Prediction method and device for multimedia file
By performing feature recognition and user behavior analysis on multimedia files, the problem of low user interest matching in traditional systems is solved, and more accurate multimedia file push is achieved.
Patent Information
- Application Number
- CN202510804528.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-17
- Publication Date
- 2025-09-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional multimedia file recommendation systems rely on content metadata, resulting in low user interest matching and reduced push accuracy.
By obtaining multimedia file source data, dividing the data into segments, extracting identification features and assigning progress stamps, building a user behavior feature set, setting warning thresholds, matching feature sets and predicting target users, the push accuracy can be improved.
Improves the accuracy of multimedia file push and reduces the release of duplicate and abnormal content.
Smart Images
Figure CN120705341A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of multimedia data processing, and in particular to a method and device for predicting multimedia files. Background Art
[0002] With the explosive growth of the Internet content ecosystem, the average daily upload volume of multimedia files such as videos, audios, and pictures has exceeded 100 million.
[0003] Traditional content recommendation systems mostly rely on multimedia file content metadata for publishing, such as multimedia file titles and tags, which leads to a low match between multimedia files and adapted user interests, thereby reducing the accuracy of multimedia file push; therefore, it is very necessary to propose a multimedia file prediction method and device that can improve the accuracy of multimedia file push. Summary of the Invention
[0004] The present invention aims to provide a method and device for predicting multimedia files, so as to improve the accuracy of multimedia file push.
[0005] To achieve the above object, the present invention adopts a multimedia file prediction method comprising the following steps:
[0006] Obtain multimedia file source data, divide the data into segments, extract identification features of each data segment, assign progress stamps to the identification features, and output a multimedia file identification feature set;
[0007] Obtaining a user behavior feature set, predicting the target user of the multimedia file based on the identification feature set of the multimedia file, and triggering a multimedia file publishing request;
[0008] Set up an alert threshold, match the identification feature set of multiple multimedia files and obtain the matching value, compare the alert threshold and the matching value, and then execute the multimedia file publishing request.
[0009] Among them, in the steps of obtaining multimedia file source data, dividing data segments, extracting identification features of each data segment, assigning progress stamps to the identification features, and outputting a multimedia file identification feature set:
[0010] Setting a time interval and dividing the multimedia file source data into data segments according to the time interval;
[0011] Extracting identification features of each data segment, and using the identification features as type features of the data segment;
[0012] Output multimedia file recognition feature set.
[0013] After extracting the identification features of each data segment and using the identification features as the type features of the data segment:
[0014] According to the total duration of the multimedia file, progress stamp data is assigned to the identification feature of each data segment.
[0015] Wherein, in the step of outputting the multimedia file recognition feature set:
[0016] Obtain the identification features of each data segment, sort them according to the progress stamp, obtain the multimedia file identification feature set, and output it.
[0017] Among them, in the step of obtaining a user behavior feature set, predicting the target user of the multimedia file based on the identification feature set of the multimedia file, and triggering a multimedia file publishing request:
[0018] Construct user behavior feature sets and output user preference types;
[0019] Calculate the matching degree between the multimedia file identification feature set and the user preference type, obtain the target users of the multimedia file, and predict the spread of the multimedia file;
[0020] Get the prediction results and trigger a multimedia file publishing request.
[0021] Among them, in the step of constructing the user behavior feature set and outputting the user preference type:
[0022] Obtain user behavior data, convert user behavior into behavior features, build a user behavior feature set, identify user preferences based on the user behavior feature set, and output user preference types.
[0023] Among them, in the steps of setting a warning threshold, matching identification feature sets of multiple multimedia files and obtaining matching values, and then executing a multimedia file publishing request after comparing the warning threshold and the matching values:
[0024] Set up warning thresholds and calculate the matching value between the current file identification feature set and the existing file identification feature set;
[0025] Compare the warning threshold with the matching value and output the comparison result.
[0026] Among them, in the step of comparing the warning threshold and the matching value and outputting the comparison result:
[0027] When the matching value is less than or equal to the warning threshold, the current file is identified as a normal file and the file publishing operation continues.
