Editing content optimization system and method based on user feedback quantization signal

Abstract

The invention discloses an edited content optimization system and method based on a user feedback quantization signal, and belongs to the technical field of video content generation and optimization. The system comprises a user feedback acquisition module, a data modeling and feedback vector generation module, an editing optimization and regeneration module and an automatic comparison and iteration closed loop module. The method comprises the steps of collecting user watching behavior data; modeling the feedback data to generate a feedback vector; inputting the feedback vector into a director language model to adjust video structure parameters; and generating an optimized version and carrying out multiple rounds of test iteration. By establishing an intelligent closed loop of generation-feedback-optimization-regeneration, automatic and continuous optimization of video content based on user feedback is realized, the complete playing rate, click rate and conversion rate of the video are remarkably improved, and the problem that an existing AI editing system lacks learning optimization from user behavior data is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of video content generation and optimization technology, specifically to an editing content optimization system and method based on user feedback quantization signals. Background Technology

[0002] Existing AI editing systems mostly employ generative methods, generating video content based on preset rules or templates, lacking dynamic response and learning mechanisms for user behavior data. These systems cannot adaptively optimize content structure based on actual user viewing behavior, leading to a discrepancy between generated content and user preferences, impacting content conversion efficiency and user experience. Therefore, there is an urgent need for an intelligent editing optimization method that can achieve a closed loop of "generation-feedback-regeneration". Summary of the Invention

[0003] This invention provides a video editing content optimization system and method based on user feedback quantization signals. By modeling user behavior data and embedding a director's language model, it achieves continuous optimization and iteration of video content.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a content editing optimization system based on user feedback quantization signals, comprising:

[0005] The user feedback collection module is used to collect user behavior data;

[0006] The data modeling and feedback vector generation module is used to model the user behavior data and generate feedback vectors.

[0007] The editing optimization and regeneration module is used to embed the feedback vector into the director's language model and adjust the video structure parameters;

[0008] An automatic comparison and iterative closed-loop module is used to test and iteratively optimize multiple video versions.

[0009] Preferably, the user behavior data includes completion rate, click-through rate, user dwell time, bounce rate, and hotspot click data.

[0010] Preferably, the data modeling module uses statistical models or machine learning methods to fit the user behavior data.

[0011] Preferably, the feedback vector is used to control the output parameters of the director's language model, including video rhythm, shot structure, and visual stimulus intensity.

[0012] Preferably, the system supports multi-platform adaptation, including Douyin, Bilibili, and TikTok.

[0013] Preferably, a method for optimizing edited content based on user feedback quantization signals includes:

[0014] Collect user feedback data on video content;

[0015] The feedback data is modeled to generate a feedback vector;

[0016] The feedback vector is input into the director's language model to adjust the video structure parameters;

[0017] Optimized video versions are generated and subjected to multiple rounds of testing and iteration.

[0018] The beneficial effects of this invention are as follows: it realizes a quantitative mapping from user feedback to content structure, forming a closed loop of "feedback-driven content evolution"; it significantly improves the completion rate, click-through rate and conversion rate of video content; it supports multi-platform adaptation and automatic comparison of multiple versions, improving the accuracy and efficiency of content delivery; and it is suitable for highly feedback-sensitive scenarios, such as advertising, short video generation, and e-commerce content production. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 The overall architecture of the system of the present invention is shown;

[0021] Figure 2 This is a flowchart illustrating the method of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] like Figure 1 As shown, an editing content optimization system based on user feedback quantization signals is presented. This system consists of four core modules, which work collaboratively with each other through data interfaces and model parameter transfer.

[0024] 1. User Feedback Collection Module

[0025] This module acquires user behavior data through data acquisition components integrated into the video player or through application programming interfaces provided by the platform. The collected metrics include, but are not limited to:

[0026] Completion rate: Reflects the percentage of video content that is watched in its entirety.

[0027] Click-through rate: Reflects the user's level of interest in interactive elements in the video.

[0028] User dwell time: Indicates how long a user focuses on video content.

[0029] Bounce point: Records the specific time and location when a user exits the viewing session.

[0030] Hotspot click data: Displays user attention to specific areas of a video using a heatmap format.

[0031] The collected data is organized according to a preset data structure, including video identifiers and various indicator values, and stored in a dedicated time-series database for subsequent analysis and processing.

