Personalized marketing recommendation engine data analysis method and platform

Abstract

The present invention provides a personalized marketing recommendation engine data analysis method and platform, relating to the field of data engine technology. The method includes acquiring target marketing data and target demand search data for different marketing targets, performing data classification analysis based on demand changes, and generating target cycle marketing data corresponding to the marketing targets; acquiring platform product information data, and combining the target cycle marketing data with demand change analysis based on marketing product characteristics to generate marketing adjustment data for the marketing targets; and performing product information recommendation analysis based on the marketing adjustment data to generate target marketing product recommendation data. This method uses reasonable demand data analysis to more accurately identify demand and achieve more accurate marketing information recommendations.

Description

Technical Field

[0001] The present invention relates to the field of data engine technology, and in particular to a personalized marketing recommendation engine data analysis method and platform. Background Art

[0002] Data engines are core components of modern data infrastructure, responsible for efficiently storing, processing, managing, and analyzing large amounts of data. E-commerce marketing is a form of online marketing that leverages the internet to complete a series of marketing steps and achieve marketing goals. With the development of e-commerce marketing, a marketing content recommendation method has emerged that utilizes data engines to analyze the needs of marketing targets, improving the accuracy of product recommendations to marketing targets.

[0003] However, due to the large variability in the needs of marketing targets, using data engines to conduct more accurate data analysis to further improve the accuracy of marketing information recommendations is becoming an increasingly difficult problem plaguing e-commerce marketing.

[0004] Therefore, designing a personalized marketing recommendation engine data analysis method and platform to conduct more accurate demand positioning through reasonable demand data analysis to achieve more accurate marketing information recommendation effects is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of the present invention is to provide a personalized marketing recommendation engine data analysis method, which obtains historical data of different marketing objects to cluster data according to changes in object demand, so as to better utilize historical data to mine demand information, and at the same time uses clustered data and platform product data to analyze demand changes, and then adjusts the current marketing recommendation information to form reasonable adjustment data based on changes in object demand, and then combines the demand situation to perform recommendation analysis on the adjustment data according to priority, to form recommended product information with clear priority, effectively ensuring the accuracy of recommended product information, fully ensuring the satisfaction of recommended object needs, further improving the accuracy of marketing information recommendations, and making marketing information recommendations more effective and accurate.

[0006] The purpose of the present invention is also to provide a personalized marketing recommendation engine data analysis platform, which collects historical data and platform data through a data collection unit, and combines with a feature analysis unit to perform in-depth feature extraction and analysis of marketing products, and grasps the needs of the recommendation objects from more in-depth feature information. On this basis, the recommendation adjustment unit is used to adjust the recommendation information in combination with the current needs of the recommendation objects. Different functional units are closely integrated to form an information interactive whole, which is an important material basis for realizing recommendation engine analysis.

[0007] In a first aspect, the present invention provides a personalized marketing recommendation engine data analysis method, including: obtaining object marketing data and object demand search data of different marketing objects, performing data classification analysis based on demand changes, and forming object cycle marketing data corresponding to the marketing object; obtaining platform product information data, and combining the object cycle marketing data, performing demand change analysis based on marketing product characteristics, and forming marketing adjustment data for the marketing object; performing product information recommendation analysis based on the marketing adjustment data, and forming object marketing product recommendation data.

[0008] In the present invention, the method obtains historical data of different marketing objects to cluster data according to changes in object demand, so as to better utilize historical data to mine demand information, and at the same time uses clustered data and platform product data to analyze demand changes, and then adjusts the current marketing recommendation information to form reasonable adjustment data based on changes in object demand, and then combines the demand situation to perform recommendation analysis on the adjustment data according to priority, to form recommended product information with clear priorities, effectively ensuring the accuracy of recommended product information, fully ensuring the satisfaction of recommended object needs, further improving the accuracy of marketing information recommendations, and making marketing information recommendations more effective and accurate.

[0009] As a possible implementation method, object marketing data of different marketing objects are obtained, and data classification analysis based on demand changes is performed to form object cycle marketing data corresponding to the marketing objects, including: obtaining all corresponding recommended products for the object marketing data of different marketing objects, and extracting feature labels of product information to form object product feature label groups corresponding to different recommended products; extracting product feature labels for the object product feature label groups corresponding to all recommended products based on the object demand search data of different marketing objects, and performing demand change analysis in the following manner: setting the demand change analysis time, and using the demand change analysis time as the clustering unit, clustering the extracted search labels within the entire data collection cycle of the object demand search data. The search product feature tags are classified to form a set of demand product feature tags corresponding to multiple demand change analysis durations; for the number of search product feature tags in different demand product feature tag sets, if the reduction rate of the number of search product feature tags in the earliest demand product feature tag set in the time dimension reaches the demand quantity change rate, the demand change analysis duration corresponding to the demand product feature tag set that meets the conditions is determined as the data classification time point at the time point defined in the data collection cycle: according to the data classification time point, the object marketing data collected before the data classification time point is recorded as historical object marketing data; according to the data classification time point, the object marketing data collected before the data classification time point is recorded as current object marketing data.

[0010] In the present invention, the recommended products are feature-labeled, the main purpose of which is to obtain feature information about the description of the product, and then provide a lower-level data reference for subsequent analysis based on the feature information to determine the same type or guide the acquisition of new recommended product information. For feature labeling, since e-commerce marketing data is mainly text, video, photo and other types of data, feature extraction can be performed on different types of data respectively. For example, for text, semantic-based feature extraction can be performed to obtain feature information including product appearance, performance, efficacy, etc.; for photos, comparison extraction can be performed between multiple photos of the product to obtain repetitive detailed picture information that is different from other products; for video, voice information can be obtained and converted into text for semantic-based feature extraction. At the same time, the video picture can be frame-extracted and the feature information of the picture can be obtained in the form of photos, and then the text features and picture features are combined to form feature data corresponding to the video type information. Of course, in order to unify the form of data, the feature labels corresponding to the products are formed in the form of a combination of pictures and text. It should be understood that for personalized marketing recommendations, it is necessary to accurately grasp the needs of the current object and provide product information that is more targeted and meets the needs of the object. Therefore, for the recommendation data that has already occurred for the target, it is necessary to fully consider the adjustments in the target's demand direction in the recent period. Therefore, it is necessary to classify the marketing data. The purpose of this classification is to more accurately determine whether the marketing target has experienced a demand change that is different from the past in the current period. This demand change is often reflected in the marketing target's search information. In order for the platform to recommend product information that better meets the marketing target's current needs, it is necessary to determine whether the recommendation data it has already generated meets the current target's needs. Therefore, reasonable data classification is necessary, so that the demand data that has already occurred can be accurately located and analyzed. The demand change analysis duration can be set in conjunction with the overall duration span of the data collection cycle to ensure the rationality of the data collection cycle segmentation. Of course, this segmentation should start from the most recent time. If there is any excess or insufficient duration in the final division, it can be attributed to the adjacent demand change analysis duration.

