A push method, device and equipment of multimedia information and storage medium

Abstract

Embodiments of the present application disclose a multimedia information pushing method and device, equipment and a storage medium, which are used for maintaining healthy pushing of multimedia information. The method comprises the following steps: obtaining to-be-processed multimedia information, respectively calculating the similarity between the to-be-processed multimedia information and the data in a multimedia risk database, a multimedia negative feedback database and a multimedia playing database, obtaining a first similarity set, a second similarity set and a third similarity set, respectively obtaining initial risk values corresponding to the multimedia risk database and the multimedia negative feedback database, calculating a target risk value corresponding to the to-be-processed multimedia information based on the initial risk values, the first similarity set, the second similarity set and the third similarity set, determining an audit result of the to-be-processed multimedia information based on the target risk value and a multimedia information audit condition, and pushing the to-be-processed multimedia information to a multimedia information receivable object if the audit result is pushable.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, device and storage medium for pushing multimedia information. Background Technology

[0002] With the rapid development of internet technology, many advertisers choose to place ads in different ad slots so that browsing users can notice the ads and thus achieve a promotional effect.

[0003] Advertising placement typically only tracks ad exposure time and click count, without analyzing ad risks or security issues. This could potentially lead to ads being placed in unsuitable locations, resulting in negative consequences.

[0004] In the current internet advertising market, reducing the risk of advertising generally involves manually reviewing ads after they have been placed, using a lot of human resources. Only when high-risk ads are identified are the ads controlled and dealt with.

[0005] However, with the increasing complexity and diversity of advertising types and the varying standards of manual review, errors are prone to occur during manual review. In addition, some advertising materials are quite special, and without a good reference value, manual review relies solely on the different interpretations of the review standards by the reviewers. This can lead to situations where, on the one hand, advertising platforms cannot effectively control risky ads, and on the other hand, advertisers, in pursuit of promotional effects, may imitate and place risky ads, further exacerbating the volume of views for risky ads and hindering the healthy placement of advertisements. Summary of the Invention

[0006] This application provides a method, apparatus, device, and storage medium for pushing multimedia information. It is used to construct a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database as multi-dimensional reference information to more comprehensively consider the target risk value corresponding to the multimedia information to be processed. This enables automated review and push of multimedia information based on the target risk value, which can not only reduce the possibility of human error in review, but also reduce the risk of the multimedia information push environment, thereby maintaining a healthy environment for multimedia information push.

[0007] This application provides a method for pushing multimedia information, including:

[0008] Acquire multimedia information to be processed;

[0009] The similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database is calculated respectively to obtain the first similarity set, the second similarity set, and the third similarity set;

[0010] Obtain the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database respectively;

[0011] Based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set, the target risk value corresponding to the multimedia information to be processed is calculated;

[0012] Based on the target risk value and the multimedia information review conditions, the review results of the multimedia information to be processed are determined.

[0013] If the review result indicates that the multimedia information to be processed can be pushed, then the multimedia information to be processed will be pushed to the recipients of the multimedia information.

[0014] Another aspect of this application provides a multimedia information push device, including:

[0015] The acquisition unit is used to acquire multimedia information to be processed.

[0016] The processing unit is used to calculate the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database and the multimedia playback database, respectively, to obtain a first similarity set, a second similarity set and a third similarity set;

[0017] The acquisition unit is also used to acquire the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, respectively.

[0018] The processing unit is also used to calculate the target risk value corresponding to the multimedia information to be processed based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set;

[0019] The determination unit is used to determine the review result of the multimedia information to be processed based on the target risk value and the multimedia information review conditions.

[0020] The processing unit is also used to push the multimedia information to be processed to the multimedia information receiving object if the review result is pushable.

[0021] In one possible design, in another implementation of the embodiments of this application,

[0022] The acquisition unit is also used to acquire multimedia risk information, which includes multimedia risk reporting data and multimedia risk capture data.

[0023] The processing unit is also used to store the multimedia risk information to the multimedia risk database if the multimedia risk information matches the multimedia risk condition.

[0024] In one possible design, in another implementation of the embodiments of this application, the acquisition unit may specifically be used for:

[0025] The reported multimedia data is traversed, and the multimedia data marked with risk is used as the multimedia risk reporting data.

[0026] Regularly crawl data from multimedia information publishing platforms to obtain multimedia risk capture data;

[0027] Specifically, the processing unit can be used to: if multimedia risk reporting data or multimedia risk capture data contains multimedia risk keywords, then store the multimedia risk reporting data or multimedia risk capture data into the multimedia risk database.

[0028] In one possible design, in another implementation of the embodiments of this application,

[0029] The acquisition unit is also used to receive raw negative feedback data corresponding to multimedia information, wherein the raw negative feedback data includes historical exposure data of multimedia information;

[0030] The acquisition unit is also used to acquire the historical playback log corresponding to the multimedia information, wherein the historical playback log includes at least the playback position information and the group object information corresponding to the multimedia information;

[0031] The processing unit is also used to integrate historical exposure data, playback location information, and group object information, and store the integrated multimedia negative feedback data in the multimedia negative feedback database.