[0032] 2. Data Modeling and Feedback Vector Generation Module

[0033] This module performs in-depth analysis of the collected user behavior data and extracts key feature parameters of the video structure, including scene duration distribution, shot switching frequency, timing and duration of subtitle appearance, and time point when a call to action element appears.

[0034] Statistical modeling or machine learning methods are used to establish a mapping relationship between user behavior metrics and video structural features. By analyzing the influence of each structural feature on user feedback, a feedback vector with clear physical meaning is generated. This feedback vector is a normalized weight array that accurately reflects the influence weight of different video structural parameters on user experience.

[0035] 3. Editing Optimization and Regeneration Module

[0036] This module uses the generated feedback vectors as control parameters input to the director's language model to guide the optimization and adjustment of video content. Specific optimization strategies include:

[0037] Based on scenario-level user bounce rate data, automatically adjust scenario duration or delete underperforming content segments.

[0038] Analyze the click effects of calls to action elements to optimize their timing and presentation.

[0039] Enhance the visual presentation of video areas that attract high user attention.

[0040] Adjust the video pacing and camera switching frequency based on user dwell patterns.

[0041] The optimized video structure is output in a structured configuration format, clearly annotating parameters such as the duration of each scene, covered elements, and key action time points, providing precise guidance for video regeneration.

[0042] 4. Automatic comparison and iterative closed-loop module

[0043] This module manages the testing and optimization process for multiple video versions. The specific implementation process includes:

[0044] Stream different optimized versions of video content to the target platform in parallel.

[0045] Real-time collection of user feedback data from each version and calculation of overall performance score.

[0046] The best-performing version is automatically selected as the benchmark for the next round of optimization based on the scoring results.

[0047] The preferred version of the feedback data is incorporated into the training sample set to update the data model.

[0048] The above process forms a complete closed loop of "generation, testing, optimization, and regeneration," enabling continuous improvement of video content. Experimental data shows that after multiple rounds of iterative optimization, the completion rate of video content can be increased by more than 25%, the click-through rate by more than 20%, and user engagement significantly improved.

[0049] like Figure 2 As shown, the system's workflow includes the following steps:

[0050] The first step is to collect user viewing behavior data in real time through data acquisition components;

[0051] The second step is to clean and extract features from the collected data to build a user feedback model.

[0052] The third step is to adjust the video structure parameters based on the feedback vector output by the model.

[0053] The fourth step is to regenerate the video content based on the optimized parameters;

[0054] The fifth step is to conduct multi-version testing and collect feedback data;

[0055] The sixth step is to update and optimize the model based on the test results, and start a new optimization cycle.

[0056] This system can adapt to the content characteristics of various video platforms, including but not limited to Douyin, Bilibili, and TikTok, and achieves precise content optimization through platform-specific user behavior pattern analysis.

[0057] In practical applications, this system effectively enhances the performance of video content. Taking e-commerce promotional videos as an example, after three rounds of iterative optimization, the video completion rate increased from the initial 48% to 61%, the click-through rate increased from 12% to 17%, and the conversion rate increased by more than 35%, significantly outperforming traditional editing methods.

[0058] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A content editing optimization system based on user feedback quantization signals, characterized in that, include: The user feedback collection module is used to collect user behavior data; The data modeling and feedback vector generation module is used to model the user behavior data and generate feedback vectors. The editing optimization and regeneration module is used to embed the feedback vector into the director's language model and adjust the video structure parameters; An automatic comparison and iterative closed-loop module is used to test and iteratively optimize multiple video versions.

2. The system according to claim 1, characterized in that, The user behavior data includes completion rate, click-through rate, user dwell time, bounce rate, and hotspot click data.

3. The system according to claim 2, characterized in that, The data modeling module uses statistical models or machine learning methods to fit user behavior data.

4. The system according to claim 3, characterized in that, The feedback vector is used to control the output parameters of the director's language model, including video rhythm, shot structure, and visual stimulus intensity.

5. The system according to claim 1, characterized in that, The system supports multi-platform adaptation, including Douyin, Bilibili, and TikTok.

6. A method for optimizing edited content based on user feedback quantization signals, characterized in that, include: Collect user feedback data on video content; The feedback data is modeled to generate a feedback vector; The feedback vector is input into the director's language model to adjust the video structure parameters; Optimized video versions are generated and subjected to multiple rounds of testing and iteration.

Images