, calculate the age distribution feature similarity, gender ratio feature similarity, interest keyword feature similarity, and typical consumption behavior feature similarity of each hotspot portrait and the target audience portrait respectively; Based on the preset feature weights, weighted scores are performed on the age distribution feature similarity, the gender ratio feature similarity, the interest keyword feature similarity, and the typical consumer behavior feature similarity to obtain the matching degree of each hotspot portrait; The matching degree of each hotspot portrait is compared with a preset matching degree threshold, and the hotspot portrait with the highest matching degree is determined as the hotspot portrait of the target audience portrait.

4. The method for generating creative ideas with the help of AI according to claim 1, characterized in that: According to the hot spot portrait of the target audience portrait, the steps of performing dual-dimensional screening in combination with the popularity of the hot spot to obtain the target hot spot portrait include: Counting the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait; Calculate the hot spot attention of each hot spot portrait based on the search volume index and the number of social media mentions; Counting the current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait since their appearance, and determining the hot spot duration of each hot spot portrait; Obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence; The hot spot portrait with the largest hot spot heat is selected from the hot spot portraits of the target audience portrait as the target hot spot portrait.

5. The device for realizing creative generation by means of AI according to claim 4, characterized in that: The following steps are also included: When calculating the search volume index and social media mention times of the hot spots corresponding to each hot spot portrait of the target audience portrait, the following formula is used: S = current social media search volume / maximum search volume × 100; M = current number of social media mentions / maximum number of mentions × 100; Among them, S is the search volume index, and M is the number of social media mentions; Based on the search volume index and the number of social media mentions, the following formula is used to calculate the hot spot attention of each hot spot portrait: C = α × S + β × M; Among them, C is the hot spot attention, α is the weight coefficient of the search volume index, and β is the weight coefficient of the number of mentions on social media; The current duration of the hot spots corresponding to each hot spot portrait of the target audience portrait is counted, and the hot spot duration of each hot spot portrait is determined by using the following formula: T = current duration / maximum duration × 100; Where T is the hotspot duration, and the unit of the current duration is day; When obtaining the hotspot heat of each hotspot portrait according to the hotspot attention and the hotspot persistence, the following formula is used: I = γ × C + δ × T; Among them, I is the heat of the hot spot, γ is the weight coefficient of the hot spot attention, and δ is the weight coefficient of the hot spot persistence.

6. The method for generating creative ideas with the help of AI according to claim 1, characterized in that: The target hot spot portrait and the associated product information are input into a preset deep learning model based on semantic matching, and the steps of outputting creativity include: Based on the deep learning model of semantic matching, the target hot spot portrait is associated with the selling point of the product and the market demand of the product to generate creative ideas in the form of advertising copy, visual design and video script.

7. The method for generating creative ideas with the help of AI according to any one of claims 1 to 6, characterized in that: After inputting the target hot spot portrait and the associated product information into a preset semantic matching-based deep learning model and outputting the creative idea, the method further includes: Compare and score the ideas in multiple dimensions to determine the target ideas.

8. The method for generating creative ideas with the help of AI according to claim 7, characterized in that: The steps of comparing and scoring the ideas in multiple dimensions and determining the target ideas include: Use the preset prediction model to compare and score each creative idea in multiple dimensions; Wherein, the prediction model includes an input layer, a feature extraction layer, a multi-dimensional evaluation layer and an output layer; The input layer obtains and outputs historical creative effect data, market feedback data, audience response data, and current market trends and competitive environment data; Based on the preset evaluation feature, the feature extraction layer extracts the creative feature associated with the evaluation feature from the output data of the input layer; The multi-dimensional evaluation layer scores the creative features output by the feature extraction layer and calculates a comprehensive score, and then matches the evaluation report containing pros and cons analysis and optimal recommendation of the target creative according to the comprehensive score; The output layer outputs the comprehensive score and the evaluation report of the target creativity by the multi-dimensional evaluation layer.

9. The method for generating creative ideas with the help of AI according to claim 8, characterized in that: The training steps of the prediction model include: Inputting large-scale data sets in vertical fields into the prediction model for training; Optimizing the model parameters of the prediction model through cross-validation and parameter adjustment; When the prediction accuracy of the prediction model reaches a preset accuracy threshold, the current prediction model is output.

10. The method for generating creative ideas with the help of AI according to claim 8, characterized in that: The pros and cons analysis of the evaluation report includes describing the performance of the target creativity in each evaluation dimension; The preferential recommendation of the evaluation report includes, based on the comprehensive score, selecting the idea corresponding to the largest comprehensive score as the target idea.

11. A device for realizing creative generation by means of AI, characterized in that: include, The data module is used to obtain user-generated content from various social media platforms in real time and collect hot topics in the current market; A hot topic portrait construction module is used to perform semantic analysis and sentiment analysis based on the hot topic to generate a hot topic portrait; A target audience hotspot portrait module is used to calculate the matching degree between the hotspot portrait and the target audience portrait of the product, and select the hotspot portrait with the highest matching degree as the hotspot portrait of the target audience portrait; A target hotspot screening module is used to perform dual-dimensional screening based on the hotspot portrait of the target audience portrait and the heat of the hotspot to obtain a target hotspot portrait; The creative generation module is used to input the target hot spot portrait and the associated product information into a preset deep learning model based on semantic matching to output creativity.

12. The device for realizing creative generation by means of AI according to claim 11, characterized in that: Also includes, The creative prediction module is used to compare and score the creative ideas in multiple dimensions to determine the target creative ideas.

13. A computer device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method according to any one of claims 1 to 10.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

15. A computer program product, characterized in that The invention comprises a computer program, which implements the steps of the method according to any one of claims 1 to 10 when being executed by a processor.

Images