[0011] As a possible implementation method, platform product information data is obtained, and combined with the object periodic marketing data, a demand change analysis based on the marketing product characteristics is conducted to form marketing adjustment data for the marketing object, including: obtaining platform product information data, and conducting a demand change analysis based on the product feature tags on the historical object marketing data of the marketing object to form historical marketing adjustment data corresponding to the marketing object; obtaining platform product information data, and conducting a demand change analysis based on the product feature tag combination on the current object marketing data in combination with the historical object marketing data corresponding to the marketing object to form current marketing feature data corresponding to the marketing object.

[0012] In the present invention, the data of different periodic segments formed by data clustering are adjusted and analyzed for demand changes, which mainly includes two aspects. The first aspect is to split the feature label information corresponding to the product, and analyze the demand changes based on the attention of a single label, so as to accurately grasp the needs of the recommended objects to adjust the recommended marketing information. The second aspect is to consider that the demand conditions of the two periodic segments are different, so it is necessary to adopt a reasonable demand change method accordingly. For the historical periodic segments, the demand analysis is conducted through the attention of a single label, and for the current period, the demand change analysis is conducted in the form of a label group.

[0013] As a possible implementation method, platform product information data is obtained, and demand change analysis based on product feature tags is performed on historical object marketing data of marketing objects to form historical marketing adjustment data corresponding to the marketing objects, including: clustering analysis of attention time of product feature tags of recommended products on historical object marketing data of different marketing objects to form feature tag attention time clustering data corresponding to historical object marketing data; demand change trend analysis of different object product feature tags is performed based on feature tag attention time clustering data of different marketing objects to form object product feature tag trend change analysis data; product recommendation adjustment analysis is performed based on object product feature tag trend change analysis data in combination with platform product information data to form historical marketing adjustment data.

[0014] In the present invention, the demand change analysis of the marketing data of historical objects, since the marketing data is relatively old data, cannot accurately reflect the current demand, especially the demand that is different from the past, but the demand for continued attention to the recommended object can be accurately determined. Therefore, the main purpose of the demand change analysis of the marketing data of historical objects is to determine the continued demand of the recommended object. It is understandable that for e-commerce marketing, the attention of the recommended object to the demand product can be determined by the length of time the recommended object pays attention to the electronic data. For example, if a product is in demand by the recommended object, the recommended object will increase the time spent on the video, text, and picture of the product, such as continuously watching the video, photo, and text corresponding to the product. Therefore, it is reasonable to use the attention time as a reference data standard. Of course, the length of time of attention will also change after the demand changes. Since the recommended information is not a single product, the length of time of attention to a certain product is discrete. It is reasonable and accurate to use the cumulative time as the basis for trend change analysis to analyze data.

[0015] As a possible implementation method, a cluster analysis of the product feature label attention time of recommended products is performed on the historical object marketing data of different marketing objects to form feature label attention time cluster data corresponding to the historical object marketing data, including: determining the cumulative object attention time of all recommended products involved in the historical object marketing data in the time dimension sequence for the historical object marketing data of different marketing objects; assigning the cumulative object attention time to all object product feature labels in the object product feature label group corresponding to each recommended product to form a cumulative change curve of product object feature label attention time corresponding to the object product feature label; for the product object feature label attention time cumulative change curve of all object product feature labels corresponding to all recommended products in the historical object marketing data of the marketing object, the product object feature label attention time cumulative change curves corresponding to the same object product feature label are superimposed on the time dimension to form a cumulative change curve of object feature label attention time corresponding to the object product feature label; and the object feature label attention time cumulative change curves corresponding to all object product feature labels under the collective marketing object form feature label attention time cluster data.

[0016] In the present invention, it can be understood that different products may have the same feature labels. The clustering of attention duration here is mainly to merge the same feature labels and the corresponding attention duration. Of course, since it is a demand change analysis, this attention duration also needs to be reflected in the time dimension to form the attention duration change trend corresponding to the feature label. It should be noted that compared with the product itself, the feature label corresponding to the product is essentially the point of attention of the recommendation object, because the demand for the recommendation object may be discrete. For example, shampoo is needed, but which type of shampoo is not determined. Instead, discrete demand information such as suitable for oily hair, anti-hair loss, anti-dandruff, etc. is considered. Therefore, extracting these features to form labels for demand trend analysis can more accurately grasp the continuous demand of the recommendation object.

[0017] As a possible implementation method, based on the clustering data of the attention time of feature tags of different marketing objects, a demand change trend analysis of different object product feature tags is performed to form object product feature tag trend change analysis data, including: performing the following trend change analysis on the object feature tag attention time cumulative change curve corresponding to different object product feature tags in the feature tag attention time clustering data: for any object feature tag attention time cumulative change curve, if the average change rate belongs to the stable change rate range, the corresponding object product feature tag is calibrated as a stable feature; for any object feature tag attention time cumulative change curve, if the average change rate does not belong to the stable change rate range and is greater than the maximum value of the stable change rate range, the corresponding object product feature tag is calibrated as an enhanced feature; for any object feature tag attention time cumulative change curve, if the average change rate does not belong to the stable change rate range and is less than the minimum value of the stable change rate range, the corresponding object product feature tag is calibrated as a weakened feature; complete the calibration of all object product feature tags, and collect all object product feature tags to form object product feature tag trend change analysis data.

[0018] In the present invention, as time goes by, the demand for recommended objects changes, which is essentially a decrease in attention to feature tags. This decrease in attention can be reflected by the change in the cumulative duration of the tag. This application mainly classifies the changing trends of feature tags into three categories: one is a feature tag with continuous and stable attention, one is a feature tag with gradually decreasing attention, and the third is a feature tag with gradually increasing attention. The corresponding judgment method is to determine the average change rate of the curve. The stable change rate range can be set according to actual conditions, or it can be determined based on big data analysis.

[0019] As a possible implementation method, based on the object product feature label trend change analysis data, product recommendation adjustment analysis is conducted in combination with the platform product information data to form historical marketing adjustment data, including: based on the platform product information, feature labeling is extracted for the product information corresponding to all products on the platform to form platform product feature label groups corresponding to different products; based on the object product feature label trend change analysis data, all object product feature labels that have undergone stable feature calibration and all object product feature labels that have undergone enhanced feature calibration are obtained to form a historical demand continuous feature label set; based on the platform product feature label groups corresponding to different products and the historical demand continuous feature label set, product recommendation adjustment analysis is conducted to form historical marketing adjustment data.

[0020] In this invention, after completing trend analysis, we can determine the characteristic tags that the recommended target continues to follow or is increasing their attention. The products associated with these tags are very likely to be products that the recommended target needs. Of course, for e-commerce platforms, the products they provide are determined by their supply chain, so the range of product selection is also determined by the products the platform can provide. By labeling these products and combining them with the characteristic tags of the recommended target's continued attention and increased attention under historical data, we can determine the product information that continues to be recommended based on historical data, thereby completing the adjustment of historical marketing.

