Clothing design online advertisement service platform based on user feedback

Abstract

The invention relates to the technical field of costume design online advertising, and discloses a costume design online advertising service platform based on user feedback, which comprises the following steps: connecting display videos of all presale costumes end to end to obtain a costume advertisement; if the user pauses when watching the clothing advertisement, recording the display video at the pause position as a marked video; the display adding strategy is executed, a picture set is displayed at the pause position of the user, and the picture set comprises a plurality of presale clothing display pictures; calculating the average playing speed of the clothing advertisement watched by each user; executing a multi-speed constant value strategy, predicting the probability of purchasing pre-sold clothes when the user watches the clothes advertisement at the average playing multi-speed, and recording the probability as a prediction probability; executing the matched display strategy, and displaying the matched video of the presale clothing; the average playing speed of the clothing advertisement watched by the user is calculated, the probability that the user purchases the presale clothing is predicted, clothing merchants are helped to determine the sales amount of the presale clothing, the insight of the market is increased, and the marketing effect is improved.

Description

Technical Field

[0001] The invention relates to the technical field of clothing design online advertising, in particular to a clothing design online advertising service platform based on user feedback. Background Art

[0002] In the advertising strategy of pre-sale clothing, it is very important to provide user-controllable playback speed and pause functions, which involves user behavior analysis and its potential impact on sales. By tracking and analyzing the interactive data of users when watching clothing advertisements, in-depth insights into user interests and preferences can be obtained. Using big data and machine learning technology, companies can analyze these user behavior data and optimize advertising content and positioning strategies accordingly to improve the conversion rate and effectiveness of advertisements. In addition, the introduction of this playback control function not only enhances the user's viewing experience, but also provides brands with a way to fine-tune products and marketing strategies through user interaction. In short, users' playback speed selection and pause behavior for clothing advertisements is an effective tool for predicting pre-sale clothing sales, which helps companies better understand market demand and thus formulate more effective sales strategies.

[0003] In the prior art, when users watch advertisements for pre-sale clothing, there will be different viewing speeds and multiple pauses during the viewing process. These viewing behaviors represent the users' purchase intention for the pre-sale clothing. However, the prior art often ignores this important user behavior when predicting the sales of pre-sale clothing or formulating sales strategies, resulting in poor sales results.

[0004] This solution proposes to add a picture collection to the user's pause point, calculate the average playback speed based on the browsing time, actual viewing time and advertising time, obtain the first average purchase ratio in the target interval where the average playback speed is located, and then predict the sales of each pre-sale clothing. Summary of the invention

[0005] The present invention provides a clothing design online advertising service platform based on user feedback, which is used to promote the solution of the problems mentioned in the above background technology.

[0006] The present invention provides the following technical solution: a clothing design online advertising service platform based on user feedback, comprising:

[0007] The clothing pre-sold by clothing merchants is recorded as pre-sold clothing;

[0008] For each pre-sale garment, shoot a display video of the pre-sale garment;

[0009] Connect all pre-sale clothing display videos end to end to obtain clothing advertisements;

[0010] Get the slowest speed for playing clothing advertisements;

[0011] Get the fastest speed for playing clothing advertisements;

[0012] Execute the speed setting strategy and set several playback speeds for clothing advertisements;

[0013] When users view an apparel ad:

[0014] If the user pauses while watching a clothing advertisement, the displayed video at the pause point is recorded as a marked video;

[0015] Execute an additional display strategy to display a picture set at the user's pause point, wherein the picture set includes multiple display pictures of pre-sale clothing;

[0016] When a user opens a photo collection, the time from when the user opens the photo collection to when the user closes the photo collection is obtained and recorded as the browsing time;

[0017] Get the total actual playing time of the clothing advertisement for each user when the advertisement is played, and record it as the actual viewing time;

[0018] Get the duration of the clothing advertisement, recorded as the advertisement duration;

[0019] Calculate the advertisement duration / (actual viewing duration + browsing duration), and record the result as the average playback speed; calculate the average playback speed of each user watching the clothing advertisement;

[0020] Implement speed division strategy and define multiple speed intervals for playing clothing advertisements;

[0021] Get the speed interval where the average playback speed is located, and record it as the target interval;

[0022] Execute the speed setting strategy to predict the probability of users purchasing pre-sale clothing when watching clothing advertisements at the average playback speed, which is recorded as the predicted probability;

[0023] Get the predicted probabilities of all users who watched the clothing ads and calculate the sum as the predicted sales of each pre-sold clothing item;

[0024] Get matching outfits for each pre-sale outfit;

[0025] Shoot a display video for the matching clothes of each pre-sale clothing to obtain the matching video of each pre-sale clothing;

[0026] Implement a matching display strategy to show matching videos for pre-sale clothing.

