Search

CN-121998712-A - Advertisement delivery resource scheduling management method and system based on big data analysis

CN121998712ACN 121998712 ACN121998712 ACN 121998712ACN-121998712-A

Abstract

The invention discloses a method and a system for scheduling and managing advertisement delivery resources based on big data analysis, which relate to the technical field of data processing and comprise the steps of collecting multi-source advertisement delivery data for batch processing and fusion calculation to obtain real-time user behavior characteristics; introducing historical conversion rule data, performing multi-objective optimization on the real-time user behavior characteristics and the historical conversion rule data to construct a preliminary scheduling strategy, setting advertisement putting constraint conditions to execute the preliminary scheduling strategy to perform resource allocation optimization to construct a resource scheduling scheme, simulating to execute the resource scheduling scheme to perform effect tracking, obtaining simulation scheduling parameters to perform incremental update, and constructing a closed-loop optimization report of advertisement putting resource scheduling. The invention solves the technical problems that the distribution of advertisement putting resources in the prior art lacks a dynamic optimization mechanism and is difficult to accurately schedule according to real-time data, and achieves the technical effects of improving the distribution precision of advertisement resources and putting conversion effects.

Inventors

  • LIN ZIYUE

Assignees

  • 武汉遐迩数媒科技有限公司

Dates

Publication Date
20260508
Application Date
20260226

Claims (10)

