Search

CN-121998716-A - AI intelligent advertisement automatic putting method and system

CN121998716ACN 121998716 ACN121998716 ACN 121998716ACN-121998716-A

Abstract

The invention relates to the technical field of advertisement delivery, in particular to an automatic AI intelligent advertisement delivery method and system, comprising the steps of collecting behavior characteristics of a current use session of a user, generating scene fingerprints representing the current use state, matching corresponding advertisement types from pre-established advertisement demand portraits based on the scene fingerprints, determining advertisement type sets which are allowed to participate in the delivery, acquiring corresponding conversion flow information aiming at advertisements in the advertisement type sets, comparing continuous interaction capacity of the current use session with the conversion flow information, screening out advertisements with conversion conditions in the current use session, recording interaction types and occurrence sequences of the delivered advertisements in the use session, forming session delivery tracks, combining with user behavior continuity, generating session evolution states, controlling the range of advertisement participation, and collecting and storing user feedback information according to use state identification after the advertisement delivery is completed.

Inventors

  • ZHENG BIN

Assignees

  • 杭州艾羚网络科技有限公司

Dates

Publication Date
20260508
Application Date
20260129

Claims (10)

  1. An AI intelligent advertisement automatic putting method is characterized by comprising the following steps: s1, collecting behavior characteristics of a current use session of a user, and generating scene fingerprints of the current use state; S2, according to the scene fingerprint, matching corresponding advertisement types from pre-established advertisement demand portraits, and determining an advertisement type set which is allowed to participate in the current advertisement request processing; s3, aiming at advertisements in the advertisement type set, acquiring corresponding conversion flow information, comparing the continuous interaction capacity of the current use session with the conversion flow information, and screening out advertisements with conversion conditions in the current use session; S4, in the current use session, aiming at the screened advertisement record interaction type and occurrence sequence of the advertisement, forming a session delivery track, matching the session delivery track with a predefined conversion guide sequence, and determining the advertisement type range corresponding to the next advertisement delivery in the advertisement type set; S5, generating a session evolution state based on the session release track and the continuity of user behaviors in the current use session, and only allowing advertisements meeting release unlocking conditions corresponding to the session evolution state to participate in release within the advertisement type range; s6, after the advertisement is put, collecting feedback information of the user on the put advertisement, recording a corresponding use state identifier when feedback is generated, and storing the advertisement feedback information according to the use state identifier.
  2. 2. The AI-intelligent advertisement delivery method according to claim 1, wherein in S1, the collecting the behavior characteristics of the current use session of the user includes: After a user uses a session, acquiring position change information, session duration information and page stay and page switching behavior information of a user terminal in the use session; And carrying out time serialization processing on the collected behavior information to form a session behavior feature set reflecting the operation continuity and the use stability of the user in the use session.
  3. 3. The AI-intelligent advertisement delivery method according to claim 1, wherein in S1, the generating a scene fingerprint of the current usage state includes: constructing a conversation behavior feature set into a time sequence feature input, and inputting the time sequence feature input into a scene recognition model; the scene recognition model is constructed based on a neural network for processing time sequence characteristics, wherein the neural network comprises at least one of a cyclic neural network, a long-term and short-term memory network or a sequence modeling network based on an attention mechanism; Sequence feature learning is carried out on session behavior features through a neural network, and a state vector representing the current use session state of a user is output; Based on the state vector, determining a use scene identifier corresponding to the current use session of the user through a state classification algorithm, and taking the use scene identifier as a scene fingerprint of the current use state.
  4. 4. The AI-intelligent automatic advertising method according to claim 1, wherein in S2, the matching the corresponding advertisement type from the pre-established advertisement demand portraits comprises: Acquiring advertisement demand portraits corresponding to candidate advertisements, wherein the advertisement demand portraits are used for describing demand attributes of the advertisements on the aspects of interaction complexity, continuous operation requirements and exposure dependence characteristics; performing suitability analysis on the scene fingerprint and the advertisement demand portrait, and determining the meeting condition of the scene fingerprint on the advertisement demand portrait; And classifying the advertisements meeting the preset adaptation conditions into advertisement types corresponding to the current use state according to the suitability analysis result, and determining an advertisement type set which is allowed to participate in the current advertisement request processing according to the advertisement types.