[0028] Among them, in the step of comparing the warning threshold and the matching value and outputting the comparison result:
[0029] When the matching value is greater than the warning threshold, the current file is identified as an abnormal file and the file publishing operation is terminated.
[0030] The present invention also provides a multimedia file prediction device, comprising a recognition feature set acquisition module, a target user prediction module, and a publishing request execution module; wherein:
[0031] The identification feature set acquisition module is used to obtain multimedia file source data, divide the data into segments, extract identification features of each data segment, assign progress stamps to the identification features, and output the multimedia file identification feature set;
[0032] The target user prediction module is used to obtain a user behavior feature set, predict the target user of the multimedia file based on the identification feature set of the multimedia file, and trigger a multimedia file publishing request;
[0033] The publishing request execution module is used to set an alert threshold, match the identification feature sets of multiple multimedia files and obtain a matching value, and execute the multimedia file publishing request after comparing the alert threshold and the matching value.
[0034] A multimedia file prediction method and device of the present invention respectively adopt the identification feature set acquisition module, the target user prediction module, and the release request execution module to perform the following steps: obtaining multimedia file source data, dividing data segments, extracting identification features of each data segment, assigning a progress stamp to the identification features, and outputting a multimedia file identification feature set; obtaining a user behavior feature set, predicting the target user of the multimedia file based on the identification feature set of the multimedia file, and triggering a multimedia file release request; setting an alert threshold, matching the identification feature sets of multiple multimedia files and obtaining a matching value, and executing the multimedia file release request after comparing the alert threshold and the matching value; improving the accuracy of multimedia file push by performing feature recognition based on the multimedia file and predicting the target user of the multimedia file based on the user's behavior characteristics. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0036] Figure 1 4 is a flowchart of the steps of the multimedia file prediction method of the present invention.
[0037] Figure 2 It is a step flow chart of S100 of the present invention.
[0038] Figure 3 It is a step flow chart of S200 of the present invention.
[0039] Figure 4 It is a step flow chart of S300 of the present invention.
[0040] Figure 5 It is a structural principle diagram of the multimedia file prediction device of the present invention.
[0041] Figure 6 It is a structural principle diagram of the electronic device of the present invention.
[0042] 401-Identification feature set acquisition module, 402-Target user prediction module, 403-Release request execution module. DETAILED DESCRIPTION
[0043] Exemplary embodiments are described in detail herein, with examples illustrated in the accompanying drawings. When the following description refers to the drawings, identical numerals in different drawings represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with this application.
[0044] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
[0045] It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".
[0046] See also Figures 1 to 4 The present invention provides a method for predicting multimedia files, comprising the following steps:
[0047] S100: Acquire multimedia file source data, divide the data into segments, extract identification features of each data segment, assign a progress stamp to the identification features, and output a multimedia file identification feature set.
[0048] In this embodiment, the multimedia file source data is obtained, divided into data segments, the identification features of each data segment are extracted, and a progress stamp is assigned to the identification features, and a multimedia file identification feature set is output. The specific process is as follows:
[0049] S101: Setting time intervals, dividing multimedia file source data into data segments according to the time intervals;
[0050] S102: extracting identification features of each data segment, and using the identification features as type features of the data segment;
[0051] S103: assigning progress stamp data to the identification feature of each data segment based on the total duration of the multimedia file;
[0052] S104: Obtain identification features of each data segment, sort them according to progress stamps, obtain a multimedia file identification feature set, and output it.
[0053] In the above process, multimedia file source data, such as video files, audio files and their total duration, is obtained; a time interval Δt is set, and the multimedia file source data is divided into data segments according to the time interval Δt; for example, if the time interval is set to 10s, the file is divided into N = ceil (T_total / Δt) data segments according to Δt (the last segment may be less than Δt). Example:
[0054] The total length of the video is 55 seconds, and Δt = 10 seconds. It is divided into 6 segments, with the first 5 segments of 10 seconds each and the last segment of 5 seconds. Output the segmented data block list [Segment_1, Segment_2, ..., Segment_N].
