WO2023217194A1

WO2023217194A1 - Dynamic video cover generation method

Info

Publication number: WO2023217194A1
Application number: PCT/CN2023/093327
Authority: WO
Inventors: 陈羽飞; 刘奎龙; 詹鹏鑫; 许佳琪
Original assignee: 阿里巴巴（中国）有限公司
Priority date: 2022-05-13
Filing date: 2023-05-10
Publication date: 2023-11-16
Also published as: CN114996553A

Abstract

The present application provides a dynamic video cover generation method, comprising: acquiring static materials related to a video subject; classifying the acquired static materials related to the video subject into corresponding static material categories according to content properties; permuting and combining the static materials according to a predetermined policy and the acquired static material categories, the predetermined policy at least comprising using different permutation and combination sequences for different target customer objects; and forming the permuted and combined static materials into a whole, and generating a dynamic video cover. In the present application, the dynamic video cover is generated by means of the static material according to the predetermined policy, such that the manufacturing cost of a high-quality dynamic video cover can be effectively reduced, personalized cover display is achieved for different consumer requirements, and then the click rate and the page view of a user are improved.

Description

A dynamic video cover generation method

This application claims priority to the Chinese patent application submitted to the China Patent Office on May 13, 2022, with application number 202210519403.1 and the application title "A dynamic video cover generation method", the entire content of which is incorporated into this application by reference. .

Technical field

The present application relates to the field of computers, and in particular to a dynamic video cover generation method, device, electronic equipment and storage medium.

Background technique

With the rapid development of Internet technology, e-commerce application platforms are playing an increasingly important role in people's daily lives. The display of goods in e-commerce application platforms also presents increasingly rich forms of expression. The form of product display has a vital impact on consumers' click-through rate, purchase rate, conversion rate, user stickiness, etc. Therefore, it is increasingly meaningful to explore the product display method of e-commerce application platforms.

Among existing technologies, commonly used e-commerce platforms mainly display products through pictures and videos. Among them, displaying products in the form of pictures on the product search interface and recommendation interface cannot allow consumers to intuitively feel the quality of the product. However, using automatically playing audio and video dynamic video covers makes consumers visually more intuitive and effective. It can improve consumers' awareness of products in the shortest possible time, promote consumers to make purchasing decisions, and significantly increase the click-through rate and purchase conversion rate of products. Therefore, more and more stores are beginning to use dynamic short video covers instead of original picture covers to display and introduce products, and realize dynamic three-dimensional display of products through the automatic playback of dynamic short videos.

In the practical application of dynamic short videos, most of the dynamic video covers on e-commerce platforms are shot and uploaded by the merchants themselves. Therefore, the coverage rate of high-quality dynamic short videos on e-commerce platforms is low, and due to the high cost of shooting high-quality short videos and the mismatch between the key points of the product content displayed by the merchants and the key points of user needs, how to use it at a lower cost? Generating high-quality dynamic video covers and improving the personalized display of dynamic video covers in response to different consumer needs has become an urgent problem for e-commerce application platforms.

Contents of the invention

The present invention provides a dynamic video cover generation method, which solves the existing technical problems of high dynamic video cover generation costs and inability to perform dynamic video cover display in a targeted manner in the face of different consumer demands. The present invention also provides a device, electronic equipment and storage medium for dynamic video cover generation. The technical solutions are as follows:

The embodiment of this application provides a dynamic video cover generation method, including:

Obtain static materials related to the main body of the video;

Classify the obtained static materials related to the main body of the video into the corresponding static material categories according to the nature of the content;

According to a predetermined strategy, the static materials are arranged and combined according to the obtained static material categories; the predetermined strategy at least includes using different arrangements and combinations for different target customer objects;

The arranged and combined static materials are formed into a whole to generate the dynamic video cover.

Optionally, according to the different predetermined strategies, the obtained static materials related to the video body can generate multiple versions of dynamic video covers; for the same predetermined strategy, due to the different selection of the static materials, also A variety of different versions of dynamic video covers can be generated.

Optionally, forming the arranged and combined static materials into a whole to generate the dynamic video cover includes:

Before generating the dynamic video cover, the expression method of the static material related to the video body is deformed.

Optionally, the deformation processing of the expression method of the static material related to the video subject includes: beautifying the obtained static material by selecting and using a design style template.

Optionally, the number of dynamic video covers is at least the first number, and the method further includes:

Obtain the first number of dynamic video covers;

For the first number of dynamic video covers, select a number of sample customers from the target customer group and conduct a first prediction test;

Based on the test feedback data and predetermined evaluation criteria, a target dynamic video cover that matches the target group is obtained.

Optionally, after determining the target dynamic video cover, further perform the following steps:

Obtain the target dynamic video cover;

Adjust the target dynamic video cover to obtain multiple fine-tuned target dynamic video covers. noodle;

For the plurality of fine-tuned target dynamic video covers, extract a number of sample customers from the target customer group and conduct a second prediction test;

Based on the test feedback data, a fine-tuned dynamic video cover that matches the target group is obtained.

Optionally, the target customer group is a customer group with target attributes obtained after classifying the customer groups.

Optionally, classifying customer groups includes: dividing customers with the same demand characteristic labels into the same type of customer groups.

Optionally, the target attribute is an attribute marked according to the customer's demand feature tag.

