Search

CN-121998708-A - Intelligent delivery algorithm optimization method and system applied to short video delivery platform

CN121998708ACN 121998708 ACN121998708 ACN 121998708ACN-121998708-A

Abstract

The invention provides an intelligent delivery algorithm optimization method and system applied to a short video delivery platform, which belong to the technical field of Internet advertisement delivery, and comprise the steps of firstly determining an initial content pool containing short video materials and content labels classified according to topics and a target user group touch range covering user behaviors and regional characteristics, then collecting real-time interaction data of the target user group on the initial content pool, generating a content adaptation coefficient reflecting matching association of the materials and users based on the real-time interaction data, adjusting the delivery priority of the initial content pool and the coverage dimension of the target user group touch range according to the content adaptation coefficient, and finally executing the operations circularly until a stable delivery strategy is formed and applied to continuous delivery, thereby improving the accuracy and efficiency of short video delivery.

Inventors

  • LIN YUQI

Assignees

  • 飞客网红(杭州)科技有限公司

Dates

Publication Date
20260508
Application Date
20260123

Claims (10)

  1. 1. An intelligent delivery algorithm optimization method applied to a short video delivery platform is characterized by comprising the following steps: Determining an initial content pool of short video streaming and a touch range of a target user group, wherein the initial content pool comprises short video materials classified according to topics and corresponding content labels, and the touch range of the target user group comprises user behavior characteristics and region distribution characteristics; Collecting real-time interaction data of a target user group on the initial content pool, wherein the real-time interaction data comprises browsing duration, interaction operation and jump path of a user on short video materials; Generating a content adaptation coefficient based on the real-time interaction data, wherein the content adaptation coefficient is used for reflecting the matching association of the short video material and the target user group; Adjusting the delivery priority of the initial content pool and the coverage dimension of the reaching range of the target user group according to the content adaptation coefficient; and circularly executing the operations of collecting real-time interaction data, generating a content adaptation coefficient and adjusting the delivery parameters until a stable delivery strategy is formed and the method is applied to a continuous delivery process.
  2. 2. The intelligent delivery algorithm optimization method applied to the short video delivery platform according to claim 1, wherein the determining the reach of the initial content pool and the target user group of the short video delivery comprises: screening short video materials conforming to streaming topics from a content library of a short video streaming platform, dividing the short video materials into a plurality of topic subsets according to topic classification standards, wherein each topic subset comprises at least one short video material; adding a content tag for each short video material, wherein the content tag comprises core elements, expression modes and emotion tendency descriptions related to the material; Integrating the short video materials added with the content labels and the corresponding topic subsets into an initial content pool, and establishing an index directory of the initial content pool, wherein the index directory comprises the corresponding relation between the topic subsets and the short video materials and the retrieval path of the content labels; Extracting user registration information and historical behavior records from a user database of a short video streaming platform, wherein the user registration information comprises basic materials filled by a user, and the historical behavior records comprise topics and content labels of short video materials browsed and interacted by the user in the past; dividing regional distribution characteristics of a user according to regional fields in user registration information and position associated data in a historical behavior record, wherein the regional distribution characteristics comprise administrative division information of a resident region and an active region of the user; Analyzing the attention frequency and interaction depth of short video materials of different topics and content tags in a user history behavior record, and extracting user behavior characteristics, wherein the user behavior characteristics comprise topic types, content tag combinations and interaction modes of user preference; and associating and integrating the user behavior characteristics and the region distribution characteristics, defining the touch range of the target user group, and forming a description document of the touch range, wherein the description document comprises the corresponding relation between the user behavior characteristics and the region distribution characteristics and an identification set covering the user.
  3. 3. The intelligent delivery algorithm optimization method applied to the short video delivery platform according to claim 2, wherein the adding content tags for each short video material comprises: Sharing the complete content of each short video material, and extracting core elements appearing in the short video material, wherein the core elements comprise at least one identifiable key object, and the key objects comprise characters, objects, scenes and events; analyzing shooting techniques, editing styles and narrative structures of the short video materials, and determining expression modes of the short video materials, wherein the expression modes comprise era, animation, interview, demonstration and scenario type description; Judging emotion tendencies transmitted by the short video materials through picture color tones, background music and language style elements in the short video materials, wherein the emotion tendencies comprise positive, neutral, stable and lively qualitative expressions; Combining the core element, the expression mode and the emotion tendency description into content labels according to a preset format, wherein each content label consists of a plurality of description items, and the description items are distinguished by separators; Adding a weight identifier for each description item in the content tag, wherein the weight identifier reflects the prominence degree of the description item in the short video material, and the prominence degree is determined according to the frequency, the duration and the influence of the description item; Binding the content tag added with the weight identifier with the short video material, storing the content tag in attribute information of the short video material, and establishing a one-to-one correspondence between the content tag and the short video material.