[0021] As a possible implementation method, product recommendation adjustment analysis is performed based on the platform product feature label group and the historical demand continuous feature label set corresponding to different products to form historical marketing adjustment data, including: for the historical demand continuous feature label set, if there is a platform product feature label group that satisfies that all product feature labels in the platform product feature label group belong to the historical demand continuous feature label set, then the product corresponding to the platform product feature label group is determined as a historical demand continuous type 1 product; for the historical demand continuous feature label set, if there is a platform product feature label group that satisfies that the number of product feature labels in the platform product feature label group that belong to the historical demand continuous feature label set is greater than the number of product feature labels in the platform product feature label group. If the quantity ratio is not less than the high demand ratio threshold, the product corresponding to the platform product feature label group is determined as a historical demand continuous category II product; for the historical demand continuous feature label set, if there is a platform product feature label group that satisfies the requirement that there are product feature labels belonging to the historical demand continuous feature label set in the platform product feature label group, and the number of product feature labels belonging to the historical demand continuous feature label set accounts for less than the high demand ratio threshold in the platform product feature label group, then the product corresponding to the platform product feature label group is determined as a historical demand continuous category III product; all historical demand continuous category I products, historical demand continuous category II products and historical demand continuous category III products are collected to form historical marketing adjustment data.

[0022] In the present invention, it can be understood that there may be many products involved in a single feature tag. The more feature tags involved in the product that the recommended object continues to pay attention to and enhances its attention to, the more likely the product is to be wanted by the recommended object. Therefore, the classification of products adjusted for historical marketing is achieved based on the number of feature tags that the recommended object continues to pay attention to and enhances its attention to. The high demand proportion threshold can be set according to actual conditions, and can also be determined based on big data analysis.

[0023] As a possible implementation method, platform product information data is obtained, and the current object marketing data is analyzed based on the product feature label combination in combination with the historical object marketing data corresponding to the marketing object to form the current marketing feature data corresponding to the marketing object, including: for different marketing objects, obtaining the object product feature label group corresponding to different recommended products in the current object marketing data, and combining the object product feature label group corresponding to different recommended products in the corresponding historical object marketing data to perform feature comparison analysis to form feature comparison analysis result data; based on the feature comparison analysis result data, combined with the platform product information data, product demand change analysis is performed to form the current marketing feature data corresponding to the marketing object.

[0024] In the present invention, the current target marketing data may contain the latest demand information. For example, if the recommended target has new needs, search information will be provided to the platform. This search information will index some products. However, due to the lack of thorough analysis, especially feature tag analysis, the products may be biased. Therefore, the analysis of the current target marketing data mainly involves extracting the feature tags of the new needs and then filtering the most matching products on the platform for recommendation. Of course, because the feature tags of the newly recommended related products in the marketing data are representative, analyzing a single feature tag will lose the ability to locate the needs. Therefore, analyzing in the form of feature tag groups can more accurately determine the new needs of the recommended target.

[0025] As a possible implementation method, for different marketing objects, the object product feature label groups corresponding to different recommended products in the current object marketing data are obtained, and combined with the object product feature label groups corresponding to different recommended products in the corresponding historical object marketing data, feature comparison analysis is performed to form feature comparison analysis result data, including: performing the following label group overlap comparison analysis on the object product feature label groups corresponding to different recommended products in the current object marketing data and the object product feature label groups corresponding to different recommended products in the corresponding historical object marketing data: for the object product feature label group corresponding to any recommended product in the current object marketing data, if the number of object product feature labels that overlap with the object product feature label group corresponding to any recommended product in the historical object marketing data exceeds the similarity ratio threshold in the object product feature label group corresponding to the recommended product in the current object marketing data, then the object product feature label group corresponding to the recommended product in the current object marketing data is determined as the new demand feature label group; the object product feature label groups corresponding to all recommended products in the current object marketing data are traversed, and all new demand feature label groups are collected to form feature comparison analysis result data.

[0026] In the present invention, the first step in determining new demands is to identify the feature tags corresponding to the new demands from the marketing data. It is understandable that since the products provided by the platform based on the search information of the recommended objects may not be accurate, there are also unnecessary feature tags in the feature tag group corresponding to the products. In order to avoid overly strict screening of the feature tag group, the tag groups with a certain degree of feature tag overlap need to be determined as new demand feature tags. The similarity ratio threshold can be set according to the actual situation or determined based on big data analysis.

[0027] As a possible implementation method, based on the feature comparison analysis result data, combined with the platform product information data, product demand change analysis is performed to form current marketing feature data corresponding to the marketing object, including: for different new demand feature label groups in the feature comparison analysis result data, combined with the platform product feature label groups of different products in the platform product information data, the following demand change analysis is performed: for the new demand feature label group, if there is a platform product feature label group that contains all the object product feature labels in the new demand feature label group, then the product corresponding to the platform product feature label group is determined as a current demand type 1 product; for the new demand feature label group, if there is some object product feature labels in the platform product feature label group that match some in the new demand feature label group If the object product feature labels are the same, and the proportion of the number of identical object product feature labels in the new demand feature label group is not less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as the current demand category 2 product; for the new demand feature label group, if some object product feature labels in the platform product feature label group are the same as some object product feature labels in the new demand feature label group, and the proportion of the number of identical object product feature labels in the new demand feature label group is less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as the current demand category 3 product; all current demand category 1 products, current demand category 2 products and current demand category 3 products are collected to form the current marketing feature data.

[0028] In the present invention, similarly, for a new demand feature tag group, the feature tags may all or partly represent the feature information required by the recommended object. Therefore, when performing product positioning, the demand for products is classified according to the percentage of overlap. This fully ensures the accuracy of the products recommended and adjusted, and also provides a reference for the priority of subsequent product information recommendations. The percentage of changing demand can be determined based on actual conditions or based on big data analysis.

[0029] As a possible implementation method, product information recommendation analysis is performed based on marketing adjustment data to form object marketing product recommendation data, including: based on historical marketing adjustment data, determining historical demand for continuous category one products and current demand category one products as category one recommended products, historical demand for continuous category two products and current demand category two products as category two recommended products, and historical demand for continuous category three products and current demand category three products as category three recommended products; allocating recommended traffic for category one recommended products, category two recommended products, and category three recommended products, and ensuring that the allocated recommended traffic gradually decreases as the category increases; obtaining the recommended traffic allocated to recommended products of different categories to form object marketing product recommendation data corresponding to the marketing object.

[0030] In the present invention, after obtaining corresponding adjustment results using historical and current object marketing data, the method of recommendation also needs to be considered. This is because the demand for recommended objects varies greatly among different product categories. Considering that e-commerce marketing is mostly carried out in the form of traffic push, the recommended traffic can be distributed according to product category, ensuring that category 1 products have the highest demand, category 2 the second, and category 3 the least. Of course, for different product information of the same type, the recommended traffic can be divided equally or allocated according to demand.