Optionally, the number of dynamic video covers is a second number greater than the first number, and the method further includes: targeting customer groups with specific target attributes, before conducting the first estimated test, based on the specific The target attribute selects the first number of dynamic video covers for testing from a second number of dynamic video covers more than the first number in a predetermined method.

The dynamic video cover generation method provided by this application obtains static materials related to the main body of the video, and classifies the obtained static materials related to the main body of the video into corresponding static material categories according to the nature of the content; according to the predetermined strategy, according to the Obtained static material categories, arrange and combine the static materials; the predetermined strategy at least includes using different arrangements and combinations for different target customer objects; form the arranged and combined static materials into a whole, and generate the dynamic video cover. This method of generating dynamic video covers uses existing static materials as the source of video production. The same static material group can generate multiple versions of dynamic video covers for different target customer groups according to different predetermined strategies. This technical solution can greatly reduce the cost of producing dynamic video covers, and by generating dynamic video covers according to predetermined strategies, it can meet the needs of different target customer groups and generate targeted dynamic videos that adapt to them. Video cover has the effect of increasing users’ click-through rate and views.

In addition, this application also provides a device, an electronic device, and a computer-readable storage medium for a dynamic video cover generation method, which also have the above beneficial effects.

It should be understood that what is described in this section is not intended to identify key or important features of the application, nor is it intended to limit the scope of the application. Features of the present application will become readily understood through the following detailed description of typical examples.

Description of the drawings

In order to explain the technical solutions of the embodiments of the present application more clearly, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. , for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative labor.

Figure 1 is a schematic diagram of an application scenario for a dynamic video cover generation method provided by an embodiment of the present application;

Figure 2 is a schematic flow chart of a dynamic video cover generation method provided by an embodiment of the present application;

Figure 3 is a schematic flowchart of a method for generating a target dynamic video cover provided by an embodiment of the present application;

Figure 4 is a schematic flowchart of a method for generating a fine-tuned dynamic video cover that matches the target group provided by an embodiment of the present application;

Figure 5 is a schematic structural diagram of a device for generating a dynamic video cover provided by an embodiment of the present application;

Figure 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present application;

Figure 7 is a schematic structural diagram of a storage medium provided by an embodiment of the present application;

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description serve to explain the principles of the disclosure.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, many specific details are set forth in the following description to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described here. Those skilled in the art can make similar extensions without violating the connotation of the present invention. Therefore, the present invention is not limited by the specific implementation disclosed below.

It should be noted that the terms “first”, “second” and “third” in various parts of the embodiments of this application and the drawings are used to distinguish similar objects and do not indicate any specific differences between them. Sequence or sequence. Such data are interchangeable under appropriate circumstances so that the embodiments of the application described herein can be practiced in other sequences than illustrated or described herein.

The terms "including" and "having" and any variations thereof in this application are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units and need not be limited to those explicitly listed. may include other steps or elements not expressly listed or inherent to the process, method, product or apparatus.

In this application, unless expressly stated otherwise, the term "or" includes all possible combinations except where impracticable. For example, if it is expressed that a database may include A or B, then unless otherwise specified or impracticable, it may include database A, or B, or A and B. As another example, if it is expressed that a certain database may include A, B or C, then unless otherwise specified or infeasible, the database may include database A, or B, or C, or A and B, or A and C, or B and C, or A and B and C.

First, some technical terms involved in this application will be explained:

AB testing is to create two (A/B) or multiple (A/B/n) versions of a Web or App interface or process, and test them with the same (similar) visitor groups (targets) at the same time. People) randomly access these versions, collect user experience data and business data from each group, and finally analyze and evaluate the best version, which is officially adopted. Its role is to eliminate disputes between different opinions in customer experience (UX) design and determine the best solution based on actual results; through comparative experiments, find the real cause of the problem and improve product design and operation levels; to establish a data-driven, continuously optimized Closed-loop process; through A/B testing, the risk of launching new products or new features is reduced and product innovation is guaranteed.

CTR (Click-Through-Rate) estimation is one of the core algorithms in calculating advertising. It is to predict the clicks of each advertisement and predict whether the user will click or not click. Its estimation is related to many factors, such as historical click rate, ad position, time, users, etc. The CTR prediction model is a model that comprehensively considers various factors and characteristics and is trained on a large amount of historical data. Therefore, the CTR prediction model can be widely used in personalized recommendations, information retrieval, online advertising and other fields to learn and predict user feedback. Among them, user feedback mainly includes clicks, collections, purchases, etc.

Recommendation algorithm, in the era of massive information data, recommendation algorithm can filter and filter massive information, and display the information that users are most concerned about and interested in in front of users. It not only increases work efficiency, but also saves users time in filtering information. By connecting users and information, on the one hand, it helps users discover information that is valuable to them, and on the other hand, it allows information to be displayed among people who are interested in them, thus achieving a win-win situation for information providers and users. Recommendation algorithms can be roughly divided into three categories: content-based recommendation algorithms, collaborative filtering recommendation algorithms and knowledge-based recommendation algorithms.

Below, an example is provided to illustrate the application scenarios of the solution provided by this application.

Figure 1 is a schematic diagram of an application scenario for a dynamic video cover generation method provided by an embodiment of the present application. As shown in Figure 1, this application scenario includes client 101 and server 102, so The number of clients 101 is not limited. The client 101 obtains static materials related to the main body of the video and sends the user's static materials to the server 102. The server 102 generates a dynamic video cover based on the static materials according to a predetermined strategy and sends the The dynamic video cover is sent to the client 101, and the client 101 displays the dynamic video cover to consumers for them to browse and purchase. For the specific implementation process, please refer to the solutions of the following embodiments.