  4. 4. The intelligent delivery algorithm optimization method applied to the short video streaming platform according to claim 2, wherein the analyzing the attention frequency and the interaction depth of the short video materials of different topics and content labels in the user history behavior record, extracting the user behavior characteristics, comprises: Screening all short video materials browsed by a user from a historical behavior record of the user, and extracting the topics and content labels of all the short video materials; Counting the browsing times and total browsing time of a user on short video materials of each theme in a preset period, and calculating the attention frequency of each theme, wherein the attention frequency is the proportion of the browsing times of the theme to the total browsing times of all the themes; analyzing the interaction behavior of a user on short video materials of each content tag, wherein the interaction behavior comprises comments, sharing and collection, and calculating the interaction times and the interaction quality corresponding to each content tag, and the interaction quality is comprehensively evaluated according to the comment length and the sharing range factors; According to the ordering of the attention frequencies, selecting N topics with the front attention frequency as the topic types preferred by the user; performing combination analysis on content labels with the user interaction times larger than the set times and the interaction quality larger than the set quality, identifying content label combinations which occur at least M times simultaneously, and determining the content label combinations as content label combinations preferred by users; counting the distribution characteristics of the interaction modes used by the user in the interaction process, and determining the interaction modes preferred by the user; and integrating the theme types, the content tag combinations and the interaction modes of the user preferences into the user behavior characteristics to form a description file of the user behavior characteristics.
  5. 5. The intelligent delivery algorithm optimization method applied to the short video streaming platform according to claim 1, wherein the collecting real-time interaction data of the target user group on the initial content pool comprises the following steps: embedding a data acquisition module in a content display interface of the short video streaming platform, wherein the data acquisition module is used for recording the operation behaviors of a target user group on short video materials in an initial content pool; when users in the target user group open the short video material, the data acquisition module starts timing until the users close the short video material or jump to other contents, and a corresponding time period is recorded as the browsing duration of the users on the short video material; Monitoring operation actions of a user in the process of browsing short video materials, and recording time points, duration and associated objects of each operation action as interactive operation data, wherein the operation actions comprise clicking, sliding, commenting, sharing and collecting; Tracking the switching sequence of a user among different short video materials in an initial content pool, and recording a trigger mode of the user for jumping from one short video material to another short video material and a page path passing through the middle to form jumping path data; The data acquisition module gathers browsing duration, interactive operation data and jump path data into real-time interactive data snapshots according to a preset time interval, and each real-time interactive data snapshot comprises all operation records of a target user group on an initial content pool in the time interval; Adding a time identifier and a user identifier for each real-time interaction data snapshot, wherein the time identifier comprises the starting time and the ending time of data acquisition, and the user identifier comprises a unique code of a user in a target user group for generating operation behaviors; and storing the snapshot of the real-time interaction data added with the time identifier and the user identifier into a real-time database, and establishing a corresponding relation index of the real-time interaction data and the short video material and the target user group in the initial content pool.
  6. 6. The method for optimizing intelligent delivery algorithm applied to short video streaming platform according to claim 5, wherein tracking the switching sequence of the user between different short video materials in the initial content pool, recording the trigger mode of the user to jump from one short video material to another short video material and the page path passing in the middle, and forming jump path data, comprises: Setting a tracking identifier at a page skip node of a short video streaming platform, wherein the tracking identifier is used for identifying the operation of a user to skip from a current page to other pages; when a user executes a skip operation on one short video material page of the initial content pool, tracking, marking and recording the mark of the current short video material, marking the skip target short video material and the skip triggering operation types comprise clicking a recommended link and sliding switching; Recording intermediate pages passed in the user jumping process, wherein the intermediate pages comprise a platform home page, a classification page and a search result page, and recording the access sequence and the stay time of the intermediate pages according to time sequence; Integrating the current short video material identification, the skip triggering mode, the intermediate page access sequence and the target short video material identification according to time lines to form a single skip path record; Correlating a plurality of jump path records of the same user in the same time period, analyzing the continuity and correlation of the jump paths, and identifying the browsing rule of the user in the initial content pool; and summarizing all the jump path records into jump path data, wherein the jump path data comprises a user identifier, a jump path record set and a path analysis conclusion, and the path analysis conclusion is used for describing a habit mode of switching the short video material in the initial content pool by a user.