[0027] Optionally, the shooting of a display video of the pre-sale clothing includes:

[0028] Get any pre-sale clothing;

[0029] Get the front and back of pre-sale clothing;

[0030] Recording the two sides of the pre-sale garment as a first side and a second side respectively;

[0031] Get the above and below of pre-sale clothing;

[0032] The display video of the pre-sale clothing is shot in the order of the front, the first side, the back, the second side, the top and the bottom.

[0033] Optionally, the execution speed setting strategy includes:

[0034] Setting the division interval for determining the speed of clothing advertisement playback;

[0035] Calculate (fastest speed - slowest speed) / division interval = number of divisions n;

[0036] Calculate the slowest speed + divide the interval, and record the result as the first speed;

[0037] Calculate the slowest speed + division interval * 2, and record the result as the second speed;

[0038] …

[0039] Calculate the slowest speed + divided interval * n = fastest speed;

[0040] Then, the slowest speed, the first speed, the second speed ... the fastest speed are all the set playback speeds for playing the clothing advertisement.

[0041] Optionally, the step of executing the adding display strategy includes:

[0042] Extract the image of each frame of the labeled video;

[0043] The extracted images are grouped into a frame set in chronological order;

[0044] Setting a plurality of selection numbers for grouping pictures in a frame set;

[0045] The multiple selected quantities are recorded in ascending order as the first selected quantity, the second selected quantity, ... the rth selected quantity;

[0046] Get the number of selections in order from small to large;

[0047] For the obtained number of selections, the obtained number of selections is recorded as the number of markings;

[0048] S1, dividing the frame image set into groups of the number of marks;

[0049] S2, obtaining the last image in each group, recorded as the analog image of each group;

[0050] S3, obtaining any two adjacent analog images and performing similarity comparison;

[0051] S4, setting a threshold similarity for determining a set of images;

[0052] S5. When the similarity between any two adjacent analog images is less than the threshold similarity, all analog images form a picture set;

[0053] S6. When there are two adjacent analog images whose similarity is greater than or equal to the threshold similarity, the next selected number of the number of marks is obtained as the new number of marks, and S1-S6 are repeated.

[0054] Optionally, the execution of the speed division strategy includes:

[0055] Get the partition interval;

[0056] Calculate 0 + division interval, and record the result as the first boundary speed;

[0057] Calculate the first boundary speed + the division interval, and record the result as the second boundary speed;

[0058] Calculate the second boundary speed + the division interval, and record the result as the third boundary speed;

[0059] …

[0060] Calculate the t-1th boundary speed + division interval, and record the result as the tth boundary speed, where the value of t is infinite;

[0061] Then, the interval formed by the first boundary speed and the second boundary speed is recorded as the first speed interval;

[0062] The interval formed by the second boundary speed and the third boundary speed is recorded as the second speed interval;

[0063] …

[0064] The section formed by the t-1th boundary speed and the tth boundary speed is recorded as the t+1th speed section.

[0065] Optionally, the execution of the speed setting strategy includes:

[0066] When a user has previously watched a clothing advertisement of a clothing merchant and then purchased clothing, the user is recorded as the first user;

[0067] The clothing advertisement where the purchased clothing appeared is recorded as a historical clothing advertisement;

[0068] For each historical clothing ad:

[0069] Get the total number of clothing displayed in historical clothing advertisements, and record it as the total display amount;

[0070] Obtain the number of clothing items in the historical clothing advertisements purchased by the user, and record it as the total purchase amount;

[0071] Calculate total purchase amount / total display amount, and record the result as purchase ratio.

[0072] Optionally, the execution of the speed setting strategy includes:

[0073] For each first user:

[0074] Obtain the average playback speed of all historical clothing advertisements purchased by the first user, and obtain multiple historical speeds;

[0075] Classify the historical speed into corresponding speed intervals;

[0076] Obtain the purchase ratios corresponding to all historical speed multiples in the target interval and record them as the first purchase ratio, the second purchase ratio, ... the uth purchase ratio;

[0077] Execute the following formula to calculate the first average purchase ratio of the target interval;

[0078] (first purchase ratio + second purchase ratio + ... + uth purchase ratio) / u = first average purchase ratio;

[0079] Then, the first average purchase ratio in the target interval is used as the predicted probability of the first user.