It should be noted that Figure 1 is only a schematic diagram of the application scenario of a dynamic video cover generation method provided by the embodiment of the present application. The embodiment of the present application does not limit the devices included in Figure 1, nor does it limit the communication between the devices in Figure 1. Positional relationships are limited. For example, to meet the application scenario shown in Figure 1, a data storage device may also be included. The data storage device may be an external memory relative to the client 101 or the server 102, or may be an internal memory integrated in the client 101 or the server 102. The client 101 can be a smart phone, a computer, a TV, a wearable device, a multimedia player, an e-reader and other devices, on which various types of shopping application software are installed; the server 102 can be a server or several A centralized server composed of servers can also be a cloud computing service center. The client 101 and the server 102 can be connected through wireless network communication, wired communication, or optical fiber cable communication.

It can be understood that according to actual needs, the number of devices of the client 101 and the server 102 in Figure 1 may be changed.

Next, the dynamic video cover generation method is introduced through specific embodiments.

The first embodiment of the present application provides a dynamic video cover generation method.

FIG. 2 is a flow chart of a dynamic video cover generation method provided by an embodiment of the present application, which can be used in the implementation environment shown in FIG. 1 . As shown in Figure 2, the method in the embodiment of this application includes the following steps:

S201: Obtain static material related to the main body of the video.

This step collects relevant basic information for the implementation of this embodiment, that is, static materials related to the video theme, for generating dynamic video covers.

Among them, the video subject refers to the target object that will be displayed in the form of video, including people, objects, events, etc.; when video is used for product promotion, it mainly refers to various commodities. In this embodiment, the scenario is limited to the above product promotion occasions. The above products are used for display and sale on the corresponding e-commerce application platform. Static materials are materials used to display product features or information.

For example, the subject of the video can be various objects such as mobile phones, books, vehicles, bread, apples, etc. Regarding the main products for display and sale on e-commerce platforms, where e-commerce platforms are various shopping application platforms common in daily life, this application does not limit the specific form. Taking a mobile phone as an example, the above static material can be a text introduction to the beauty function of the product mobile phone, a picture of the product mobile phone's appearance, or a sales data report for the three consecutive quarters of the product mobile phone, etc. Materials related to product mobile phones.

The above-mentioned static materials obtained are derived from the existing materials of the product, including but not limited to product information, product pictures, product videos, product audio and other forms. The product information can be text and graphics in the product introduction, or materials specially selected from a special material management library for users to use, or text content extracted from pictures through text recognition; pictures of the goods It can be a physical picture actually taken, a virtual image constructed using application software, or a hand-drawn portrait; the video of the product can also come from short video materials related to online applications, or it can be actually recorded by the user. short video material; the audio of the product can be actual recorded audio, or it can be material in the existing audio material library.

For example, the materials obtained in this step are: the text message of the beauty function of the product mobile phone "xxx model mobile phone beauty camera can help you return to 18 years old", the appearance picture of the product mobile phone, the product mobile phone The real-life vlog short video (videoweblog, video web log) as well as the seller’s audio commentary and static materials about the product mobile phone, and the sales data report of the product mobile phone for three consecutive quarters.

S202: Classify the obtained static materials related to the video subject into corresponding static material categories according to the nature of the content;

This step is used to classify the static materials collected in the previous step.

In this step, in order to enhance the user performance of the above dynamic video cover, after obtaining multiple related static materials of the video subject, the expression method of the above static materials needs to be deformed to obtain the deformed static materials, and then according to the predetermined The strategy uses the deformed static material to produce dynamic video covers.

Specifically, in this step, the expression method of the static material related to the video theme is mainly transformed, and the static material obtained above is beautified by selecting and using a design style template. The above design styles include but are not limited to deformation of product information, fonts, font sizes, colors, animations, etc. of product text information; cutout, processing, animation, special effects enhancement, etc. of product images; product video Transition, playback material, playback area adjustment, etc.; commercial audio clipping, noise reduction, reverberation, speed change, pitch change processing, etc. The above static materials related to various senses such as vision and hearing will be beautified and adjusted through design templates to enhance consumers' intuitive experience. Moreover, using existing static materials as production materials for dynamic video covers can reduce the cost of video production to a certain extent.

For example, the above-obtained text information about the beauty function of the product mobile phone "xxx model mobile phone beauty camera can help you return to 18 years old", the appearance picture of the product mobile phone, and the actual vlog of the product mobile phone The short video (video weblog, video web log) and the seller's audio commentary static materials about the product mobile phone, and the three consecutive quarter sales data report of the product mobile phone's static materials are deformed. Transform the expression method of material 1: the text information of the beauty function of the commercial mobile phone "xxx model mobile phone beauty camera" is transformed into a font using Song Dynasty, size 20, the font color is a gradient color, and the font is jumped to the right in the screen, with 0.2 Second/frame jump speed text information of beauty function after transformation. Transform the expression method of material 2: the appearance of the product mobile phone through dehazing, contrast enhancement and lossless enlargement to transform the image; transform the expression method of material 3: the real-shot vlog short video (videoweblog, video network log) of the product mobile phone Use editing, splicing, adding filters, and adding background music to achieve transformation; transform the expression method of material 4: the seller's audio commentary static material about the product's mobile phone is processed through amplification, filters, and equalizers to achieve transformation; transform the expression method of material 5 Transformation: The sales data report of the product mobile phone for three consecutive quarters is transformed into a sales data report in the form of a three-dimensional column chart. The deformation processing of the above expression method increases customers' overall sensory perception of the mobile phone, and uses existing materials for deformation. Its production cost is low and it lowers the threshold for high-quality short video production.