[0031] In a second aspect, the present invention provides a personalized marketing recommendation engine data analysis platform, including: a data acquisition unit, used to obtain platform product information and object marketing data and object demand search data of different marketing objects; a feature analysis unit, used to perform data classification analysis based on demand changes according to the object marketing data and object demand search data of different marketing objects acquired by the data acquisition unit, to form object cycle marketing data corresponding to the marketing object; a recommendation adjustment unit, used to perform demand change analysis based on marketing product characteristics according to the object cycle marketing data formed by the feature analysis unit in combination with the platform product information data acquired by the data acquisition unit, to form marketing adjustment data of the marketing object, and to perform product information recommendation analysis to form object marketing product recommendation data.

[0032] In the present invention, the platform collects historical data and platform data through the data collection unit, and combines with the feature analysis unit to conduct in-depth feature extraction and analysis of marketing products, grasp the needs of the recommendation objects from more in-depth feature information, and on this basis, uses the recommendation adjustment unit to adjust the recommendation information in combination with the current needs of the recommendation objects. Different functional units are closely integrated to form an information interactive whole, which is an important material basis for realizing recommendation engine analysis.

[0033] The personalized marketing recommendation engine data analysis method and platform provided by the present invention have the following beneficial effects:

[0034] This method obtains historical data of different marketing objects to cluster data according to changes in object demand, so as to better utilize historical data to mine demand information. At the same time, it uses clustered data and platform product data to analyze demand changes, and then adjusts the current marketing recommendation information to form reasonable adjustment data based on changes in object demand. The adjusted data is then analyzed according to the priority of the recommendation based on the demand situation to form recommended product information with clear priorities, which effectively guarantees the accuracy of the recommended product information, fully guarantees the satisfaction of the recommended object's needs, further improves the accuracy of marketing information recommendations, and makes marketing information recommendations more effective and accurate.

[0035] The platform collects historical data and platform data through the data collection unit, and combines it with the feature analysis unit to conduct in-depth feature extraction and analysis of marketing products, grasp the needs of the recommendation objects from more in-depth feature information, and then uses the recommendation adjustment unit to adjust the recommendation information based on the current needs of the recommendation objects. Different functional units are closely integrated to form an information interactive whole, which is an important material basis for realizing recommendation engine analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 A diagram showing the steps of a personalized marketing recommendation engine data analysis method provided by an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of the structure of a personalized marketing recommendation engine data analysis platform provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.

[0040] Data engines are core components of modern data infrastructure, responsible for efficiently storing, processing, managing, and analyzing large amounts of data. E-commerce marketing is a form of online marketing that leverages the internet to complete a series of marketing steps and achieve marketing goals. With the development of e-commerce marketing, a marketing content recommendation method has emerged that utilizes data engines to analyze the needs of marketing targets, improving the accuracy of product recommendations to marketing targets.

[0041] However, due to the large variability in the needs of marketing targets, using data engines to conduct more accurate data analysis to further improve the accuracy of marketing information recommendations is becoming an increasingly difficult problem plaguing e-commerce marketing.

[0042] refer to Figure 1-Figure 2 An embodiment of the present invention provides a data analysis method for a personalized marketing recommendation engine. The method obtains historical data of different marketing objects to cluster data according to changes in object demand, so as to better utilize historical data to mine demand information. At the same time, the method uses the clustered data and platform product data to analyze demand changes, and then adjusts the current marketing recommendation information to form reasonable adjustment data based on changes in object demand. The adjustment data is then analyzed according to the priority of the recommendation based on the demand situation to form recommended product information with clear priorities, which effectively ensures the accuracy of the recommended product information, fully guarantees the satisfaction of the recommended object's needs, further improves the accuracy of marketing information recommendations, and makes marketing information recommendations more effective and accurate.

[0043] The data analysis method of the personalized marketing recommendation engine specifically includes the following steps:

[0044] S1: Obtain object marketing data and object demand search data of different marketing objects, conduct data classification analysis based on demand changes, and form object cycle marketing data corresponding to the marketing objects.

[0045] Obtain object marketing data of different marketing objects, conduct data classification analysis based on demand changes, and form object cycle marketing data corresponding to the marketing objects, including: obtaining all corresponding recommended products for the object marketing data of different marketing objects, and extracting feature labels of product information to form object product feature label groups corresponding to different recommended products; extracting product feature labels for the object product feature label groups corresponding to all recommended products based on the object demand search data of different marketing objects, and conducting demand change analysis in the following way: setting the demand change analysis time, and using the demand change analysis time as the clustering unit, clustering the extracted search product feature labels within the entire data collection cycle of the object demand search data. The tags are classified to form multiple demand change analysis time corresponding to the demand product feature tag sets; the number of search product feature tags in different demand product feature tag sets, when the reduction rate of the number of search product feature tags in the earliest demand product feature tag set in the time dimension reaches the demand quantity change rate, the demand change analysis time corresponding to the demand product feature tag set that meets the conditions is determined as the data classification time point at the time point defined in the data collection cycle: according to the data classification time point, the object marketing data collected before the data classification time point is recorded as the historical object marketing data; according to the data classification time point, the object marketing data collected before the data classification time point is recorded as the current object marketing data.

[0046] The main purpose of feature labeling recommended products is to obtain characteristic information describing the products, thereby providing a lower-level data reference for subsequent feature-based analysis to identify similar products or guide the acquisition of new recommended product information. Regarding feature labeling, since e-commerce marketing data primarily consists of text, video, and photos, feature extraction can be performed on different types of data. For example, for text, semantic-based feature extraction can be performed to obtain characteristic information such as product appearance, performance, and efficacy. For photos, multiple product photos can be compared and extracted to obtain repetitive image information with details that distinguish the product from other products. For videos, voice information can be converted into text and then semantic-based feature extraction can be performed. Furthermore, video frames can be extracted and image feature information can be obtained similarly to photos. The text features and image features can then be combined to form feature data corresponding to the video type information. Of course, to unify the data format, the corresponding feature labels for products are formed using a combination of images and text. It is important to understand that personalized marketing recommendations are all about accurately understanding the needs of the current target audience and providing more targeted product information that meets their needs. Therefore, for the recommendation data that has already occurred for the target, it is necessary to fully consider the adjustments in the target's demand direction in the recent period. Therefore, it is necessary to classify the marketing data. The purpose of this classification is to more accurately determine whether the marketing target has experienced a demand change that is different from the past in the current period. This demand change is often reflected in the marketing target's search information. In order for the platform to recommend product information that better meets the marketing target's current needs, it is necessary to determine whether the recommendation data it has already generated meets the current target's needs. Therefore, reasonable data classification is necessary, so that the demand data that has already occurred can be accurately located and analyzed. The demand change analysis duration can be set in conjunction with the overall duration span of the data collection cycle to ensure the rationality of the data collection cycle segmentation. Of course, this segmentation should start from the most recent time. If there is any excess or insufficient duration in the final division, it can be attributed to the adjacent demand change analysis duration.