[0055] Get each data segment Segment_i and call the corresponding feature extraction algorithm according to the data segment type:
[0056] In the video segment: frame-level features (such as key frames, shot switching frequency), audio features (such as background music type), subtitle text, etc. are extracted.
[0057] In the audio segment: extract spectral features (such as Mel-frequency cepstral coefficients MFCC), rhythm features (BPM), human voice / background sound ratio, etc.
[0058] Among them, in the image segment: extract color histogram, object detection label (such as face, scene classification).
[0059] Integrate the extracted features into structured data, such as JSON format, as the type feature Feature_i of the segment. Example:
[0060] Features of the first segment of the video (0-10 seconds):
[0061] {"frame_count":240,"key_frames":[0,5,10],"audio_genre":"pop"}.
[0062] Output a list of type features for each segment [Feature_1,Feature_2,...,Feature_N].
[0063] Get the type feature Feature_i of each segment and the total duration T_total of the multimedia file.
[0064] The progress stamp Timestamp_i of each segment is calculated to indicate the time position of the segment in the file:
[0065] Timestamp_i = (i-1)*Δt (the first segment is 0 seconds, the second segment is 10 seconds, and so on).
[0066] Add Timestamp_i to Feature_i to form a feature with a timestamp:
[0067] Feature_with_time_i. Example:
[0068] The progress stamp of the third segment is Timestamp_3 = 20 seconds, and the feature update is:
[0069] {
[0070] "frame_count":300,"key_frames":[20,25],"timestamp":20}.
[0071] Output a list of features with progress stamps [Feature_with_time_1,...,Feature_with_time_N].
[0072] Sort the feature list in ascending order according to the timestamp field to ensure the correct time sequence; encapsulate the sorted feature list into a multimedia file recognition feature set Feature_Set. Example: Sorted feature set:
[0073] [{"frame_count":240,"timestamp":0},
[0074] {"frame_count":250,"timestamp":10},
[0075] {"frame_count":300,"timestamp":20}, ... ]
[0078] Output the structured recognition feature set Feature_Set for subsequent use.
[0079] S200: Obtain a user behavior feature set, predict the target user of the multimedia file based on the identification feature set of the multimedia file, and trigger a multimedia file publishing request.
[0080] In this embodiment, a user behavior feature set is obtained, and the target user of the multimedia file is predicted based on the identification feature set of the multimedia file, and a multimedia file publishing request is triggered. The specific process is as follows:
[0081] S201: Constructing a user behavior feature set and outputting user preference types;
[0082] S202: Calculate the matching degree between the multimedia file identification feature set and the user preference type, obtain the target user of the multimedia file, and predict the spread of the multimedia file;
[0083] S203: Obtain prediction results and trigger a multimedia file publishing request.
[0084] In the above process, a user behavior feature set is constructed and the user preference type is output; user behavior data is obtained and converted into behavior features to construct a user behavior feature set. Based on the user behavior feature set, user preferences are identified and the user preference type is output. User behavior data includes viewing history, likes, comments, shares, duration of stay, complete play rate and user profile; user profile includes age, gender, region, and interest tags. User behavior data is quantified, for example:
[0085] Watch Time: Calculates the average viewing time of users for a certain type of content (such as technology or entertainment).
[0086] Interaction intensity: User activity is calculated based on the weighted number of likes, comments, and shares.
[0087] Time decay: Assign time weight to historical behaviors (e.g., the weight of behaviors in the past week is greater than that of behaviors a month ago).
[0088] Example:
[0089] Construct user behavior feature vectors, such as viewing time percentage (e.g., technology videos account for 30%, entertainment videos account for 50%), interactive behavior statistics (number of likes, comments, and shares), and time decay factors (recent behavior is given more weight).
[0090] Example feature vector:
[0091] [Technology_Viewing Time=0.3, Entertainment_Viewing Time=0.5, Technology_Interactions=2, Entertainment_Interactions=10]
[0092] Normalize the features to avoid the influence of different dimensions on clustering effect.