In embodiments of the present disclosure, the categories of static materials that may exist are all possible categories classified according to the content of the static materials. This category is used to express the selling point characteristics or selling point characteristics of the product. The above possible categories include but are not limited to: product price, product function, sales volume, cost performance, quality, brand, after-sales service, concept, process/origin advantages, configuration, core consumer groups, packaging advantages, added value, creativity, copyright, Delivery services and many other categories are not specifically limited in the embodiments of this application.

Illustratively, according to the examples in the above steps, the categories of commercial mobile phones can be: mobile phone appearance, mobile phone brand, mobile phone functions, mobile phone sales, mobile phone cost performance, mobile phone after-sales service, mobile phone price, mobile phone configuration, mobile phone copyright, etc. Many categories.

Illustratively, according to the example in step S201 above, this step obtains in step S201 The text information of the beauty function of the product mobile phone "xxx model mobile phone beauty camera", the appearance picture of the product mobile phone, the actual vlog short video (videoweblog, video network log) of the product mobile phone, and the seller's information about the product The audio commentary static materials of a product mobile phone and the sales data report of the product mobile phone for three consecutive quarters', after being transformed and beautified, are classified into corresponding categories according to their content information. For example, according to the content information, the above-obtained text information about the beauty function of the product mobile phone "xxx model mobile phone beauty camera" material is classified into the product function category, the appearance picture material of the product mobile phone is classified into the design category, and the product mobile phone's appearance picture material is classified into the design category. The real-shot vlog short video (videoweblog, video network log) material is classified into the corresponding cost-effective category, the seller's commentary audio static material is classified into the corresponding after-sales service category, and the three consecutive quarters sales data report materials of the product mobile phone are classified into Sales Category.

S203: Arrange and combine the static materials according to the categories of the obtained static materials according to a predetermined strategy; the predetermined strategy at least includes using different arrangements and combinations for different target customer objects.

This step is the core step of this embodiment and is used to generate a dynamic video cover from the static materials collected in the previous step.

The predetermined strategy is a specific way of arranging and combining the static materials according to the categories of the static materials that can exist.

The above-mentioned predetermined strategy is a specific way to select, arrange and combine the static materials according to the categories of the static materials that can exist. Specifically, the predetermined strategy refers to the method in which the materials of the above-mentioned categories are selected. choices, and how to arrange and combine them. For example, the predetermined strategy A is: commodity price - core consumer group - sales volume; the selected static materials include: commodity price, core consumer group information, and sales volume information. The specific arrangement is: according to "commodity price - core consumer group -Sales volume" arrangement.

Similarly, there can also be a predetermined strategy B: product introduction-process/origin advantage-delivery service; predetermined strategy C: after-sales service-cost-effectiveness-product function; predetermined strategy D: product price-product function; predetermined strategy E It is: cost-effectiveness-brand; in short, multiple strategies can be formulated according to the situation. Of course, a certain type of element can also appear multiple times, and there can be more specific selection strategies for which elements to choose in each type of element. The expression forms of the above strategies are only simplified expressions. There can be more detailed material selection and material arrangement and combination methods. The specific descriptions of these methods can also be used as part of the strategy. In this embodiment, only the above simplified policy description method is adopted.

Illustratively, according to the examples in the above steps, the predetermined strategy for the mobile phone product is: predetermined strategy A: mobile phone sales - mobile phone functions - mobile phone appearance; predetermined strategy B: mobile phone brand - mobile phone price - mobile phone after-sales service; predetermined strategy Strategy C: Mobile phone after-sales service - mobile phone appearance - mobile phone cost-effectiveness, and other possible strategies.

The above-mentioned predetermined strategy not only refers to the arrangement sequence itself, but also includes using different arrangement and combination sequences for different target customer objects. The different target customer objects refer to target customer groups with different attributes; because target customer groups with different attributes have different main information that they care about, different permutations and combinations can be used to highlight the information that they mainly care about. To specifically select the order of arrangement and combination for target customer objects with different attributes, AB testing can be used, which will be introduced in detail later.

In this step, in order to combine the above-mentioned deformed static materials, according to the predetermined strategy, the deformed static materials are matched to the predetermined strategy according to their own categories to form a corresponding static material group. Different reservation strategies have particularly significant differences in the corresponding static material groups generated. Due to the different static materials used in the same scheduled strategy, the corresponding static material groups generated are also different.

Illustratively, according to the examples in the above steps, the combination is carried out: first, randomly select the predetermined strategy A of the above-mentioned mobile phone products: mobile phone sales - mobile phone functions - mobile phone appearance, and the predetermined strategy C: mobile phone after-sales service - mobile phone appearance - Cell phone value for money. Secondly, take the above example in S202-1, the text message of the beauty function of the product mobile phone "xxx model mobile phone beauty camera can help you return to 18 years old", the appearance picture of the product mobile phone, and the actual shot of the product mobile phone. The vlog short video (videoweblog, video network log) and the seller's audio commentary static material about the product mobile phone, and the product mobile phone's sales data report material for three consecutive quarters are combined according to the predetermined strategy A and predetermined strategy C to form a static Material group. Finally, the corresponding static material group A combination formed is: the sales data report of the product mobile phone for three consecutive quarters - the beauty function text message of the product mobile phone "xxx model mobile phone beauty camera will help you return to 18 years old" - the product mobile phone's Appearance pictures; the corresponding static material group C combination is: the seller's audio commentary static material about the product mobile phone - the product mobile phone's appearance picture - the actual vlog short video (videoweblog, video network log) of the product mobile phone.