  5. 5. The AI-intelligent advertisement delivery method according to claim 1, wherein in S3, the obtaining the corresponding conversion flow information includes: pre-establishing a conversion flow description corresponding to the conversion behavior of each candidate advertisement, wherein the conversion flow description is used for representing the sequence of operation steps required for completing the conversion of the advertisement; associating, for at least some of the operating steps, corresponding continuous operating conditions for describing minimum continuous operating requirements required to complete the corresponding operating step; and taking the operation step sequence and the corresponding continuous operation condition as the conversion flow information of the advertisement.
  6. 6. The AI-intelligent automatic advertising method according to claim 1, wherein in S3, the comparison between the continuous interactive capability based on the current use session and the conversion flow information comprises: Generating continuous interactive capability descriptions for representing continuous operation capability of the user in the using session based on page stay behavior, operation interval and operation interruption conditions in the current using session of the user; according to the sequence of operation steps in the conversion flow information, sequentially matching the continuous interactive capability description with continuous operation conditions of corresponding operation steps; Only when the continuous interactive capability description can cover the continuous operation condition corresponding to the operation step, the advertisement is judged to have the conversion completing condition in the current use session.
  7. 7. The AI-intelligent automatic advertising method according to claim 1, wherein in S4, the forming a session-delivery trajectory comprises: Recording the interaction type of the advertised in the current use session of the user, wherein the interaction type is used for representing the interaction mode of the user on the advertisement; sequentially processing the interaction types according to the sequence of the advertisements in the use session to form a session interaction sequence reflecting the process of the advertisements in the use session; And taking the session interaction sequence as a session putting track.
  8. 8. The AI-intelligent advertisement delivery method according to claim 1, wherein generating a session evolution state based on the session delivery trajectory and the continuity of user behavior in the currently used session in S5 comprises: based on the session putting track, acquiring the interaction type of the advertisement put in the current using session and the putting sequence information thereof; In the use session, collecting page stay behavior, continuous operation duration and operation interruption condition of the user, and representing behavior continuity of the user in the use session; and carrying out joint analysis on the session release track and the behavior continuity, determining the session stage where the current session is used, and taking the determined session stage as a session evolution state.
  9. 9. The AI-intelligent advertisement automatic delivery method according to claim 1, wherein in S5, the delivery unlock condition includes: presetting corresponding release unlocking conditions for advertisements with different advertisement types or conversion flow complexity levels, wherein the release unlocking conditions are used for limiting session evolution states which are required to be met by the participation of the advertisements in release; After the session evolution state is generated, the session evolution state is matched with the release unlocking condition, and whether the current use session meets the release unlocking condition of the corresponding advertisement is determined; and only when the session evolution state meets the release unlocking condition of the corresponding advertisement, allowing the advertisement to participate in release processing in the current use session.
  10. An AI-intelligent automatic advertisement delivery system, characterized in that it is used in the AI-intelligent automatic advertisement delivery method according to any one of claims 1 to 9, said system comprising the following modules: The session behavior analysis module is used for collecting behavior characteristics of a current use session of a user and generating scene fingerprints representing the current use state; The advertisement type matching module is used for matching corresponding advertisement types from pre-established advertisement demand portraits according to the scene fingerprints and determining an advertisement type set which is allowed to participate in the current advertisement request processing; The conversion completeness judging module is used for acquiring conversion flow information corresponding to the advertisements in the advertisement type set, comparing the conversion flow information with the continuous interaction capacity of the current use session, and screening out the advertisements with the conversion completion conditions in the current use session; the session delivery track management module is used for recording interaction types and occurrence sequences of delivered advertisements in a current use session to form a session delivery track, matching the session delivery track with a predefined conversion guide sequence, and determining an advertisement type range corresponding to the next advertisement delivery in the advertisement type set; the release unlocking control module is used for generating a session evolution state based on the session release track and the continuity of user behaviors in the current use session, and only allowing advertisements meeting release unlocking conditions corresponding to the session evolution state to participate in release within the advertisement type range; And the feedback information management module is used for collecting feedback information of the user on the advertised after the advertisement is placed, recording a corresponding use state identifier when feedback is generated, and storing the advertisement feedback information according to the use state identifier.