[0047] S2: Obtain platform product information data, and combine it with the target period marketing data to conduct demand change analysis based on the marketing product characteristics to form marketing adjustment data for the marketing target.

[0048] Obtain platform product information data, and combine it with the object's periodic marketing data to conduct a demand change analysis based on the marketing product characteristics to form marketing adjustment data for the marketing object, including: obtaining platform product information data, and conducting a demand change analysis based on product feature tags on the historical object marketing data of the marketing object to form historical marketing adjustment data corresponding to the marketing object; obtaining platform product information data, and combining it with the historical object marketing data corresponding to the marketing object to conduct a demand change analysis based on the product feature tag combination on the current object marketing data to form current marketing feature data corresponding to the marketing object.

[0049] The data of different cycle segments formed by data clustering are adjusted and analyzed based on demand changes, which mainly includes two aspects. The first aspect is to split the feature label information corresponding to the product, and analyze the demand changes based on the attention of a single label, so as to accurately grasp the needs of the recommended objects to adjust the recommended marketing information. The second aspect is to consider that the demand conditions of the two cycle segments are different, so it is necessary to adopt reasonable demand change methods accordingly. For the historical cycle segments, the demand analysis is conducted through the attention of a single label, while for the current cycle segment, the demand change analysis is conducted in the form of a label group.

[0050] Obtain platform product information data, and conduct demand change analysis based on product feature tags on historical object marketing data of marketing objects to form historical marketing adjustment data corresponding to the marketing objects, including: conducting cluster analysis on product feature tag attention time for recommended products on historical object marketing data of different marketing objects to form feature tag attention time cluster data corresponding to historical object marketing data; conducting demand change trend analysis on product feature tags of different objects based on feature tag attention time cluster data of different marketing objects to form object product feature tag trend change analysis data; conducting product recommendation adjustment analysis based on object product feature tag trend change analysis data in combination with platform product information data to form historical marketing adjustment data.

[0051] The demand change analysis of historical marketing data of objects, since marketing data is relatively old data, cannot accurately reflect the current changes in demand, especially those that are different from the past. However, the demand for continued attention to the recommended objects can be accurately determined. Therefore, the main purpose of the demand change analysis of historical marketing data of objects is to determine the continued demand of the recommended objects. It is understandable that for e-commerce marketing, the attention of the recommended objects to the demand products can be determined by the length of time they pay attention to the electronic data. For example, if a product is in demand by the recommended objects, then the recommended objects will increase the time they stay on the product's videos, texts, and pictures, such as continuously watching the corresponding videos, photos, and texts of the product. Therefore, it is reasonable to use attention time as a reference data standard. Of course, the length of attention will also change after the demand changes. Since the recommended information is not a single product, the length of attention to a certain product is discrete. It is reasonable and accurate to use the cumulative time as the basis for trend change analysis to analyze data.

[0052] For the historical object marketing data of different marketing objects, a cluster analysis of the product feature label attention time of recommended products is performed to form feature label attention time cluster data corresponding to the historical object marketing data, including: for the historical object marketing data of different marketing objects, determining the cumulative object attention time of all recommended products involved in the historical object marketing data in the time dimension sequence; for all object product feature labels in the object product feature label group corresponding to each recommended product, assigning the cumulative object attention time to all object product feature labels to form a cumulative change curve of product object feature label attention time corresponding to the object product feature label; for the product object feature label attention time cumulative change curve of all object product feature labels corresponding to all recommended products in the historical object marketing data of the marketing object, the product object feature label attention time cumulative change curves corresponding to the same object product feature label are superimposed on the time dimension to form a cumulative change curve of object feature label attention time corresponding to the object product feature label; and the object feature label attention time cumulative change curves corresponding to all object product feature labels under the collective marketing object are formed to form feature label attention time cluster data.

[0053] It is understandable that different products may have the same feature labels. The clustering of attention duration here is mainly to merge the same feature labels and the corresponding attention duration. Of course, since it is a demand change analysis, this attention duration also needs to be reflected in the time dimension to form the attention duration change trend corresponding to the feature label. It should be noted that compared to the product itself, the feature label corresponding to the product is essentially the focus of the recommendation object, because the demand for the recommendation object may be discrete. For example, the need for shampoo will not determine which type of shampoo, but the discrete demand information such as suitable for oily hair, anti-hair loss, anti-dandruff, etc. is considered. Therefore, extracting these features to form labels for demand trend analysis can more accurately grasp the continuous demand of the recommendation object.

[0054] Based on the clustering data of the attention time of the feature tags of different marketing objects, the demand change trend analysis of the feature tags of different object products is performed to form the object product feature tag trend change analysis data, including: performing the following trend change analysis on the cumulative change curve of the object feature tag attention time corresponding to the different object product feature tags in the feature tag attention time clustering data: for any object feature tag attention time cumulative change curve, if the average change rate falls within the stable change rate range, the corresponding object product feature tag is calibrated as a stable feature; for any object feature tag attention time cumulative change curve, if the average change rate does not fall within the stable change rate range and is greater than the maximum value of the stable change rate range, the corresponding object product feature tag is calibrated as an enhanced feature; for any object feature tag attention time cumulative change curve, if the average change rate does not fall within the stable change rate range and is less than the minimum value of the stable change rate range, the corresponding object product feature tag is calibrated as a weakened feature; complete the calibration of all object product feature tags, and collect all object product feature tags to form the object product feature tag trend change analysis data.

[0055] As time goes by, the demand for recommended objects changes, which essentially means that the attention to feature tags decreases. This decrease in attention can be reflected by the change in the cumulative duration of the tag. This application mainly classifies the changing trends of feature tags into three categories: one is the feature tags that are continuously and stably paid attention to, one is the feature tags with gradually decreasing attention, and the third is the feature tags with gradually increasing attention. The corresponding judgment method is to determine the average change rate of the curve. The stable change rate range can be set according to the actual situation, or it can be determined based on big data analysis.

[0056] Based on the data on trend changes in object product feature labels, product recommendation adjustment analysis is conducted in combination with platform product information data to form historical marketing adjustment data, including: based on the platform product information, feature labeling is extracted for the product information corresponding to all products on the platform to form platform product feature label groups corresponding to different products; based on the data on trend changes in object product feature labels, all object product feature labels that have undergone stable feature calibration and all object product feature labels that have undergone enhanced feature calibration are obtained to form a historical demand continuous feature label set; based on the platform product feature label groups corresponding to different products and the historical demand continuous feature label set, product recommendation adjustment analysis is conducted to form historical marketing adjustment data.

[0057] After completing trend analysis, we can identify the characteristic tags that the recommended target continues to follow or is increasing their attention. The products associated with these tags are likely to be the products the recommended target is looking for. Of course, for e-commerce platforms, the products they offer are determined by their supply chain, so the range of product selection is also determined by the platform's offerings. By labeling these products and combining them with the characteristic tags that the recommended target has followed over time based on historical data, we can identify products that continue to be recommended based on historical data, thereby completing historical marketing adjustments.