S204: Combine the arranged and combined static materials into a whole to generate the dynamic video cover.

Dynamic video covers contain multiple shots, and each shot can be used independently or jointly to express the selling point features and characteristics of the product. Combine the static material group and use video production to generate a single After the elements are matched, transformed, and edited, the whole is formed as a dynamic video cover. Of course, it is not excluded to use various general video generation tools to generate dynamic videos. Due to the different permutations and combinations used for different target customer groups, a variety of dynamic video covers will be generated in this step.

Illustratively, according to the example in step S203 above, the static material group A will be formed: the three-quarter sales data report of the product mobile phone - the text information of the beauty function of the product mobile phone "xxx model mobile phone beauty camera" - Pictures of the appearance of the product mobile phone; and static material group C: the seller's audio explanation of the product mobile phone static material - Pictures of the appearance of the product mobile phone - Real-shot vlog short videos of the product mobile phone (videoweblog, The above material group of the video network log) is input into the video production and generation unit, and dynamic cover A and dynamic cover C are generated correspondingly. The above dynamic cover A presents the mobile phone sales, mobile phone functions, and mobile phone appearance related to the product mobile phone. The above dynamic cover C presents content related to the mobile phone after-sales service of this product mobile phone - the appearance of the mobile phone - the cost-effectiveness of the mobile phone.

Obviously, there are multiple generation strategies for the above dynamic video covers, so that multiple versions of dynamic video covers can be generated for the same video. Even for the same generation strategy, if the specific material selection is different, multiple fine-tuned versions can be generated. . Therefore, after generating multiple dynamic video covers for the same video subject, it may be necessary to evaluate their actual effects; through evaluation, the version with the best effect is finally selected among multiple possible versions. To this end, embodiments of the present disclosure may further include the following steps:

Please refer to Figure 3, which is a schematic flowchart of a method for generating a target dynamic video cover provided by an embodiment of the present application.

S301. Obtain the first number of dynamic video covers.

The number of generated dynamic video covers is at least a first number. The above first number of dynamic video covers should be understood as a plurality of dynamic video covers, and this embodiment does not limit the specific number.

As mentioned before, due to the wide range of sources of static materials for the main body of the video, the rich and diverse forms of materials, and the ever-changing content of the materials, the number of dynamic video covers that can be formed according to the same predetermined strategy is variable, and the number of dynamic video covers generated according to different predetermined strategies is The number of covers is also variable. In this step, it is assumed that the first number of dynamic video covers obtained are a certain number of dynamic video covers generated according to multiple different predetermined strategies. In other embodiments, it may also refer to a certain number of dynamic video covers generated according to the same predetermined strategy, and the relationship between these dynamic video covers is different fine-tuned versions under the same predetermined strategy.

In the initial step of generating dynamic video covers, the number of generated dynamic video covers is a second number greater than the first number.

For example, what is obtained in this step is the predetermined strategy for the mobile phone product in the above step S202: Predetermined strategy A: mobile phone sales volume - mobile phone function - mobile phone appearance; predetermined strategy B: mobile phone brand - mobile phone price - mobile phone after-sales service ; Predetermined strategy C: mobile phone after-sales service - mobile phone appearance - mobile phone cost-effectiveness, three types of dynamic video covers generated correspondingly under the three predetermined strategies. For example, the total number of three types of dynamic video covers generated is 300, which is the second The quantity is 300. In this step, the first number of dynamic video covers obtained is 15, that is, the first number is 15. Among them, the predetermined strategy A obtained: mobile phone sales-mobile phone functions-dynamic video covers of mobile phone appearance The number of reservations obtained is strategy B: mobile phone brand-mobile phone price-mobile phone after-sales service dynamic video cover is 5; the reservation strategy C obtained: mobile phone after-sales service-mobile phone appearance-mobile phone cost-effective dynamic video cover is 5 indivual. There can be many methods for selecting the first number of dynamic video covers from the second number of initial dynamic video covers, including: using a specially trained machine learning model, expert selection, or even random selection. The above-mentioned second number of initial dynamic video covers can also be automatically generated using a machine learning model according to a predetermined strategy.

S302. For the first number of dynamic video covers, select several sample customers from the target customer group and conduct a first prediction test.

The target customer group is a customer group with target attributes after classifying the customer groups. Target attributes refer to attributes labeled according to customer demand characteristics.

For a customer group with a specific target attribute, before conducting the first estimated test, select from a second number of dynamic video covers more than the first number in a predetermined method based on the specific target attribute. The first number of dynamic video covers used for testing.

What needs to be understood here is that the positioning of the dynamic video cover of the e-commerce application platform is to show consumers the core characteristics of the product and the most attractive characteristics to consumers in a short period of time. Different customers pay different attention to the same product. For example, some customers are price-sensitive, some have higher requirements for after-sales service, and some customers are more trusting in soft power such as brand influence and sales volume. Therefore, it is necessary to label users with different demand characteristics for the same product.