[0080] Optionally, the execution of the speed setting strategy includes:

[0081] When the user has not purchased clothing by viewing clothing advertisements of clothing merchants before, the user is recorded as a second user;

[0082] Obtaining a target interval for a second user;

[0083] The first average purchase ratios corresponding to all first users in the target interval are obtained, the mean is calculated, and the result is recorded as the second average purchase ratio, which is used as the predicted probability of the second user.

[0084] Optionally, the executing supporting display strategy includes:

[0085] For each user:

[0086] Get the predicted probability of the user;

[0087] Set the threshold probability for determining whether to recommend a matching video to a user;

[0088] When the predicted probability ≧ threshold probability;

[0089] Connect the supporting videos of each pre-sale clothing from beginning to end to obtain a first supporting advertisement, and recommend the first supporting advertisement to the user;

[0090] When the predicted probability < threshold probability;

[0091] Acquire the matching videos of the pre-sale clothing in the picture collection viewed by the user, connect the matching videos end to end to obtain a second matching advertisement, and recommend the second matching advertisement to the user.

[0092] The present invention has the following beneficial effects:

[0093] 1. The clothing design online advertising service platform based on user feedback shoots a display video for each pre-sale clothing, connects the display videos end to end to obtain clothing advertisements, and only displays one pre-sale clothing at a time, so that only one pre-sale clothing can be displayed on a display interface, which is convenient for users to watch only one pre-sale clothing at a time, avoiding visual confusion and improving user satisfaction in watching clothing advertisements.

[0094] 2. The clothing design online advertising service platform based on user feedback can meet users' different viewing habits and time limits by setting multiple playback speeds. Users can choose the appropriate viewing speed according to their needs. If users just want to quickly browse clothing advertisements, providing accelerated playback can help users save time. If users are particularly interested in a certain piece of clothing, providing slowed-down playback or normal speed can give users more time to observe the texture, style and other details of the clothing. Or for users with slower visual processing speeds, a slower playback speed can help them better understand and receive information. Therefore, setting multiple playback speeds for clothing advertisements can play an important role in enhancing user satisfaction, improving advertising effectiveness and strengthening user engagement.

[0095] 3. The clothing design online advertising service platform based on user feedback cannot determine whether the user pauses at a certain point in a clothing advertisement because the user wants to carefully observe the details of the marked clothing or because he / she is temporarily busy and cannot continue to watch the clothing advertisement. At this time, a picture album is displayed at the point where the user paused. If the user opens the picture album, it means that the user wants to carefully observe the details of the marked clothing, so the picture album can show the details of the marked clothing in more detail. If the user does not open the picture album, it means that the user is temporarily busy. Users who are temporarily busy do not have browsing time, and whether the user wants to carefully observe the details of the marked clothing is determined by displaying the picture album. At the same time, the picture album contains multiple display pictures of the marked clothing, which is convenient for the user to observe carefully.

[0096] 4. The online advertising service platform for clothing design based on user feedback obtains a frame image set by extracting the image of each frame of the marked video, but the frame image set contains multiple similar images; by setting multiple selection quantities, and selecting each selection quantity in order from small to large, the selected selection quantity is recorded as the marking quantity, the frame image set is divided into the marking number of groups, the last image is obtained from each group, the analog image of each group is obtained, and the similarity of adjacent analog images is compared. When the similarity of adjacent analog images is less than the threshold similarity, it is considered that each analog image shows an image of a different orientation of the marked clothing, that is, each analog image is different, and all analog images are formed into a picture set; when there are adjacent analog images whose similarity is greater than the threshold similarity, it is considered that at least two analog images are the same, and the marking quantity is updated at this time until each analog image is different, and different analog images are selected by the threshold similarity to form a picture set, thereby reducing the number of pictures viewed by the user while displaying pictures of different orientations of the marked clothing, which is convenient for the user to observe carefully.

[0097] 5. The online advertising service platform for clothing design based on user feedback, when a user opens a picture collection, obtains the browsing time of the user viewing the picture collection, obtains the actual viewing time of the user, calculates the advertisement time / (actual viewing time + browsing time), and obtains the average playback speed of the user viewing the clothing advertisement, that is, the time the user browses the picture collection is counted as the time the user views the clothing advertisement, and then calculates the average playback speed, and counts the time the user views the picture collection as the time the user views the clothing advertisement. The reason is that the purpose of the user browsing the picture collection and watching the clothing advertisement is to better view the clothing, but the viewing form is different. For the convenience of calculation, the time of browsing the picture collection is counted as the time of watching the clothing advertisement, and then calculates the average playback speed. The user's willingness to buy is reflected by the average playback speed, which simplifies the calculation, facilitates the subsequent recommendation of supporting advertisements to the user, and improves the user's satisfaction.