[0093] Based on a clustering algorithm, user behavior features are mapped to interest tags. An optimal K is set. When K = 3, the silhouette coefficient is highest (0.7), indicating that users can be divided into three categories. K user feature vectors are randomly selected as initial centers. The distance (e.g., Euclidean distance) from each center is calculated, and the user is assigned to the closest cluster. The cluster center is updated to the mean vector of all users in the cluster, and the iteration is repeated until convergence.
[0094] For each cluster, calculate the mean of its eigenvector to obtain the "typical characteristics" of the cluster. Example:
[0095] Cluster 1: [Technology_Viewing Time = 0.6, Entertainment_Viewing Time = 0.1, Technology_Interactions = 5, Entertainment_Interactions = 1]
[0096] Cluster 2: [Technology_Viewing Time = 0.1, Entertainment_Viewing Time = 0.7, Technology_Interactions = 1, Entertainment_Interactions = 8]
[0097] Generate interpretable labels based on the typical features of the cluster. Example:
[0098] Cluster 1 → Label “Technology Enthusiast” (high proportion of technology-related behaviors).
[0099] Cluster 2 → Label “Entertainment Expert” (high proportion of entertainment-related behaviors).
[0100] Assign each user to the cluster to which they belong and label the corresponding interest tags. Example:
[0101] User A belongs to cluster 1 and has the label “Technology Enthusiast”.
[0102] For each user, calculate the similarity between their preferred type and the type of multimedia file features; example:
[0103] Multimedia file feature type: ["technology","music"].
[0104] User A's preference type: ["Technology", "Education"], similarity = 1 (matching "Technology") / 2 (total number of preferences) = 0.5.
[0105] User B's preference type: ["Entertainment", "Life"], similarity = 0.
[0106] Target user screening: set a matching threshold (e.g., θ = 0.3) and select users with matching degrees higher than the threshold as target users. Example:
[0107] User A's matching degree is 0.5>0.3, making it the target user; user B's matching degree is 0, so it is filtered out.
[0108] Predict the scope of dissemination based on target user scale, user activity, and content popularity.
[0109] Output the target user list and propagation prediction results:
[0110]
[0111] After obtaining the target user list and the propagation prediction results, a multimedia file publishing request is triggered.
[0112] S300: Setting a warning threshold, matching identification feature sets of multiple multimedia files and obtaining matching values, and executing a multimedia file publishing request after comparing the warning threshold and the matching values.
[0113] In this embodiment, a warning threshold is set, and identification feature sets of multiple multimedia files are matched and matching values are obtained. After comparing the warning threshold and matching values, a multimedia file publishing request is executed. The specific process is as follows:
[0114] S301: Setting a warning threshold and calculating a matching value between the current file identification feature set and the existing file identification feature set;
[0115] S302: Compare the warning threshold with the matching value and output the comparison result;
[0116] When the matching value is less than or equal to the warning threshold, the current file is identified as a normal file and the file publishing operation continues;
[0117] When the matching value is greater than the warning threshold, the current file is identified as an abnormal file and the file publishing operation is terminated.
[0118] In the above process, a current file identification feature set and an existing file identification feature library are obtained, wherein the existing file identification feature library stores feature data of historically released files.
[0119] Configure an alert threshold (e.g., θ = 0.8); calculate feature similarity between the current file and each file in the existing file library. Common methods include: hash fingerprint matching, such as calculating the Hamming distance of images / videos based on perceptual hashing (pHash) (the smaller the distance, the more similar they are). Text similarity, such as TF-IDF or cosine similarity (for text features such as titles and descriptions). Comprehensive similarity, combining multi-dimensional feature weighting calculations (e.g., similarity = 0.6 × visual similarity + 0.4 × text similarity). Example:
[0120] The pHash Hamming distance between the current file and the existing file A is 5 (total number of bits is 64), and the similarity = 1-(5 / 64)≈0.92.