Among them, the way to classify customer groups is to analyze each customer's purchasing behavior on multiple shopping application platforms, and label the customer's purchasing demand characteristics. It can also be based on the customer's focus on pre-purchased goods, Label the demand points, and then label the demand characteristics The same customers are divided into the same group.

For example, on a certain shopping application platform, the product is a mobile phone, and the customers whose purchase demand for mobile phones is labeled as sales volume are divided into one type of customer group, and the customers whose purchase demand is labeled as price are divided into one type of customer group. The purchasing requirements can also be divided into one type of customer group. Customers whose needs are tagged with product functions and sales volume are divided into one type of customer group. Customers whose purchase needs are tagged with brand, price, process/origin advantages are divided into one type of customer group. The above divisions can be in various forms, and the classification process Reference may be made to relevant prior art, which will not be described in detail in this disclosure.

Here, it should be reminded that the target customer group is a customer group that matches the test content of the first estimated test. Select a number of sample customers from the target customer group and conduct the first prediction test.

The first prediction test refers to a test based on the content characteristics of the dynamic video cover. In this step, the dynamic video cover generated according to the predetermined strategy will be matched with the corresponding target customer group according to the content characteristics of the dynamic video cover, and some sample customers will be selected to perform a first prediction test on the above dynamic video cover. The first estimated test form is not specifically limited in this disclosure. For example, the first prognostic test is in the form of an AB test.

Illustratively, according to the example in step S301 above, the first number of 150 dynamic video covers obtained is subjected to the first prediction test. Taking the dynamic video cover of the predetermined strategy A as an example, the five dynamic video covers of the predetermined strategy A: mobile phone sales - mobile phone functions - mobile phone appearance are first estimated and tested; according to the content characteristics of the dynamic video cover, the mobile phone The three characteristics of sales volume, mobile phone functions, and mobile phone appearance match the corresponding target customer groups as those who are concerned about the above three characteristics. We will not do a weight analysis of the three content characteristics of the target customer groups here. For example, the target customer group that pays attention to the above three characteristics is 150,000 customers. At this time, a number of 500 sample customers are randomly selected to do the first estimate test of the dynamic video cover content. The first estimate test is carried out using the AB test method. . The above five dynamic video covers of mobile phone sales - mobile phone functions - mobile phone appearance were randomly displayed to the above 500 sample customers. The customers browsed and purchased the product mobile phones according to their personal preference and interest. In the same way, the first estimated test process is performed on the 5 dynamic video covers of the predetermined strategy B: mobile phone brand - mobile phone price - mobile phone after-sales service; and the predetermined strategy C: mobile phone after-sales service - mobile phone appearance - mobile phone cost-effectiveness 5 A dynamic video cover is processed for the first prediction test.

S303: Obtain a target dynamic video cover that matches the target group based on the test feedback data and the predetermined evaluation criteria.

Test data based on the first estimated test feedback, such as browsing time data of sample users during the test period, purchase order data, order payment ratio data, user forwarded recommended product data, repurchase data of purchased users, and evaluations of purchased users Data and other aspects of data for comprehensive evaluation.

The predetermined evaluation standard is to grade the above-mentioned first estimated test feedback test data, and use the dynamic video cover with the highest comprehensive score as the target dynamic video cover for this type of user.

Illustratively, according to the example in step S302 above, the third-party platform is used to track the mobile phone browsing and purchasing status of the product of the above 500 sample customers, and the comprehensive performance will be included in the five predetermined dynamic video covers of strategy A. The dynamic video cover with the highest score is used as the target dynamic video cover for the target customer group of mobile phone sales-mobile phone functions-mobile phone appearance. At the same time, the target dynamic video cover of the mobile phone brand-mobile phone price-mobile phone after-sales service target customer group can also be determined from the five predetermined dynamic video covers of strategy B, and the target dynamic video cover of the five predetermined strategy C can be determined. In the cover, the target dynamic video cover of the target customer group of mobile phone after-sales service-mobile phone appearance-mobile phone cost-effectiveness is determined.

Among them, for different categories of target customer groups, the determined target dynamic video covers are different, so that the dynamic video covers of the goods to be purchased seen by each type of users on the shopping application platform have real-time pertinence, and increase the user's attitude toward the purchased goods. The stickiness, click-through rate and number of views can further enhance users’ exposure and purchase behavior of products.

Optionally, after obtaining the first estimated test data, the CTR estimation method, recommendation algorithm and other methods can be combined to optimize the order or weight ratio of the logical expression of the product selling points according to the predetermined strategy.

Optionally, after obtaining the first estimated test data, you can combine CTR estimation methods, recommendation algorithms and other methods, and then intervene in the step of generating dynamic video covers based on the static materials according to the predetermined strategy to predict in advance. A reservation strategy with better user performance can reduce the number of dynamic video covers generated. At the same time, it can also significantly reduce the number of dynamic video covers that need to be tested during the first estimated test process.

After determining the target dynamic video cover, the embodiment of the present disclosure further includes the following steps:

Please refer to Figure 4. Figure 4 is a schematic flow chart of a method for generating a fine-tuned dynamic video cover that matches the target group provided by an embodiment of the present application. It can be used in the implementation environment as shown in Figure 1;

S401: Obtain the target dynamic video cover.

Obtain the above-mentioned target dynamic video cover that matches the target group. What needs to be understood here is that the target customer group has preliminarily sorted and screened the content characteristics of the dynamic video cover based on their own satisfaction and personal preferences. However, in terms of the expression form of the content characteristics of the target dynamic video cover, no corresponding screening has been made. Therefore, further optimization and adjustment are needed to obtain the target dynamic video cover of the target group.