[0098] 6. The clothing design online advertising service platform based on user feedback, a clothing advertisement displays multiple pre-sale clothing. The first user may purchase multiple clothing after watching the clothing advertisement. Therefore, the average playback speed of the first user's historical clothing advertisements is obtained, and the purchase ratio of the historical clothing advertisements is calculated. In a historical clothing advertisement, an average playback speed corresponds to a purchase ratio. All historical speeds are divided into speed intervals, all historical speeds in the target interval are obtained, the purchase ratio of each historical speed is calculated, the average is calculated, and the first average purchase ratio of the target interval is obtained. The first average purchase ratio is the probability that the first user purchases each pre-sale clothing. The predicted probability of each pre-sale clothing is obtained. The first average purchase ratio of the target interval is calculated through the correspondence between the average playback speed and the purchase ratio of the historical clothing advertisements. When the first user is watching the clothing advertisement of the pre-sale clothing, the playback speed is in the target interval, and the probability that the first user purchases each pre-sale clothing is the first average purchase ratio, thereby predicting the sales of the pre-sale clothing, which is convenient for clothing merchants to allocate goods.

[0099] 7. The clothing design online advertising service platform based on user feedback obtains the first average purchase ratio of all first users in the target interval and calculates the mean to obtain the predicted probability of the second user, that is, predicts the probability of the second user purchasing each pre-sale clothing item based on the experience value, so that the data has a basis and the prediction is more reasonable.

[0100] 8. The clothing design online advertising service platform based on user feedback obtains the supporting clothing of each pre-sale clothing. When the predicted probability calculated after the user watches the clothing advertisement is greater than or equal to the threshold probability, it indicates that the user has a high probability of purchasing each pre-sale clothing. At this time, the supporting videos of each pre-sale clothing are connected end to end to obtain a first supporting advertisement, and the first supporting advertisement is pushed to the user to increase the user's satisfaction and increase the sales of the clothing merchant. When the predicted probability calculated after the user watches the clothing advertisement is less than the threshold probability, if the user opens a picture set of a pre-sale clothing while watching the clothing advertisement, it indicates that the user is interested in the pre-sale clothing in the opened picture set. Then, the supporting videos of the pre-sale clothing in the picture set viewed by the user are obtained, connected end to end to obtain a second supporting advertisement, and the second supporting advertisement is recommended to the user. According to the user's predicted probability and the situation of opening the picture set, the supporting advertisement is personalized recommended to the user, thereby increasing the user's satisfaction and the possibility of the user purchasing the pre-sale clothing. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0102] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0103] Embodiment 1, refer to Figure 1 , a clothing design online advertising service platform based on user feedback, including:

[0104] The clothing pre-sold by clothing merchants is recorded as pre-sold clothing;

[0105] For each pre-sale garment, shoot a display video of the pre-sale garment;

[0106] Connect all pre-sale clothing display videos end to end to obtain clothing advertisements;

[0107] Get the slowest speed for playing clothing advertisements;

[0108] Get the fastest speed for playing clothing advertisements;

[0109] Execute the speed setting strategy and set several playback speeds for clothing advertisements;

[0110] When users view an apparel ad:

[0111] If the user pauses while watching a clothing advertisement, the displayed video at the pause point is recorded as a marked video;

[0112] Execute an additional display strategy to display a picture set at the user's pause point, wherein the picture set includes multiple display pictures of pre-sale clothing;

[0113] When a user pauses at a certain point in a clothing advertisement, it cannot be determined whether the user pauses the advertisement because he or she wishes to observe the details of the marked clothing carefully or because he or she has something else to do and cannot continue viewing the clothing advertisement. In this case, a picture album is displayed at the point where the user paused. If the user opens the picture album, it means that the user wishes to observe the details of the marked clothing carefully, and the picture album can display the details of the marked clothing in more detail. If the user does not open the picture album, it means that the user has something else to do, and users who have something else to do do not have the time to browse. By displaying the picture album, it can be determined whether the user wishes to observe the details of the marked clothing carefully. At the same time, the picture album contains multiple display pictures of the marked clothing, which is convenient for users to observe carefully.