Description

AI intelligent advertisement automatic putting method and system Technical Field The invention relates to the technical field of advertisement delivery, in particular to an AI intelligent advertisement automatic delivery method and system. Background With the popularization of mobile internet and intelligent terminals, advertisement delivery gradually develops from a traditional manual configuration mode to an automatic delivery mode based on an algorithm. The existing automatic advertisement putting method generally screens, sorts and puts the advertisements based on the historical browsing behavior, interest preference or long-term portrait information of the user, and partial technical schemes also combine the geographical position information or equipment information of the user to perform area-oriented or crowd-oriented processing on the advertisements, so that the automation and the scale of advertisement putting are realized to a certain extent. In the practical application process, the use state of the user on the same terminal has obvious dynamic change characteristics, for example, the user can be in a state of commuting, short-time browsing application and stable stay and long-time application, the conventional advertisement putting method mainly depends on static user portraits or coarse-granularity directional information, and the real-time use state of the current use session of the user is difficult to accurately describe, so that advertisement putting decisions cannot be distinguished and processed aiming at different use states, and the suitability of advertisement putting is affected. Disclosure of Invention In order to make up for the defects, the invention provides an AI intelligent automatic advertisement delivery method and system, aiming at solving the problem that advertisement delivery decisions cannot be distinguished according to different use states, so that the suitability of advertisement delivery is affected. In a first aspect, the present invention provides a method for automatically delivering an AI-intelligent advertisement, including the following steps: s1, collecting behavior characteristics of a current use session of a user, and generating scene fingerprints of the current use state; S2, according to the scene fingerprint, matching corresponding advertisement types from pre-established advertisement demand portraits, and determining an advertisement type set which is allowed to participate in the current advertisement request processing; s3, aiming at advertisements in the advertisement type set, acquiring corresponding conversion flow information, comparing the continuous interaction capacity of the current use session with the conversion flow information, and screening out advertisements with conversion conditions in the current use session; S4, in the current use session, aiming at the screened advertisement record interaction type and occurrence sequence of the advertisement, forming a session delivery track, matching the session delivery track with a predefined conversion guide sequence, and determining the advertisement type range corresponding to the next advertisement delivery in the advertisement type set; S5, generating a session evolution state based on the session release track and the continuity of user behaviors in the current use session, and only allowing advertisements meeting release unlocking conditions corresponding to the session evolution state to participate in release within the advertisement type range; s6, after the advertisement is put, collecting feedback information of the user on the put advertisement, recording a corresponding use state identifier when feedback is generated, and storing the advertisement feedback information according to the use state identifier. By adopting the technical scheme, the acquisition and analysis of the current use session behavior characteristics of the user are introduced, and the advertisement delivery flow is controlled in stages based on the generated scene fingerprint, so that the advertisement delivery decision process can be combined with the use state of the current use session of the user to be processed, the advertisement delivery decision can be distinguished according to different use states, and the problem that the delivery suitability is insufficient due to the fact that the real-time use state of the user is difficult to be described in the conventional advertisement delivery method is solved. Preferably, in S1, the collecting the behavior characteristics of the current session used by the user includes: After a user uses a session, acquiring position change information, session duration information and page stay and page switching behavior information of a user terminal in the use session; And carrying out time serialization processing on the collected behavior information to form a session behavior feature set reflecting the operation continuity and the use stability of the user in the use s