Illustratively, according to the example in step S303 above, the target dynamic video cover as the target customer group of mobile phone sales-mobile phone functions-mobile phone appearance is obtained.

S402: Adjust the target dynamic video cover to obtain multiple fine-tuned target dynamic video covers.

Among them, the adjustment of the target dynamic video cover refers to the adjustment of the expression form of the content characteristics of the target dynamic video cover, and the content characteristics that the target group pays attention to are presented through different expression forms. The above-mentioned fine-tuning of the expression form can be understood as, for example, changing the target dynamic video to a sitcom-type form for presentation, or it can be changed to a commentary-type form, or it can be changed to a vlog diary-type form, a talk show-type form, Various form adjustments such as the same style.

Illustratively, according to the example in step S401 above, the expression form of the target dynamic video cover, which is the target customer group of mobile phone sales-mobile phone functions-mobile phone appearance, is fine-tuned. The mobile target dynamic video cover can be fine-tuned into: a funny sitcom type mobile target dynamic video cover, a real-life vlog diary type mobile target dynamic video cover, an expert explanation type mobile target dynamic video cover, and a multi-user evaluation talkback Show-type mobile phone target dynamic video covers and other forms; for example, get 4 fine-tuned mobile phone product target dynamic video covers.

S403. For the plurality of fine-tuned target dynamic video covers, select several sample customers from the target customer group and conduct a second prediction test.

The second prediction test refers to a test based on the expression form of the content characteristics of the dynamic video cover. In this step, the multiple target dynamic video covers after fine-tuning the expression form are selected from the above-mentioned target customer groups, and the target dynamic video covers after the fine-tuning expression form are put into the second expression form. Predictive testing. The second estimated test form is not specifically limited in this disclosure. For example, the second estimate test is in the form of an AB test.

Illustratively, according to the example in step S402 above, 300 customers from the target customers of mobile phone sales - mobile phone functions - mobile phone appearance are randomly selected as sample customers, and the 4 fine-tuned mobile phone target dynamic video covers obtained are put into Conduct second estimated AB testing among sample customers.

S404: According to the test feedback data, obtain a fine-tuned dynamic video cover that matches the target group.

Test data based on the second estimated test feedback, such as browsing time data of sample users during the test period, purchase order data, order payment ratio data, user forwarded recommended product data, repurchase data of purchased users, and evaluations of purchased users Data and other aspects of data for comprehensive evaluation. The fine-tuned dynamic video cover with the highest overall score will be matched to this type of user.

Illustratively, according to the example in step S402 above, 4 fine-tuned mobile phone target dynamic video covers of the comedy sitcom type, real-life vlog diary type mobile phone target dynamic video covers, and expert commentary type mobile phone target dynamic video covers are used. Comprehensive analysis and comparison of the test data of video covers and mobile target dynamic video covers of the talk show category evaluated by multiple users; for example, if a fine-tuned mobile target dynamic video cover of the comedy sitcom type has the first overall score, then the fine-tuned The mobile phone dynamic video cover of the funny sitcom type is used as a fine-tuned dynamic video cover for the mobile phone sales-mobile phone function-mobile phone appearance target customer group, and the dynamic video cover is displayed to such target customer groups.

Optionally, after obtaining the second estimated test data, the CTR estimation method, recommendation algorithm and other methods can be combined to intervene in the step of "generating a dynamic video cover based on the static material according to the predetermined strategy" to advance the Predicting the expression forms that users perform well can reduce the number of dynamic video covers generated. At the same time, it can also significantly reduce the number of dynamic video covers that need to be tested during the first estimate test, further improving the second estimate. Test efficiency.

The above-mentioned second embodiment provides a dynamic video cover generation method. Correspondingly, the third embodiment of the present application also provides an application device of the dynamic video cover generation method. Since the device embodiment is basically similar to the method embodiment, the description is relatively simple. For details of the relevant technical features, please refer to the corresponding description of the method embodiment provided above. The following description of the device embodiment is only illustrative. of.

Please refer to Figure 5 to understand this embodiment. Figure 5 is a unit block diagram of the device provided by this embodiment. As shown in Figure 5, the device provided by this embodiment includes:

The material acquisition unit 501 is configured to acquire static materials related to the main body of the video.

The material processing unit 502 is configured to classify the obtained static materials related to the main body of the video into corresponding static material categories according to the nature of the content.

The material modification unit 503 is configured to arrange and combine the static materials according to the obtained static material categories according to a predetermined strategy; the predetermined strategy at least includes different target customer objects, using different arrangements and combinations.

The video generation unit 504 is configured to integrate the combined static materials into a whole and generate the dynamic video cover.

Obtain the first number of dynamic video covers;

Obtain the target dynamic video cover;

Adjust the target dynamic video cover to obtain multiple fine-tuned target dynamic video covers;

Optionally, the number of dynamic video covers is a second number greater than the first number, and the The method also includes: targeting a customer group with a specific target attribute, before conducting the first prediction test, according to the specific target attribute, using a predetermined method to select a plurality of dynamic videos from a second number more than the first number. Among the covers, the first number of dynamic video covers are selected for testing.

In the above embodiments, a dynamic video cover generation method and an application device of the dynamic video cover generation method are provided. In addition, the fourth embodiment of the present application also provides an electronic device. Examples of the electronic device are as follows:

Please refer to Figure 6 to understand this embodiment. Figure 6 is a schematic diagram of the electronic device provided in this embodiment.