[0114] When a user opens a photo collection, the time from when the user opens the photo collection to when the user closes the photo collection is obtained and recorded as the browsing time;

[0115] Get the total actual playing time of the clothing advertisement for each user when the advertisement is played, and record it as the actual viewing time;

[0116] Get the duration of the clothing advertisement, recorded as the advertisement duration;

[0117] Calculate the advertisement duration / (actual viewing duration + browsing duration), and record the result as the average playback speed; calculate the average playback speed of each user watching the clothing advertisement;

[0118] When a user opens a picture collection, the browsing time of the user viewing the picture collection is obtained, the actual viewing time of the user is obtained, the advertisement time / (actual viewing time + browsing time) is calculated, and the average playback speed of the user viewing the clothing advertisement is obtained. That is, the time the user browses the picture collection is counted as the time the user views the clothing advertisement, and then the average playback speed is calculated. The time the user views the picture collection is counted as the time the user views the clothing advertisement. The reason is that the time the user browses the picture collection and the time the user views the clothing advertisement are both for the purpose of better viewing the clothing, but in different forms of viewing. For the convenience of calculation, the time the user browses the picture collection is counted as the time the user views the clothing advertisement, and then the average playback speed is calculated. The user's willingness to buy is reflected by the average playback speed, which simplifies the calculation, facilitates the subsequent recommendation of supporting advertisements to the user, and improves user satisfaction.

[0119] Implement speed division strategy and define multiple speed intervals for playing clothing advertisements;

[0120] Get the speed interval where the average playback speed is located, and record it as the target interval;

[0121] Execute the speed setting strategy to predict the probability of users purchasing pre-sale clothing when watching clothing advertisements at the average playback speed, which is recorded as the predicted probability;

[0122] Get the predicted probabilities of all users who watched the clothing ads and calculate the sum as the predicted sales of each pre-sold clothing item;

[0123] Get matching outfits for each pre-sale outfit;

[0124] Shoot a display video for the matching clothes of each pre-sale clothing to obtain the matching video of each pre-sale clothing;

[0125] Implement a matching display strategy to show matching videos for pre-sale clothing.

[0126] The shooting of the display video of the pre-sale clothing includes:

[0127] Get any pre-sale clothing;

[0128] Get the front and back of pre-sale clothing;

[0129] Recording the two sides of the pre-sale garment as a first side and a second side respectively;

[0130] Get the above and below of pre-sale clothing;

[0131] The display video of the pre-sale clothing is shot in the order of the front, the first side, the back, the second side, the top and the bottom.

[0132] By shooting a display video for each pre-sale clothing item and connecting the display videos end to end to obtain a clothing advertisement, only one pre-sale clothing item is displayed each time, which is convenient for displaying only one pre-sale clothing item on a display interface, allowing users to view only one pre-sale clothing item at a time, avoiding visual confusion and improving user satisfaction in viewing clothing advertisements.

[0133] The execution speed setting strategy includes:

[0134] Setting the division interval for determining the speed of clothing advertisement playback;

[0135] Calculate (fastest speed - slowest speed) / division interval = number of divisions n;

[0136] Calculate the slowest speed + divide the interval, and record the result as the first speed;

[0137] Calculate the slowest speed + division interval * 2, and record the result as the second speed;

[0138] …

[0139] Calculate the slowest speed + divided interval * n = fastest speed;

[0140] Then, the slowest speed, the first speed, the second speed ... the fastest speed are all the set playback speeds for playing the clothing advertisement.

[0141] By setting multiple playback speeds to meet users' different viewing habits and time limits, users can choose the appropriate viewing speed according to their needs; if users just want to quickly browse clothing ads, providing accelerated playback can help users save time; if users are particularly interested in a certain piece of clothing, providing decelerated playback or normal speed can give users more time to observe the texture, style and other details of the clothing; or for users with slower visual processing speeds, a slower playback speed can help them better understand and receive information; therefore, setting multiple playback speeds for clothing ads can play an important role in enhancing user satisfaction, improving advertising effectiveness and strengthening user engagement.

[0142] The execution of increasing the display strategy includes:

[0143] Extract the image of each frame of the labeled video;

[0144] The extracted images are grouped into a frame set in chronological order;

[0145] Setting a plurality of selection numbers for grouping pictures in a frame set;

[0146] The multiple selected quantities are recorded in ascending order as the first selected quantity, the second selected quantity, ... the rth selected quantity;

[0147] Get the number of selections in order from small to large;

[0148] For the obtained number of selections, the obtained number of selections is recorded as the number of markings;

[0149] S1, dividing the frame image set into groups of the number of marks;

[0150] S2, obtaining the last image in each group, recorded as the analog image of each group;

[0151] S3, obtaining any two adjacent analog images and performing similarity comparison;

[0152] S4, setting a threshold similarity for determining a set of images;

[0153] S5. When the similarity between any two adjacent analog images is less than the threshold similarity, all analog images form a picture set;

[0154] S6. When there are two adjacent analog images whose similarity is greater than or equal to the threshold similarity, the next selected number of the number of marks is obtained as the new number of marks, and S1-S6 are repeated.