  7. 7. The intelligent delivery algorithm optimization method applied to the short video delivery platform according to claim 1, wherein the generating the content adaptation coefficient based on the real-time interaction data comprises: Extracting browsing duration, interaction operation data and jump path data corresponding to each short video material from the real-time interaction data, and grouping according to user identifiers in a target user group to obtain an interaction record of each user on each short video material; Analyzing the proportional relation between the browsing time length of a single user on a single short video material and the total time length of the short video material, and calculating the attention index of the user on the short video material by combining the interactive operation times and types of the user on the short video material; Analyzing the reason that the user jumps from one short video material to another short video material according to the jump path data of the user in the initial content pool, judging whether the jump is caused by the fact that the current short video material does not accord with the user expectation, and counting the proportion of each short video material which is actively jumped away by the user; Calculating the adaptation value of a single user to a single short video material by combining the attention index and the jumped-away proportion; Summarizing the adaptation values of all users in the target user group to the same short video material, and taking the comprehensive result as the content adaptation coefficient of the short video material and the target user group; Establishing a description rule for the content adaptation coefficient, wherein the description rule comprises a matching association degree description corresponding to a numerical range of the content adaptation coefficient, and the matching association degree description is used for explaining the fit condition of the short video material reflected by the content adaptation coefficient and a target user group; And storing the content adaptation coefficient of each short video material and the corresponding description rule thereof into a coefficient database, and establishing an associated index of the content adaptation coefficient, the short video material and the target user group.
  8. 8. The intelligent delivery algorithm optimization method applied to a short video streaming platform according to claim 1, wherein the adjusting the delivery priority of an initial content pool and the coverage dimension of the reach of the target user group according to the content adaptation coefficient comprises: Extracting a content adaptation coefficient of each short video material in a coefficient database, and sequencing the short video materials in an initial content pool according to the size of the content adaptation coefficient, wherein the sequencing result is used as a basis for adjusting the delivery priority; The short video materials in the first half part of the sorting result are defined as a preferential delivery group, the short video materials in the second half part of the sorting result are defined as a secondary preferential delivery group, and the short video materials in the preferential delivery group obtain more display opportunities in the streaming process; Performing aggregation analysis on content tags of short video materials in the preferential delivery group, extracting content tag combinations with occurrence frequency larger than a set frequency, taking the content tag combinations as supplementary screening conditions of an initial content pool, screening new short video materials conforming to the content tag combinations from a content library of a short video streaming platform, and adding the new short video materials to a corresponding theme subset of the initial content pool; analyzing interaction records of short video materials of a priority delivery group in a target user group, extracting common characteristics of user behavior characteristics and region distribution characteristics of the target user group, and taking the common characteristics as core coverage dimensions of a touch range; comparing the original coverage dimension with the core coverage dimension in the touch range of the target user group, reserving the overlapped coverage dimension, expanding the newly added feature description in the core coverage dimension, and forming an adjusted coverage dimension; Screening newly added users meeting the conditions from a user database of the short video streaming platform according to the adjusted coverage dimension, bringing the newly added users into the reach of the target user group, and removing original users which do not meet the adjusted coverage dimension; Recording a delivery priority adjustment result of the initial content pool and a coverage dimension adjustment result of the touch range of the target user group to form an adjustment log, wherein the adjustment log comprises parameter comparison and adjustment basis before and after adjustment.
  9. 9. The intelligent delivery algorithm optimization method applied to the short video streaming platform according to claim 8, wherein the step of screening newly added users meeting the conditions from the user database of the short video streaming platform according to the adjusted coverage dimension, bringing the newly added users into the reach of the target user group, and simultaneously removing original users not meeting the adjusted coverage dimension comprises the steps of: Decomposing the adjusted coverage dimension into a plurality of screening conditions, wherein each screening condition corresponds to one feature description in the coverage dimension, and the screening conditions comprise specific parameters of user behavior features and administrative division ranges of regional distribution features; extracting feature information of users which are not included in the touch range of a target user group from a user database of a short video streaming platform, wherein the feature information comprises user behavior features and region distribution features of the users; Comparing the characteristic information of the users which do not enter the touch range with screening conditions, determining the users which accord with all the screening conditions as newly added users, and recording unique codes of the newly added users; Extracting current user behavior characteristics and region distribution characteristics of each user from an original user identification set in the touch range of a target user group, and comparing the current user behavior characteristics and region distribution characteristics with the screening conditions of the adjusted coverage dimension; if the characteristics of the original user do not meet any screening condition, determining the user as the user to be removed, and recording the unique code of the user to be removed; adding unique codes of newly added users in a user identification set of a touch range of a target user group, deleting the unique codes of the users to be removed, and forming an updated user identification set; And extracting corresponding user information from the user database according to the updated user identification set, and updating the description document of the touch range of the target user group.
  10. 10. The intelligent delivery algorithm optimization system applied to the short video delivery platform is characterized by comprising a processor and a memory, wherein the memory is connected with the processor, the memory is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the memory so as to realize the intelligent delivery algorithm optimization method applied to the short video delivery platform according to any one of claims 1-9.