[0121] The cosine similarity between the current file and the existing file B is 0.7.
[0122] Comprehensive similarity = 0.6×0.92+0.4×0.7=0.832.
[0123] Extract the maximum match value. From the matching values of all existing files, the maximum value is taken as the final match value for the current file. The final match value reflects the highest similarity with the previous files. For example, if the matching values of the current file and the existing file library are [0.832, 0.65, 0.41] respectively, the final match value is 0.832.
[0124] Output the final matching value and warning threshold of the current file:
[0125]
[0126] If the matching value is less than or equal to the warning threshold, the current file is determined to be a normal file, which means there is no highly similar historical file.
[0127] If the matching value is greater than the warning threshold, the current file is determined to be an abnormal file, which means it may contain duplicate content.
[0128] For normal files: continue with the file publishing operation.
[0129] For abnormal files, the release of the file will be terminated and manual review will be triggered.
[0130] Example:
[0131] The current file matching value is 0.832>0.8, which is considered an abnormal file and the release is aborted.
[0132] If the matching value is 0.75≤0.8, it is determined to be a normal file and is allowed to be published.
[0133] Output comparison results and release decisions:
[0134]
[0135] In the present invention, first, multimedia file source data is obtained, data segments are divided, identification features of each data segment are extracted, and a progress stamp is assigned to the identification features, and a multimedia file identification feature set is output; then, a user behavior feature set is obtained, and the target user of the multimedia file is predicted based on the identification feature set of the multimedia file, and a multimedia file publishing request is triggered; finally, a warning threshold is set, the identification feature sets of multiple multimedia files are matched and a matching value is obtained, and after comparing the warning threshold and the matching value, the multimedia file publishing request is executed; by performing feature recognition based on the multimedia file and predicting the target user of the multimedia file based on the user's behavior characteristics, the accuracy of multimedia file push is improved.
[0136] Corresponding to the aforementioned embodiment of the method for predicting multimedia files, the present application also provides an embodiment of an apparatus for predicting multimedia files.
[0137] Figure 5 FIG. 1 is a block diagram of a multimedia file prediction device according to an exemplary embodiment. Figure 5 The device may include: an identification feature set acquisition module 401, a target user prediction module 402, and a publishing request execution module 403; wherein:
[0138] The identification feature set acquisition module 401 is used to acquire multimedia file source data, divide the data into segments, extract identification features of each data segment, assign progress stamps to the identification features, and output the multimedia file identification feature set;
[0139] The target user prediction module 402 is used to obtain a user behavior feature set, predict the target user of the multimedia file based on the identification feature set of the multimedia file, and trigger a multimedia file publishing request;
[0140] The publishing request execution module 403 is used to set a warning threshold, match the identification feature sets of multiple multimedia files and obtain a matching value, and execute the multimedia file publishing request after comparing the warning threshold and the matching value.
[0141] In this embodiment, the identification feature set acquisition module 401 acquires multimedia file source data, divides the data into segments, extracts identification features of each data segment, assigns a progress stamp to the identification features, and outputs a multimedia file identification feature set; the target user prediction module 402 acquires a user behavior feature set, predicts the target user of the multimedia file based on the identification feature set of the multimedia file, and triggers a multimedia file publishing request; the publishing request execution module 403 establishes a warning threshold, matches the identification feature sets of multiple multimedia files and obtains a matching value, and executes the multimedia file publishing request after comparing the warning threshold and the matching value; by performing feature recognition based on the multimedia file and predicting the target user of the multimedia file based on the user's behavior characteristics, the accuracy of multimedia file push is improved.
[0142] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.
[0143] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.
[0144] Accordingly, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned multimedia file prediction method. Figure 6 As shown in FIG. 1 , a hardware structure diagram of a device having data processing capability in which a multimedia file prediction device provided by an embodiment of the present invention is located, except for Figure 6 In addition to the processor, memory, and network interface shown, any device with data processing capabilities in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.