[0155] By extracting the image of each frame of the marked video, a frame image set is obtained, but the frame image set contains multiple similar images; by setting multiple selection quantities, and selecting each selection quantity in order from small to large, the selected selection quantity is recorded as the marking quantity, and the frame image set is divided into the marking number of groups, and the last image is obtained from each group to obtain the analog image of each group, and the similarity of adjacent analog images is compared. When the similarity of adjacent analog images is less than the threshold similarity, it is considered that each analog image shows an image of a different orientation of the marked clothing, that is, each analog image is different, and all analog images are combined into a picture set; when there are adjacent analog images with a similarity greater than the threshold similarity, it is considered that at least two analog images are the same. At this time, the marking quantity is updated until each analog image is different. Different analog images are selected by threshold similarity to form a picture set, which reduces the number of pictures viewed by the user while displaying pictures of different orientations of the marked clothing, so that the user can observe carefully.

[0156] The execution speed division strategy includes:

[0157] Get the partition interval;

[0158] Calculate 0 + division interval, and record the result as the first boundary speed;

[0159] Calculate the first boundary speed + the division interval, and record the result as the second boundary speed;

[0160] Calculate the second boundary speed + the division interval, and record the result as the third boundary speed;

[0161] …

[0162] Calculate the t-1th boundary speed + division interval, and record the result as the tth boundary speed, where the value of t is infinite;

[0163] Then, the interval formed by the first boundary speed and the second boundary speed is recorded as the first speed interval;

[0164] The interval formed by the second boundary speed and the third boundary speed is recorded as the second speed interval;

[0165] …

[0166] The section formed by the t-1th boundary speed and the tth boundary speed is recorded as the t+1th speed section.

[0167] The execution speed setting strategy includes:

[0168] When a user has previously watched a clothing advertisement of a clothing merchant and then purchased clothing, the user is recorded as the first user;

[0169] The clothing advertisement where the purchased clothing appeared is recorded as a historical clothing advertisement;

[0170] For each historical clothing ad:

[0171] Get the total number of clothing displayed in historical clothing advertisements, and record it as the total display amount;

[0172] Obtain the number of clothing items in the historical clothing advertisements purchased by the user, and record it as the total purchase amount;

[0173] Calculate total purchase amount / total display amount, and record the result as purchase ratio.

[0174] The execution speed setting strategy includes:

[0175] For each first user:

[0176] Obtain the average playback speed of all historical clothing advertisements purchased by the first user, and obtain multiple historical speeds;

[0177] Classify the historical speed into corresponding speed intervals;

[0178] Obtain the purchase ratios corresponding to all historical speed multiples in the target interval and record them as the first purchase ratio, the second purchase ratio, ... the uth purchase ratio;

[0179] Execute the following formula to calculate the first average purchase ratio of the target interval;

[0180] (first purchase ratio + second purchase ratio + ... + uth purchase ratio) / u = first average purchase ratio;

[0181] Then, the first average purchase ratio in the target interval is used as the predicted probability of the first user.

[0182] A clothing advertisement displays multiple pre-sale clothing. The first user may purchase multiple clothing after viewing the clothing advertisement. Therefore, the average playback speed of the first user's historical clothing advertisements is obtained, and the purchase ratio of the historical clothing advertisements is calculated. In a historical clothing advertisement, an average playback speed corresponds to a purchase ratio. All historical speeds are divided into speed intervals, all historical speeds in the target interval are obtained, the purchase ratio of each historical speed is calculated, the average is calculated, and the first average purchase ratio of the target interval is obtained. The first average purchase ratio is the probability that the first user purchases each pre-sale clothing. The predicted probability of each pre-sale clothing is obtained. The first average purchase ratio of the target interval is calculated through the correspondence between the average playback speed and the purchase ratio of the historical clothing advertisements. When the first user is watching the clothing advertisement of the pre-sale clothing, the playback speed is in the target interval, and the probability that the first user purchases each pre-sale clothing is the first average purchase ratio, thereby predicting the sales of the pre-sale clothing, which is convenient for clothing merchants to allocate goods.

[0183] The execution speed setting strategy includes:

[0184] When the user has not purchased clothing by viewing clothing advertisements of clothing merchants before, the user is recorded as a second user;

[0185] Obtaining a target interval for a second user;

[0186] The first average purchase ratios corresponding to all first users in the target interval are obtained, the mean is calculated, and the result is recorded as the second average purchase ratio, which is used as the predicted probability of the second user.