[0058] According to the platform product feature tag group and historical demand continuous feature tag set corresponding to different products, product recommendation adjustment analysis is performed to form historical marketing adjustment data, including: for the historical demand continuous feature tag set, if there is a platform product feature tag group that satisfies that all product feature tags in the platform product feature tag group belong to the historical demand continuous feature tag set, then the product corresponding to the platform product feature tag group is determined as a historical demand continuous type 1 product; for the historical demand continuous feature tag set, if there is a platform product feature tag group that satisfies that the number of product feature tags in the platform product feature tag group that belong to the historical demand continuous feature tag set accounts for no less than If the high demand proportion threshold is exceeded, the product corresponding to the platform product feature label group is determined as a historical demand continuous category II product; for the historical demand continuous feature label set, if there is a platform product feature label group that satisfies the requirement that there are product feature labels belonging to the historical demand continuous feature label set in the platform product feature label group, and the number of product feature labels belonging to the historical demand continuous feature label set accounts for less than the high demand proportion threshold, then the product corresponding to the platform product feature label group is determined as a historical demand continuous category III product; all historical demand continuous category I products, historical demand continuous category II products and historical demand continuous category III products are collected to form historical marketing adjustment data.

[0059] It is understandable that a single feature tag may involve many products. The more feature tags that a product involves that the recommended object continues to pay attention to and increases its attention, the more likely the product is what the recommended object wants. Therefore, the products adjusted for historical marketing are classified according to the number of feature tags that the recommended object continues to pay attention to and increases its attention. The high-demand ratio threshold can be set according to actual conditions or determined based on big data analysis.

[0060] Obtain platform product information data, and conduct a demand change analysis on the current object marketing data based on the product feature label combination in combination with the historical object marketing data corresponding to the marketing object, to form the current marketing feature data corresponding to the marketing object, including: for different marketing objects, obtain the object product feature label group corresponding to different recommended products in the current object marketing data, and conduct a feature comparison analysis in combination with the object product feature label group corresponding to different recommended products in the corresponding historical object marketing data to form feature comparison analysis result data; based on the feature comparison analysis result data, combine with the platform product information data to conduct a product demand change analysis to form the current marketing feature data corresponding to the marketing object.

[0061] The current target marketing data may contain the latest demand information. For example, if the recommended target has new needs, search information will be provided to the platform. This search information will index some products. However, due to lack of thorough analysis, especially feature tag analysis, the products may be biased. Therefore, the analysis of the current target marketing data mainly involves extracting the feature tags of the new needs and then filtering the most matching products on the platform for recommendation. Of course, because the feature tags of the newly recommended related products in the marketing data are representative, analyzing a single feature tag will lose the ability to identify the needs. Therefore, analyzing in the form of feature tag groups can more accurately determine the new needs of the recommended target.

[0062] For different marketing objects, obtain the object product feature label groups corresponding to different recommended products in the current object marketing data, and combine them with the object product feature label groups corresponding to different recommended products in the corresponding historical object marketing data to perform feature comparison analysis to form feature comparison analysis result data, including: performing the following label group overlap comparison analysis on the object product feature label groups corresponding to different recommended products in the current object marketing data and the object product feature label groups corresponding to different recommended products in the corresponding historical object marketing data: for the object product feature label group corresponding to any recommended product in the current object marketing data, if the number of object product feature labels that overlap with the object product feature label group corresponding to any recommended product in the historical object marketing data exceeds the similarity ratio threshold in the object product feature label group corresponding to the recommended product in the current object marketing data, then determine the object product feature label group corresponding to the recommended product in the current object marketing data as the new demand feature label group; traverse the object product feature label groups corresponding to all recommended products in the current object marketing data, and collect all new demand feature label groups to form feature comparison analysis result data.

[0063] The first step in identifying new needs is to identify the corresponding feature tags from marketing data. Understandably, since the products provided by the platform based on the search information of the recommended object may not be accurate, some feature tags in the corresponding feature tag group of the product are unnecessary. To avoid overly stringent screening of feature tag groups, all tag groups with a certain degree of feature tag overlap need to be identified as new demand feature tags. The similarity ratio threshold can be set based on actual conditions or determined based on big data analysis.

[0064] According to the feature comparison analysis result data, combined with the platform product information data, product demand change analysis is performed to form the current marketing feature data corresponding to the marketing object, including: for different new demand feature label groups in the feature comparison analysis result data, combined with the platform product feature label groups of different products in the platform product information data, the following demand change analysis is performed: for the new demand feature label group, if there is a platform product feature label group that contains all the object product feature labels in the new demand feature label group, then the product corresponding to the platform product feature label group is determined as a current demand type 1 product; for the new demand feature label group, if there are some object product feature labels in the platform product feature label group that match some object product feature labels in the new demand feature label group If the labels are the same, and the proportion of the number of identical object product feature labels in the new demand feature label group is not less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as the current demand category 2 product; for the new demand feature label group, if some object product feature labels in the platform product feature label group are the same as some object product feature labels in the new demand feature label group, and the proportion of the number of identical object product feature labels in the new demand feature label group is less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as the current demand category 3 product; all current demand category 1 products, current demand category 2 products and current demand category 3 products are collected to form the current marketing feature data.

[0065] Similarly, for new demand feature tag groups, the feature tags may all or only partially represent the required feature information for the recommended object. Therefore, when performing product positioning, products are classified by their demand based on the percentage of overlap. This fully ensures the accuracy of the products recommended and provides a reference for prioritizing subsequent product information recommendations. The percentage of changing demand can be determined based on actual conditions or through big data analysis.

[0066] S3: Based on the marketing adjustment data, conduct product information recommendation analysis to form target marketing product recommendation data.

[0067] Based on the marketing adjustment data, product information recommendation analysis is performed to form object marketing product recommendation data, including: based on historical marketing adjustment data, determining the historical demand for continuous category one products and the current demand for category one products as category one recommended products, determining the historical demand for continuous category two products and the current demand for category two products as category two recommended products, and determining the historical demand for continuous category three products and the current demand for category three products as category three recommended products; allocating the recommended traffic for category one recommended products, category two recommended products, and category three recommended products, and ensuring that the allocated recommended traffic gradually decreases as the category increases; obtaining the recommended traffic allocated to recommended products of different categories to form object marketing product recommendation data corresponding to the marketing object.

[0068] After utilizing both historical and current target marketing data to obtain the corresponding adjustment results, it's also important to consider how to recommend products. This is because demand for recommended products can vary significantly across product categories. Considering that e-commerce marketing is largely driven by traffic, it's possible to allocate recommended traffic by product category, ensuring that category 1 receives the highest traffic, category 2 receives the second highest, and category 3 receives the least. Of course, for different products within the same category, recommended traffic can be divided equally or allocated based on demand.