  1. 1. The advertisement delivery resource scheduling management method based on big data analysis is characterized by comprising the following steps: Collecting multi-source advertisement putting data to perform batch processing fusion calculation to obtain real-time user behavior characteristics; introducing historical conversion rule data, performing multi-objective optimization on the real-time user behavior characteristics in combination with the historical conversion rule data, and constructing a preliminary scheduling strategy; setting advertisement putting constraint conditions, executing the preliminary scheduling strategy to perform resource allocation optimization, and making a resource scheduling scheme; And simulating and executing the resource scheduling scheme to track the effect, obtaining simulation scheduling parameters to update the increment, and constructing a closed-loop optimization report of the advertisement delivery resource scheduling.
  2. 2. The method for scheduling and managing advertisement delivery resources based on big data analysis according to claim 1, wherein the method comprises the steps of collecting multi-source advertisement delivery data for batch processing and fusion calculation to obtain real-time user behavior characteristics, and comprises the following steps: setting a plurality of granularity parameters based on a plurality of advertisement putting platforms, and setting a plurality of granularity sliding windows according to the granularity parameters, wherein the granularity sliding windows comprise fine granularity windows, medium granularity windows and coarse granularity windows; clicking calculation is carried out on the user based on the fine granularity window, and clicking strength data are obtained; Tracking the preference of the user based on the middle granularity window to obtain browsing interest data; performing liveness analysis on the user based on the coarse-granularity window, and identifying a plurality of liveness periods; Carrying out batch processing on the click intensity data, the browsing interest data and the plurality of liveness time periods to generate a batch data set, wherein the batch data set comprises micro batch data and macro batch data; Performing streaming calculation on the micro batch data to generate a user real-time behavior index, and performing batch calculation on the macro batch data to generate a user historical behavior mode index; performing feature splicing fusion on the user real-time behavior index and the historical behavior mode index to obtain feature fusion data; And carrying out normalization processing on the feature fusion data to generate the real-time user behavior features.
  3. 3. The advertisement delivery resource scheduling management method based on big data analysis according to claim 2, wherein the characteristic splicing fusion is performed on the user real-time behavior index and the historical behavior pattern index to obtain characteristic fusion data, and the method comprises the following steps: traversing a plurality of users to carry out classification identification, carrying out storage allocation on the plurality of users according to the plurality of user identifications, determining a plurality of feature storage spaces, and constructing a feature fusion mapping table based on the plurality of feature storage spaces; writing the user real-time behavior indexes into real-time feature areas of the feature storage spaces to perform feature analysis, and constructing real-time behavior features; Calling historical characteristic areas of the plurality of characteristic storage spaces based on the historical behavior mode indexes to perform characteristic analysis, and constructing historical behavior characteristics; Performing dimension alignment on the real-time behavior feature and the historical behavior feature according to a feature fusion mapping table, and performing transverse splicing on the real-time behavior feature and the historical behavior feature according to an alignment result to generate a first fusion feature vector; Performing feature conflict recognition on the first fusion feature vector, extracting conflict items, performing feature rejection, and generating a second fusion feature vector; and performing cross verification based on the second fusion feature vector to generate feature fusion data.
  4. 4. The advertisement delivery resource scheduling management method based on big data analysis according to claim 3, wherein the construction process of the history transformation law data comprises the following steps: an advertisement history conversion basic data set is called, the advertisement history conversion basic data set is grouped according to the user identifications, and a complete history behavior time sequence of the users is constructed; performing sequence cutting on the complete historical behavior time sequences of a plurality of users to obtain a complete path sequence; Frequent sequence pattern mining screening is carried out on the complete path sequence, and a typical conversion path is determined; carrying out path characteristic analysis on the typical conversion path to generate a path efficiency index; Traversing channel conversion preference analysis of historical behaviors of a plurality of users to obtain user channel preference characteristics; Traversing the time period conversion distribution of the historical behaviors of a plurality of users to obtain user time period response characteristics; And combining and packaging the typical conversion path, the path efficiency index, the user channel preference characteristic and the user time period response characteristic to construct the historical conversion rule data.
  5. 5. The advertisement delivery resource scheduling management method based on big data analysis according to claim 1, wherein the real-time user behavior characteristics are combined with the historical transformation law data to perform multi-objective optimization, and a preliminary scheduling policy is constructed, the method comprising: performing feature alignment on the real-time user behavior features and the historical transformation rule data to generate feature alignment results; Performing feature stitching on the real-time user behavior features and the historical transformation rule data based on the feature alignment result to generate a joint feature vector; setting a combined optimization target, performing multi-target optimization based on the combined feature vector, and generating a multi-target strategy optimizing result; Performing multi-round depth reinforcement iterative learning according to the multi-target strategy optimizing result to generate multi-target gain parameters, combining according to the multi-target gain parameters, and identifying a plurality of strategy schemes; And traversing the strategy schemes to carry out screening optimization, and determining the preliminary scheduling strategy.
  6. 6. The method for scheduling and managing advertisement delivery resources based on big data analysis according to claim 5, wherein the feature-based alignment result feature-splices the real-time user behavior feature with the historical transformation law data to generate a joint feature vector, the method comprising: Performing traversal searching on the historical transformation rule data according to the real-time user behavior characteristics by taking a plurality of user identifications as index keys to generate a searching result, wherein the searching result is a presence result or an absence result; when the search result is a presence result, performing time dimension alignment on the historical behavior characteristics and the real-time user behavior characteristics, and extracting a characteristic matching result; When the search result is that the result does not exist, generating new user traversal history transformation rule data to perform global analysis, and extracting global characteristic parameters; based on the feature matching result or the global feature parameter, performing feature stitching on the real-time user behavior feature and the historical behavior feature to obtain paired features; Extracting a first type of feature based on the real-time user behavior feature, and extracting a second type of feature according to the first type of feature for the historical behavior feature, wherein the first type of feature has a corresponding relation with the second type of feature; Cross-combining the first type of features with the second type of features to generate cross feature items; And combining the paired features with the crossed feature items to generate the joint feature vector.
  7. 7. The method for scheduling and managing advertisement delivery resources based on big data analysis according to claim 1, wherein setting advertisement delivery constraint conditions to execute the preliminary scheduling policy for resource allocation optimization, and making a resource scheduling scheme, the method comprising: Introducing real-time bidding data, and performing environmental analysis according to advertisement channels based on the real-time bidding data to generate real-time market environmental parameters; executing the preliminary scheduling strategy according to the advertisement putting constraint condition to perform code conversion to obtain a plurality of coding populations, wherein the plurality of coding populations comprise a plurality of candidate resource allocation schemes; Performing fitness calculation on the plurality of candidate resource allocation schemes based on the real-time market environment parameters to obtain multi-dimensional evaluation fitness; traversing a plurality of coding ethnicities according to the multi-dimensional evaluation fitness to carry out descending order sorting, and obtaining a benefit sorting sequence; Randomly extracting the plurality of coding populations based on the benefit ordering sequence, determining a plurality of competition groups to execute cyclic cross matching, and generating a plurality of offspring parameters; performing variation iterative evaluation according to the plurality of offspring parameters, constructing a target population, performing individual decoding reduction, and obtaining a plurality of analysis parameters; And integrating the plurality of analysis parameters to construct the resource scheduling scheme.
  8. 8. The method for scheduling and managing advertisement delivery resources based on big data analysis according to claim 1, wherein the method for performing effect tracking of the resource scheduling scheme to obtain the simulation scheduling parameters comprises: Constructing an advertisement putting simulation environment, mapping the resource scheduling scheme to the advertisement putting simulation environment for simulated putting, and generating a simulated putting effect data set; introducing historical resource scheduling effects to carry out expected analysis, and setting target effect data; performing item-by-item comparison analysis on the simulated putting effect data set and the target effect data, and calculating a plurality of effect deviation indexes; performing deviation tracing attribution based on the multiple effect deviation indexes, and determining scheduling deviation parameters; And carrying out association storage on the scheduling deviation parameters in combination with the multi-effect deviation indexes to generate the simulation scheduling parameters.
  9. 9. The advertisement delivery resource scheduling management method based on big data analysis according to claim 8, wherein the simulated delivery effect data set and the target effect data are subjected to item-by-item comparison analysis, and a plurality of effect deviation indexes are calculated, the method comprising: Performing user behavior simulation analysis based on the simulated putting effect data set and the target effect data to generate a simulated response probability parameter; calculating the ratio of the simulation response probability parameter to the preset response probability parameter to generate a response deviation index; Performing market bidding simulation analysis based on the simulated putting effect data set and the target effect data, and generating simulated bidding conversion rate parameters; Calculating the ratio of the simulated bidding conversion parameter to the preset simulated bidding conversion parameter to generate a bidding deviation index; Channel response simulation analysis is carried out based on the simulated putting effect data set and the target effect data, and simulated exposure parameters are generated; Calculating the ratio of the simulated exposure parameter to the preset exposure parameter to generate an exposure deviation index; and integrating the response deviation index, the bid deviation index and the exposure deviation index to construct a plurality of effect deviation indexes.
  10. 10. An advertisement delivery resource scheduling management system based on big data analysis, characterized in that the system is used for executing the advertisement delivery resource scheduling management method based on big data analysis according to any one of claims 1-9, and the system comprises: The characteristic acquisition module is used for acquiring multi-source advertisement delivery data to perform batch processing fusion calculation so as to acquire real-time user behavior characteristics; The multi-objective optimization module is used for introducing historical transformation rule data, carrying out multi-objective optimization on the real-time user behavior characteristics in combination with the historical transformation rule data, and constructing a preliminary scheduling strategy; The system comprises a resource allocation optimization module, an effect tracking module and a closed-loop optimization report generation module, wherein the resource allocation optimization module is used for setting advertisement putting constraint conditions to execute the preliminary scheduling strategy to perform resource allocation optimization and make a resource scheduling scheme, and the effect tracking module is used for simulating and executing the resource scheduling scheme to perform effect tracking, obtaining simulation scheduling parameters to perform incremental updating and constructing a closed-loop optimization report of advertisement putting resource scheduling.