[0187] When the second user has not purchased clothing by watching clothing advertisements of clothing merchants before, the first average purchase ratio of all first users in the target interval is obtained, the mean is calculated, and the predicted probability of the second user is obtained, that is, the probability of the second user purchasing each pre-sale clothing is predicted based on the experience value, so that the data has a basis and the prediction is more reasonable.

[0188] The execution of supporting display strategies includes:

[0189] For each user:

[0190] Get the predicted probability of the user;

[0191] Set the threshold probability for determining whether to recommend a matching video to a user;

[0192] When the predicted probability ≧ threshold probability;

[0193] Connect the supporting videos of each pre-sale clothing from beginning to end to obtain a first supporting advertisement, and recommend the first supporting advertisement to the user;

[0194] When the predicted probability < threshold probability;

[0195] Acquire the matching videos of the pre-sale clothing in the picture collection viewed by the user, connect the matching videos end to end to obtain a second matching advertisement, and recommend the second matching advertisement to the user.

[0196] By obtaining the supporting clothing for each pre-sale clothing, when the predicted probability calculated after the user watches the clothing advertisement is greater than or equal to the threshold probability, it means that the user has a high probability of purchasing each pre-sale clothing. At this time, the supporting videos of each pre-sale clothing are connected end to end to obtain a first supporting advertisement, and the first supporting advertisement is pushed to the user to increase the user's satisfaction and increase the sales of the clothing merchant; when the predicted probability calculated after the user watches the clothing advertisement is less than the threshold probability, if the user opens a picture set of a pre-sale clothing while watching the clothing advertisement, it means that the user is interested in the pre-sale clothing in the opened picture set, then, the supporting videos of the pre-sale clothing in the picture set viewed by the user are obtained, connected end to end to obtain a second supporting advertisement, and the second supporting advertisement is recommended to the user. According to the user's predicted probability and the situation of opening the picture set, the supporting advertisement is personalized recommended to the user, thereby increasing the user's satisfaction and the possibility of the user purchasing the pre-sale clothing.

[0197] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0198] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A clothing design online advertising service platform based on user feedback, characterized in that: include: The clothing pre-sold by clothing merchants is recorded as pre-sold clothing; For each pre-sale garment, shoot a display video of the pre-sale garment; Connect all pre-sale clothing display videos end to end to obtain clothing advertisements; Get the slowest speed for playing clothing advertisements; Get the fastest speed for playing clothing advertisements; Execute the speed setting strategy and set several playback speeds for clothing advertisements; When users view an apparel ad: If the user pauses while watching a clothing advertisement, the displayed video at the pause point is recorded as a marked video; Execute an additional display strategy to display a picture set at the user's pause point, wherein the picture set includes multiple display pictures of pre-sale clothing; When a user opens a photo collection, the time from when the user opens the photo collection to when the user closes the photo collection is obtained and recorded as the browsing time; Get the total actual playing time of the clothing advertisement for each user when the advertisement is played, and record it as the actual viewing time; Get the duration of the clothing advertisement, recorded as the advertisement duration; Calculate the ad duration / (actual viewing duration + browsing duration), and record the result as the average playback speed; Calculate the average playback speed of clothing advertisements viewed by each user; Implement speed division strategy and define multiple speed intervals for playing clothing advertisements; Get the speed interval where the average playback speed is located, and record it as the target interval; Execute the speed setting strategy to predict the probability of users purchasing pre-sale clothing when watching clothing advertisements at the average playback speed, which is recorded as the predicted probability; Get the predicted probabilities of all users who watched the clothing ads and calculate the sum as the predicted sales of each pre-sold clothing item; Get matching outfits for each pre-sale outfit; Shoot a display video for the matching clothes of each pre-sale clothing to obtain the matching video of each pre-sale clothing; Implement a matching display strategy to show matching videos for pre-sale clothing.

2. The clothing design online advertising service platform based on user feedback according to claim 1 is characterized by: The shooting of the display video of the pre-sale clothing includes: Get any pre-sale clothing; Get the front and back of pre-sale clothing; Recording the two sides of the pre-sale garment as a first side and a second side respectively; Get the above and below of pre-sale clothing; The display video of the pre-sale clothing is shot in the order of the front, the first side, the back, the second side, the top and the bottom.