[0069] The present invention also provides a personalized marketing recommendation engine data analysis platform, which includes: a data acquisition unit, which is used to obtain platform product information and object marketing data and object demand search data of different marketing objects; a feature analysis unit, which is used to perform data classification analysis based on demand changes based on the object marketing data and object demand search data of different marketing objects acquired by the data acquisition unit, and form object cycle marketing data corresponding to the marketing object; a recommendation adjustment unit, which is used to perform demand change analysis based on marketing product characteristics based on the object cycle marketing data formed by the feature analysis unit in combination with the platform product information data acquired by the data acquisition unit, to form marketing adjustment data for the marketing object, and to perform product information recommendation analysis to form object marketing product recommendation data.

[0070] The platform collects historical data and platform data through the data collection unit, and combines it with the feature analysis unit to conduct in-depth feature extraction and analysis of marketing products, grasp the needs of the recommendation objects from more in-depth feature information, and then uses the recommendation adjustment unit to adjust the recommendation information based on the current needs of the recommendation objects. Different functional units are closely integrated to form an information interactive whole, which is an important material basis for realizing recommendation engine analysis.

[0071] In summary, the personalized marketing recommendation engine data analysis method and platform provided by the embodiments of the present invention have the following beneficial effects:

[0072] This method obtains historical data of different marketing objects to cluster data according to changes in object demand, so as to better utilize historical data to mine demand information. At the same time, it uses clustered data and platform product data to analyze demand changes, and then adjusts the current marketing recommendation information to form reasonable adjustment data based on changes in object demand. The adjusted data is then analyzed according to the priority of the recommendation based on the demand situation to form recommended product information with clear priorities, which effectively guarantees the accuracy of the recommended product information, fully guarantees the satisfaction of the recommended object's needs, further improves the accuracy of marketing information recommendations, and makes marketing information recommendations more effective and accurate.

[0073] The platform collects historical data and platform data through the data collection unit, and combines it with the feature analysis unit to conduct in-depth feature extraction and analysis of marketing products, grasp the needs of the recommendation objects from more in-depth feature information, and then uses the recommendation adjustment unit to adjust the recommendation information based on the current needs of the recommendation objects. Different functional units are closely integrated to form an information interactive whole, which is an important material basis for realizing recommendation engine analysis.

[0074] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A personalized marketing recommendation engine data analysis method, characterized in that: include: Obtain object marketing data and object demand search data for different marketing objects, perform data classification analysis based on demand changes, and form object cycle marketing data corresponding to the marketing objects; Acquire platform product information data, and combine it with the target periodic marketing data to perform demand change analysis based on marketing product characteristics to form marketing adjustment data for the marketing target; Perform product information recommendation analysis based on the marketing adjustment data to generate target marketing product recommendation data; The object marketing data and object demand search data of different marketing objects are obtained, and data classification analysis based on demand changes is performed to form object cycle marketing data corresponding to the marketing objects, including: For the object marketing data of different marketing objects, all corresponding recommended products are obtained, and feature labeling of product information is extracted to form object product feature label groups corresponding to different recommended products; Based on the object demand search data of different marketing objects, extract the product feature tags of the object product feature tag group corresponding to all the recommended products, and perform demand change analysis in the following manner: Setting a demand change analysis duration, and using the demand change analysis duration as a clustering unit, classifying the extracted search product feature tags during the entire data collection cycle of collecting the object demand search data, to form a plurality of demand product feature tag sets corresponding to the demand change analysis duration; For the number of search product feature tags in different demand product feature tag sets, if the reduction rate of the number of search product feature tags in the earliest demand product feature tag set in the time dimension reaches the demand quantity change rate, then the time point defined by the demand change analysis duration corresponding to the demand product feature tag set that meets the conditions within the data collection cycle is determined as the data classification time point: According to the data classification time point, recording the object marketing data collected before the data classification time point as historical object marketing data; According to the data classification time point, recording the object marketing data collected before the data classification time point as current object marketing data; Acquire platform product information data, and combine it with the target periodic marketing data to perform demand change analysis based on marketing product characteristics to form marketing adjustment data for the marketing target, including: Acquire the platform product information data, and perform demand change analysis based on product feature tags on the historical object marketing data of the marketing object to form historical marketing adjustment data corresponding to the marketing object; The platform product information data is obtained, and the demand change analysis based on the product feature tag combination is performed on the current object marketing data in combination with the historical object marketing data corresponding to the marketing object to form the current marketing feature data corresponding to the marketing object.

2. The personalized marketing recommendation engine data analysis method according to claim 1, characterized in that: The acquiring of the platform product information data and performing a demand change analysis based on product feature tags on the historical object marketing data of the marketing object to form historical marketing adjustment data corresponding to the marketing object include: Performing a cluster analysis of the attention duration of the product feature labels of the recommended products on the historical object marketing data of different marketing objects to form cluster data of the attention duration of the feature labels corresponding to the historical object marketing data; According to the clustering data of the attention duration of the feature tags of different marketing objects, the demand change trend analysis of the feature tags of different target products is performed to form the target product feature tag trend change analysis data; The historical marketing adjustment data is formed by performing product recommendation adjustment analysis based on the target product feature tag trend change analysis data in combination with the platform product information data.

3. The personalized marketing recommendation engine data analysis method according to claim 2, characterized in that: The performing of cluster analysis on the product feature label attention duration of the recommended product on the historical object marketing data of different marketing objects to form feature label attention duration cluster data corresponding to the historical object marketing data includes: For the historical object marketing data of different marketing objects, determining the cumulative object attention duration of all the recommended products involved in the historical object marketing data in a time dimension sequence; For all the object product feature tags in the object product feature tag group corresponding to each recommended product, assign the accumulated object attention time to all the object product feature tags to form a cumulative change curve of the product object feature tag attention time corresponding to the object product feature tag; For the cumulative change curves of the product object feature tag attention duration of all the object product feature tags corresponding to all the recommended products in the historical object marketing data of the marketing object, the cumulative change curves of the product object feature tag attention duration corresponding to the same object product feature tag are superimposed in the time dimension to form a cumulative change curve of the object feature tag attention duration corresponding to the object product feature tag; The cumulative change curves of the object feature tag attention duration corresponding to all the object product feature tags under the marketing object are collected to form the feature tag attention duration clustering data.

4. The personalized marketing recommendation engine data analysis method according to claim 3, characterized in that: The clustering data of the attention duration of the feature tags of different marketing objects is used to analyze the demand change trend of the feature tags of different target products to form the target product feature tag trend change analysis data, including: The following trend change analysis is performed on the cumulative change curve of the object feature label attention duration corresponding to the different object product feature labels in the feature label attention duration clustering data: For any of the object feature labels, pay attention to the cumulative change curve of the duration. If the average change rate falls within the stable change rate range, the corresponding object product feature label is calibrated as a stable feature. For any of the cumulative change curves of the attention duration of the object feature label, if the average change rate does not fall within the stable change rate range and is greater than the maximum value of the stable change rate range, the corresponding object product feature label is subjected to enhanced feature calibration; For any cumulative change curve of the attention duration of the object feature label, if the average change rate does not fall within the stable change rate range and is less than the minimum value of the stable change rate range, the corresponding object product feature label is subjected to weakened feature calibration; Complete the calibration of all the target product feature labels, and collect all the target product feature labels to form the target product feature label trend change analysis data.