Description

Advertisement delivery resource scheduling management method and system based on big data analysis Technical Field The invention relates to the technical field of data processing, in particular to an advertisement delivery resource scheduling management method and system based on big data analysis. Background In the advertisement putting process, the resource allocation is usually subjected to strategy formulation based on preset rules or staged statistical data, the scheduling mode is relatively fixed, and the user behavior change and market environment fluctuation are difficult to respond in time. When user interests, bidding environments or channel performances are dynamically changed, the delivery strategies cannot be synchronously adjusted, so that the resource allocation is not matched with the actual conversion trend, and the advertisement delivery effect and the resource utilization efficiency are affected. Disclosure of Invention The application provides a big data analysis-based advertisement delivery resource scheduling management method and a big data analysis-based advertisement delivery resource scheduling management system, which are used for solving the technical problems that in the prior art, advertisement delivery resource allocation lacks a dynamic optimization mechanism and accurate scheduling is difficult to carry out according to real-time data. In view of the above problems, the application provides a method and a system for scheduling and managing advertisement delivery resources based on big data analysis. In a first aspect of the present application, there is provided an advertisement delivery resource scheduling management method based on big data analysis, the method comprising: Collecting multi-source advertisement putting data to perform batch processing fusion calculation to obtain real-time user behavior characteristics, introducing historical transformation rule data, performing multi-objective optimization on the real-time user behavior characteristics and the historical transformation rule data to construct a preliminary scheduling strategy, setting advertisement putting constraint conditions to execute the preliminary scheduling strategy to perform resource allocation optimization to formulate a resource scheduling scheme, performing effect tracking by simulating the execution of the resource scheduling scheme to obtain simulation scheduling parameters to perform incremental update, and constructing a closed-loop optimization report of advertisement putting resource scheduling. In a second aspect of the present application, there is provided an advertisement delivery resource scheduling management system based on big data analysis, the system comprising: The system comprises a feature acquisition module, a multi-objective optimization module, a resource allocation optimization module and an effect tracking module, wherein the feature acquisition module is used for acquiring multi-source advertisement delivery data to carry out batch processing fusion calculation to obtain real-time user behavior features, the multi-objective optimization module is used for introducing historical transformation rule data, carrying out multi-objective optimization on the real-time user behavior features and the historical transformation rule data to construct a preliminary scheduling strategy, the resource allocation optimization module is used for setting advertisement delivery constraint conditions to execute the preliminary scheduling strategy to carry out resource allocation optimization to formulate a resource scheduling scheme, and the effect tracking module is used for simulating execution of the resource scheduling scheme to carry out effect tracking to obtain simulation scheduling parameters to carry out increment update and construct a closed-loop optimization report of advertisement delivery resource scheduling. One or more technical schemes provided by the application have at least the following technical effects or advantages: The method comprises the steps of collecting multi-source advertisement putting data, carrying out batch processing fusion calculation to obtain real-time user behavior characteristics, introducing historical transformation rule data, carrying out multi-objective optimization on the real-time user behavior characteristics and the historical transformation rule data, constructing a preliminary scheduling strategy, setting advertisement putting constraint conditions, executing the preliminary scheduling strategy to carry out resource allocation optimization, formulating a resource scheduling scheme, carrying out effect tracking by simulating execution of the resource scheduling scheme, obtaining simulation scheduling parameters to carry out incremental update, and constructing a closed-loop optimization report of advertisement putting resource scheduling. The application solves the technical problems that in the prior art, the advertisement delivery resour