3. The clothing design online advertising service platform based on user feedback according to claim 1 is characterized by: The execution speed setting strategy includes: Setting the division interval for determining the speed of clothing advertisement playback; Calculate (fastest speed - slowest speed) / division interval = number of divisions n; Calculate the slowest speed + divide the interval, and record the result as the first speed; Calculate the slowest speed + division interval * 2, and record the result as the second speed; …… Calculate the slowest speed + divided interval * n = fastest speed; Then, the slowest speed, the first speed, the second speed ... the fastest speed are all the set playback speeds for playing the clothing advertisement.

4. The clothing design online advertising service platform based on user feedback according to claim 1 is characterized by: The execution of increasing the display strategy includes: Extract the image of each frame of the labeled video; The extracted images are grouped into a frame set in chronological order; Setting a plurality of selection numbers for grouping pictures in a frame set; The multiple selected quantities are recorded in ascending order as the first selected quantity, the second selected quantity, ... the rth selected quantity; Get the number of selections in order from small to large; For the obtained number of selections, the obtained number of selections is recorded as the number of markings; S1, dividing the frame image set into groups of the number of marks; S2, obtaining the last image in each group, recorded as the analog image of each group; S3, obtaining any two adjacent analog images and performing similarity comparison; S4, setting a threshold similarity for determining a set of images; S5. When the similarity between any two adjacent analog images is less than the threshold similarity, all analog images form a picture set; S6. When there are two adjacent analog images whose similarity is greater than or equal to the threshold similarity, the next selected number of the number of marks is obtained as the new number of marks, and S1-S6 are repeated.

5. The clothing design online advertising service platform based on user feedback according to claim 1 is characterized by: The execution speed division strategy includes: Get the partition interval; Calculate 0 + division interval, and record the result as the first boundary speed; Calculate the first boundary speed + the division interval, and record the result as the second boundary speed; Calculate the second boundary speed + the division interval, and record the result as the third boundary speed; …… Calculate the t-1th boundary speed + division interval, and record the result as the tth boundary speed, where the value of t is infinite; Then, the interval formed by the first boundary speed and the second boundary speed is recorded as the first speed interval; The interval formed by the second boundary speed and the third boundary speed is recorded as the second speed interval; …… The section formed by the t-1th boundary speed and the tth boundary speed is recorded as the t+1th speed section.

6. The online clothing design advertising service platform based on user feedback according to claim 1 is characterized by: The execution speed setting strategy includes: When a user has previously watched a clothing advertisement of a clothing merchant and then purchased clothing, the user is recorded as the first user; The clothing advertisement where the purchased clothing appeared is recorded as a historical clothing advertisement; For each historical clothing ad: Get the total number of clothing displayed in historical clothing advertisements, and record it as the total display amount; Obtain the number of clothing items in the historical clothing advertisements purchased by the user, and record it as the total purchase amount; Calculate total purchase amount / total display amount, and record the result as purchase ratio.

7. The clothing design online advertising service platform based on user feedback according to claim 6 is characterized by: The execution speed setting strategy includes: For each first user: Obtain the average playback speed of all historical clothing advertisements purchased by the first user, and obtain multiple historical speeds; Classify the historical speed into corresponding speed intervals; Obtain the purchase ratios corresponding to all historical speed multiples in the target interval and record them as the first purchase ratio, the second purchase ratio, ... the uth purchase ratio; Execute the following formula to calculate the first average purchase ratio of the target interval; (first purchase ratio + second purchase ratio + ... + uth purchase ratio) / u = first average purchase ratio; Then, the first average purchase ratio in the target interval is used as the predicted probability of the first user.

8. The clothing design online advertising service platform based on user feedback according to claim 7 is characterized by: The execution speed setting strategy includes: When the user has not purchased clothing by viewing clothing advertisements of clothing merchants before, the user is recorded as a second user; Obtaining a target interval for a second user; The first average purchase ratios corresponding to all first users in the target interval are obtained, the mean is calculated, and the result is recorded as the second average purchase ratio, which is used as the predicted probability of the second user.

9. The online clothing design advertising service platform based on user feedback according to claim 1, characterized in that: The execution of supporting display strategies includes: For each user: Get the predicted probability of the user; Set the threshold probability for determining whether to recommend a matching video to a user; When the predicted probability ≧ threshold probability; Connect the supporting videos of each pre-sale clothing from beginning to end to obtain a first supporting advertisement, and recommend the first supporting advertisement to the user; When the predicted probability < threshold probability; Acquire the matching videos of the pre-sale clothing in the picture collection viewed by the user, connect the matching videos end to end to obtain a second matching advertisement, and recommend the second matching advertisement to the user.

Images