5. The personalized marketing recommendation engine data analysis method according to claim 4, characterized in that: The product recommendation adjustment analysis based on the target product feature tag trend change analysis data and the platform product information data is combined to form the historical marketing adjustment data, including: Based on the platform product information, feature labeling is performed on the product information corresponding to all products on the platform to form platform product feature label groups corresponding to different products; According to the target product feature tag trend change analysis data, all target product feature tags that have undergone stable feature calibration and all target product feature tags that have undergone enhanced feature calibration are obtained to form a historical demand continuous feature tag set; According to the platform product feature tag group and the historical demand continuous feature tag set corresponding to different products, product recommendation adjustment analysis is performed to form the historical marketing adjustment data.

6. The personalized marketing recommendation engine data analysis method according to claim 5, characterized in that: The product recommendation adjustment analysis is performed based on the platform product feature tag group and the historical demand continuous feature tag set corresponding to different products to form the historical marketing adjustment data, including: For the historical demand persistence feature tag set, if there exists a platform product feature tag group that satisfies that all product feature tags in the platform product feature tag group belong to the historical demand persistence feature tag set, then the product corresponding to the platform product feature tag group is determined as a historical demand persistence type 1 product; For the historical demand persistence feature tag set, if there exists a platform product feature tag group that satisfies the requirement that the number of product feature tags belonging to the historical demand persistence feature tag set in the platform product feature tag group accounts for no less than a high demand ratio threshold, then the product corresponding to the platform product feature tag group is determined as a historical demand persistence category II product; For the historical demand persistence feature tag set, if there exists a platform product feature tag group that satisfies the requirement that there is a product feature tag belonging to the historical demand persistence feature tag set in the platform product feature tag group, and the proportion of the product feature tags belonging to the historical demand persistence feature tag set in the platform product feature tag group is less than the high demand proportion threshold, then the product corresponding to the platform product feature tag group is determined as a historical demand persistence category 3 product; All of the products with historical demand for the first category, the products with historical demand for the second category, and the products with historical demand for the third category are collected to form the historical marketing adjustment data.

7. The personalized marketing recommendation engine data analysis method according to claim 6, characterized in that: The acquiring of the platform product information data and performing a demand change analysis based on a product feature tag combination on the current object marketing data in combination with the historical object marketing data corresponding to the marketing object to form the current marketing feature data corresponding to the marketing object include: For different marketing objects, obtaining the object product feature tag groups corresponding to different recommended products in the current object marketing data, and combining them with the object product feature tag groups corresponding to different recommended products in the corresponding historical object marketing data, performing feature comparison analysis to form feature comparison analysis result data; Based on the feature comparison and analysis result data, combined with the platform product information data, product demand change analysis is performed to form the current marketing feature data corresponding to the marketing object.

8. The personalized marketing recommendation engine data analysis method according to claim 7, characterized in that: For different marketing objects, obtaining the object product feature tag groups corresponding to different recommended products in the current object marketing data, and combining the object product feature tag groups corresponding to different recommended products in the corresponding historical object marketing data, performing feature comparison analysis to form feature comparison analysis result data, including: Perform label group overlap comparison analysis on the object product feature label groups corresponding to different recommended products in the current object marketing data and the object product feature label groups corresponding to different recommended products in the corresponding historical object marketing data in the following manner: For the object product feature label group corresponding to any of the recommended products in the current object marketing data, if there is no overlapping number of object product feature labels with those in the object product feature label group corresponding to any of the recommended products in the historical object marketing data, and the proportion of the number of object product feature labels in the object product feature label group corresponding to the recommended product in the current object marketing data exceeds a similarity proportion threshold, then the object product feature label group corresponding to the recommended product in the current object marketing data is determined as a new demand feature label group; The target product feature tag groups corresponding to all the recommended products in the current target marketing data are traversed, and all the new demand feature tag groups are collected to form feature comparison and analysis result data.

9. The personalized marketing recommendation engine data analysis method according to claim 8, characterized in that: The product demand change analysis is performed based on the feature comparison analysis result data and combined with the platform product information data to form the current marketing feature data corresponding to the marketing target, including: The following demand change analysis is performed on the different new demand feature tag groups in the feature comparison analysis result data, combined with the platform product feature tag groups of different products in the platform product information data: For the new demand feature tag group, if there is a platform product feature tag group that includes all the target product feature tags in the new demand feature tag group, then the product corresponding to the platform product feature tag group is determined as a current demand category 1 product; For the new demand feature label group, if some of the target product feature labels in the platform product feature label group are identical to some of the target product feature labels in the new demand feature label group, and the proportion of the number of identical target product feature labels in the new demand feature label group is not less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as a current demand category II product; For the new demand feature label group, if some of the target product feature labels in the platform product feature label group are identical to some of the target product feature labels in the new demand feature label group, and the proportion of the number of identical target product feature labels in the new demand feature label group is less than the change demand proportion threshold, then the product corresponding to the platform product feature label group is determined as a current demand category three product; All of the currently demanded category 1 products, the currently demanded category 2 products, and the currently demanded category 3 products are collected to form the current marketing feature data.

10. The personalized marketing recommendation engine data analysis method according to claim 9, characterized in that: The step of performing product information recommendation analysis based on the marketing adjustment data to form target marketing product recommendation data includes: According to the historical marketing adjustment data, the products with a continuous historical demand for category 1 and the products with a current demand for category 1 are determined as category 1 recommended products, the products with a continuous historical demand for category 2 and the products with a current demand for category 2 are determined as category 2 recommended products, and the products with a continuous historical demand for category 3 and the products with a current demand for category 3 are determined as category 3 recommended products; Allocate the recommended traffic of the first category of recommended products, the second category of recommended products, and the third category of recommended products, and ensure that the allocated recommended traffic gradually decreases as the categories increase; The recommended traffic allocated to the recommended products of different categories is obtained to form the object marketing product recommendation data corresponding to the marketing object.

11. A personalized marketing recommendation engine data analysis platform, using the personalized marketing recommendation engine data analysis method according to any one of claims 1 to 10, characterized in that: include: A data collection unit, used to obtain platform product information and object marketing data and object demand search data of different marketing objects; A feature analysis unit is configured to perform data classification analysis based on demand changes based on the object marketing data and object demand search data of different marketing objects acquired by the data acquisition unit, thereby forming object periodic marketing data corresponding to the marketing objects; The recommendation adjustment unit is used to perform demand change analysis based on the marketing product characteristics based on the object period marketing data formed by the feature analysis unit and the platform product information data obtained by the data acquisition unit to form marketing adjustment data of the marketing object, and perform product information recommendation analysis to form object marketing product recommendation data.

Images