As shown in Figure 7, the electronic device includes: a processor 601; a memory 602;

The memory 602 is used to store application programs. When the program is read and executed by the processor 601, it performs the following operations:

Get static material related to the main body of the video.

The combined static materials are formed into a whole to generate the dynamic video cover. Optionally, according to the different predetermined strategies, the obtained static materials related to the main body of the video can generate multiple versions of dynamic video covers; for the same predetermined strategy, due to the different selection of the static materials, also A variety of different versions of dynamic video covers can be generated.

Optionally, forming the combined static materials into a whole to generate the dynamic video cover includes:

Obtain the first number of dynamic video covers;

Based on test feedback data and predetermined evaluation criteria, obtain results that match the target group Target dynamic video cover.

Obtain the target dynamic video cover;

In order to facilitate the understanding of this embodiment, the dynamic video cover generation method provided by the disclosed embodiments of this application is generally executed by a computer device with certain computing capabilities. The computer device includes, for example: a terminal device or a server or other processing device, a terminal The device can be a user equipment (User Equipment, UE), a mobile device, a user terminal, a terminal, a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the dynamic video cover generation method can be implemented by the processor calling computer-readable instructions stored in the memory.

The fifth embodiment of the present application provides a storage medium embodiment corresponding to the above-mentioned embodiment, on which one or more computer instructions are stored; these instructions are executed by a processor to implement the method described in the above-mentioned second embodiment.

Figure 7 is a schematic structural diagram of a storage medium provided for the fourth embodiment of the present application. As shown in Figure 7, the storage medium 700 of the present application includes: a computer-readable storage medium 701 and a processor 702.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

Memory may include non-permanent storage in computer-readable media, random access memory (RAM) and/or non-volatile memory in the form of read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information may be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), and read-only memory. (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, Magnetic tape cassettes, tape magnetic disk storage or other magnetic storage devices or any other non-transmission medium can be used to store information that can be accessed by a computing device. As defined in this article, computer-readable media does not include non-transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or to equivalently replace some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application. .

Claims

A dynamic video cover generation method, which is characterized by including:

Obtain static materials related to the main body of the video;

Classify the obtained static materials related to the main body of the video into the corresponding static material categories according to the nature of the content;

According to a predetermined strategy, the static materials are arranged and combined according to the obtained static material categories; the predetermined strategy at least includes using different arrangements and combinations for different target customer objects;

The arranged and combined static materials are formed into a whole to generate the dynamic video cover.
The method according to claim 1, characterized in that the method further includes: according to the different predetermined strategies, the obtained static materials related to the video body can generate multiple versions of dynamic video covers; for The same predetermined strategy can also generate multiple different versions of dynamic video covers due to different selections of the static materials.
The method according to claim 1, forming the arranged and combined static materials into a whole to generate the dynamic video cover, including:

Before generating the dynamic video cover, the expression method of the static material related to the video body is deformed.
The method according to claim 3, wherein the deformation processing of the expression method of the static material related to the video subject includes: beautifying the obtained static material by selecting and using a template of a design style.
The method according to claim 1, wherein in the method, the number of dynamic video covers is at least a first number, and the method further includes:

Obtain the first number of dynamic video covers;

For the first number of dynamic video covers, select a number of sample customers from the target customer group and conduct a first prediction test;

Based on the test feedback data and predetermined evaluation criteria, a target dynamic video cover that matches the target customer group is obtained.
The method according to claim 5, characterized in that after determining the target dynamic video cover, the following steps are further performed:

Obtain the target dynamic video cover;

Adjust the target dynamic video cover to obtain multiple fine-tuned target dynamic video covers;

For the plurality of fine-tuned target dynamic video covers, extract a number of sample customers from the target customer group and conduct a second prediction test;

Based on the test feedback data, a fine-tuned dynamic video cover that matches the target customer group is obtained.
The method according to claim 5 or 6, characterized in that the target customer group is a customer group with target attributes obtained after classifying customer groups.
The method of claim 7, wherein classifying customer groups includes: dividing customers with the same demand characteristic labels into the same type of customer groups.
The method according to claim 7, characterized in that the target attributes are attributes marked according to customer demand characteristic tags.
The method according to claim 5 or 6, characterized in that the number of dynamic video covers is a second number greater than the first number, and the method further includes: targeting customer groups with specific target attributes, Before conducting the first estimated test, select the first number of dynamic video covers for testing from a second number of dynamic video covers more than the first number in a predetermined method according to the specific target attribute. Dynamic video cover.
A device for generating dynamic video covers, which is characterized by including:

a material acquisition unit configured to acquire static materials related to the main body of the video;

The material processing unit is configured to classify the obtained static materials related to the video subject into corresponding static material categories according to the nature of the content;

The material transformation unit is configured to arrange and combine the static materials according to the obtained static material categories according to a predetermined strategy; the predetermined strategy at least includes using different arrangements and combinations for different target customer objects;

The video generation unit is configured to integrate the arranged and combined static materials into a whole and generate the dynamic video cover.
An electronic device for dynamic video cover generation, characterized by including: a memory and a processor;

The memory is used to store program instructions;

The processor is configured to call program instructions in the memory to execute the method according to any one of claims 1 to 10.
A storage medium generated by dynamic video cover, characterized in that it is used to save calculations A computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.