Description

Intelligent delivery algorithm optimization method and system applied to short video delivery platform Technical Field The invention relates to the technical field of Internet advertisement delivery, in particular to an intelligent delivery algorithm optimization method and system applied to a short video streaming platform. Background At the moment of the vigorous development of the short video industry, the short video delivery platform becomes a key channel for promoting products and contents by advertisers and content creators. However, the existing short video streaming algorithm has a plurality of defects. At present, most streaming algorithms often adopt a coarser way when determining the content to be launched and the target user group. For the construction of an initial content pool, various short video materials are simply listed, and the subtle classification of the material subject and the accurate content label marking are lacked, so that the accurate matching of target users is difficult during the delivery. When defining the touch range of the target user group, only some basic user attributes such as age, gender and the like are considered, so that the mining of the user behavior characteristics and the regional distribution characteristics is not deep enough, and the real requirements and preferences of the user cannot be accurately mastered. Meanwhile, the existing streaming algorithm lacks a dynamic adjustment mechanism. In the process of throwing, real-time interaction data of a target user group on throwing content cannot be timely acquired, and throwing strategies cannot be optimized in real time according to actual feedback of users. This makes it difficult for the delivery effect to reach the best, and advertisers and content creators have difficulty in achieving efficient popularization goals, wasting a lot of delivery resources and costs. Disclosure of Invention In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides an intelligent delivery algorithm optimization method applied to a short video delivery platform, where the method includes: Determining an initial content pool of short video streaming and a touch range of a target user group, wherein the initial content pool comprises short video materials classified according to topics and corresponding content labels, and the touch range of the target user group comprises user behavior characteristics and region distribution characteristics; Collecting real-time interaction data of a target user group on the initial content pool, wherein the real-time interaction data comprises browsing duration, interaction operation and jump path of a user on short video materials; Generating a content adaptation coefficient based on the real-time interaction data, wherein the content adaptation coefficient is used for reflecting the matching association of the short video material and the target user group; Adjusting the delivery priority of the initial content pool and the coverage dimension of the reaching range of the target user group according to the content adaptation coefficient; and circularly executing the operations of collecting real-time interaction data, generating a content adaptation coefficient and adjusting the delivery parameters until a stable delivery strategy is formed and the method is applied to a continuous delivery process. In still another aspect, an embodiment of the present invention further provides an intelligent delivery algorithm optimization system applied to a short video delivery platform, where the intelligent delivery algorithm optimization system includes a processor and a machine-readable storage medium, where the machine-readable storage medium is connected to the processor, and the machine-readable storage medium is used to store a program, an instruction, or a code, and the processor is used to execute the program, the instruction, or the code in the machine-readable storage medium, so as to implement the method described above. Based on the above aspects, the embodiment of the invention acquires the real-time interaction data of the target user group on the initial content pool by determining the initial content pool containing the short video material classified by the subject and the corresponding content label and the target user group touch range covering the user behavior characteristic and the region distribution characteristic, can timely capture the real feedback of the user on the short video material, can accurately reflect the matching association degree of the material and the target user group based on the generated content adaptation coefficient, adjusts the throwing priority of the initial content pool and the coverage dimension of the target user group touch range according to the content adaptation coefficient, realizes the dynamic optimization of the throwing strategy, ensures that the throw