[0145] Accordingly, the present application also provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the prediction method for multimedia files as described above. The computer-readable storage medium may be an internal storage unit of any device with data processing capabilities as described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium may also be an external storage device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card (Flash Card), etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capabilities. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and may also be used to temporarily store data that has been output or is to be output.
[0146] Those skilled in the art will readily conceive of other embodiments of the present application after considering the specification and practicing the contents disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed in this application.
[0147] It will be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.
Claims
1. A method for predicting multimedia files, characterized in that: The steps include: Obtain multimedia file source data, divide the data into segments, extract identification features of each data segment, assign progress stamps to the identification features, and output a multimedia file identification feature set; Obtaining a user behavior feature set, predicting the target user of the multimedia file based on the identification feature set of the multimedia file, and triggering a multimedia file publishing request; Set up an alert threshold, match the identification feature set of multiple multimedia files and obtain the matching value, compare the alert threshold and the matching value, and then execute the multimedia file publishing request.
2. The multimedia file prediction method according to claim 1, wherein: In the steps of obtaining multimedia file source data, dividing data segments, extracting identification features of each data segment, assigning progress stamps to the identification features, and outputting a multimedia file identification feature set: Setting a time interval and dividing the multimedia file source data into data segments according to the time interval; Extracting identification features of each data segment, and using the identification features as type features of the data segment; Output multimedia file recognition feature set.
3. The multimedia file prediction method according to claim 2, wherein: After extracting the identification features of each data segment and using the identification features as the type features of the data segment: According to the total duration of the multimedia file, progress stamp data is assigned to the identification feature of each data segment.
4. The multimedia file prediction method according to claim 3, wherein: In the step of outputting the multimedia file recognition feature set: Obtain the identification features of each data segment, sort them according to the progress stamp, obtain the multimedia file identification feature set, and output it.
5. The multimedia file prediction method according to claim 1, wherein: In the steps of obtaining a user behavior feature set, predicting the target user of the multimedia file based on the identification feature set of the multimedia file, and triggering a multimedia file publishing request: Construct user behavior feature sets and output user preference types; Calculate the matching degree between the multimedia file identification feature set and the user preference type, obtain the target users of the multimedia file, and predict the spread of the multimedia file; Get the prediction results and trigger a multimedia file publishing request.
6. The multimedia file prediction method according to claim 5, wherein: In the steps of constructing the user behavior feature set and outputting the user preference type: Obtain user behavior data, convert user behavior into behavior features, build a user behavior feature set, identify user preferences based on the user behavior feature set, and output user preference types.
7. The multimedia file prediction method according to claim 1, wherein: In the steps of setting an alert threshold, matching identification feature sets of multiple multimedia files and obtaining matching values, and then comparing the alert threshold and matching values to execute a multimedia file publishing request: Set up warning thresholds and calculate the matching value between the current file identification feature set and the existing file identification feature set; Compare the warning threshold with the matching value and output the comparison result.
8. The multimedia file prediction method according to claim 7, wherein: In the step of comparing the warning threshold with the matching value and outputting the comparison result: When the matching value is less than or equal to the warning threshold, the current file is identified as a normal file and the file publishing operation continues.
9. The multimedia file prediction method according to claim 8, wherein: In the step of comparing the warning threshold with the matching value and outputting the comparison result: When the matching value is greater than the warning threshold, the current file is identified as an abnormal file and the file publishing operation is terminated.
10. A multimedia file prediction device, applied to the multimedia file prediction method according to claim 1, characterized in that: It includes an identification feature set acquisition module, a target user prediction module, and a release request execution module; among which: The identification feature set acquisition module is used to obtain multimedia file source data, divide the data into segments, extract identification features of each data segment, assign progress stamps to the identification features, and output the multimedia file identification feature set; The target user prediction module is used to obtain a user behavior feature set, predict the target user of the multimedia file based on the identification feature set of the multimedia file, and trigger a multimedia file publishing request; The publishing request execution module is used to set an alert threshold, match the identification feature sets of multiple multimedia files and obtain a matching value, and execute the multimedia file publishing request after comparing the alert threshold and the matching value.