[0032] In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

[0033] Historical exposure data is integrated with playback location information and group object information to obtain multimedia scene data;

[0034] The original negative feedback data and multimedia scene data are recorded as multimedia negative feedback data and stored in the multimedia negative feedback database.

[0035] In one possible design, in another implementation of the embodiments of this application,

[0036] The processing unit is also used to crawl data based on the crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy, wherein the crawling strategy includes one or more different crawling conditions.

[0037] The processing unit is also used to store the multimedia crawled data into the multimedia playback database.

[0038] In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

[0039] Based on time-time crawling criteria, data is crawled separately according to different time periods to obtain multimedia crawled data corresponding to each time period; and / or

[0040] Based on the group object information crawling conditions, data is crawled separately according to different group attributes to obtain multimedia crawled data corresponding to each group attribute; and / or

[0041] Based on geographic distribution information crawling conditions, data is crawled separately according to different geographic attributes to obtain multimedia crawled data corresponding to each geographic attribute; and / or

[0042] Based on the device coverage crawling condition, data is crawled separately according to different device coverage, and multimedia crawled data corresponding to each device coverage is obtained.

[0043] In one possible design, in another implementation of the embodiments of this application, the acquisition unit may specifically be used for:

[0044] Obtain the exposure, clicks, and cost corresponding to each multimedia risk information in the multimedia risk database;

[0045] Based on exposure, clicks, and consumption, determine the initial risk value corresponding to multimedia risk information;

[0046] Obtain the negative feedback quantity and negative feedback height corresponding to each multimedia negative feedback data in the multimedia negative feedback database;

[0047] Based on the amount and height of negative feedback, the initial risk value corresponding to the multimedia negative feedback data is determined.

[0048] In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

[0049] Calculate the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database to obtain the first similarity set;

[0050] Calculate the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database to obtain the second similarity set;

[0051] Calculate the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database to obtain the third similarity set.

[0052] In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

[0053] The first risk value is obtained by summing and averaging the first similarity scores that are greater than the similarity threshold and the initial risk values ​​corresponding to the first similarity scores.

[0054] The second risk value is obtained by summing and averaging the second similarity values ​​that are greater than the similarity threshold and the initial risk values ​​corresponding to the second similarity values.

[0055] Risk reduction weights are obtained based on third-degree similarity and used as risk factors for the multimedia information to be processed.

[0056] The target risk value is determined based on the first risk value, the second risk value, and the risk factors.

[0057] In one possible design, in another implementation of the embodiments of this application, the processing unit may specifically be used for:

[0058] Obtain the sum between the first risk value and the second risk value;

[0059] Calculate the product between the sum and the risk factor, and use the product as the target risk value.

[0060] In one possible design, in another implementation of the embodiments of this application, the determining unit may specifically be used for:

[0061] If the target risk value is greater than the risk threshold, or the number of times the risk audit keywords are hit is greater than the quantity threshold, then the audit result corresponding to the multimedia information to be processed is not pushable.

[0062] If the target risk value is less than or equal to the risk threshold, and the number of times the risk audit keywords are hit is less than or equal to the quantity threshold, then the audit result corresponding to the multimedia information to be processed can be pushed.

[0063] This application also provides a computer device, including: a memory, a processor, and a bus system;

[0064] The memory is used to store programs;

[0065] The processor implements the methods described above when executing a program in memory;

[0066] Bus systems are used to connect memory and processor to enable communication between them.

[0067] Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described above.

[0068] As can be seen from the above technical solutions, the embodiments of this application have the following beneficial effects:

[0069] By acquiring the multimedia information to be processed, the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database is calculated to obtain a first similarity set, a second similarity set, and a third similarity set. The initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database are also obtained. Then, based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set, the target risk value corresponding to the multimedia information to be processed can be calculated. Then, based on the target risk value and the multimedia information review conditions, the review result of the multimedia information to be processed can be determined. When the review result is pushable, the multimedia information to be processed is pushed to the multimedia information receiving objects. By constructing a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database as multi-dimensional reference information, the first, second, and third similarity sets between multimedia information to be processed can be calculated. Simultaneously, by considering the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, the target risk value corresponding to the multimedia information to be processed can be more comprehensively assessed using the initial risk value, the first similarity set, the second similarity set, and the third similarity set. This enables automated review and push of multimedia information based on the target risk value. Furthermore, by combining manual review with the target risk value to perform automated review and push of multimedia information, not only can the possibility of human error in review be reduced, but the risks of the multimedia information push environment can also be lowered, thereby maintaining a healthy environment for multimedia information push. Attached Figure Description

[0070] Figure 1 This is a schematic diagram of the architecture of a data sharing system in an embodiment of this application;

[0071] Figure 2 This is a flowchart of one embodiment of the multimedia information push method in this application;

[0072] Figure 3 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0073] Figure 4 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0074] Figure 5 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0075] Figure 6 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0076] Figure 7 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0077] Figure 8 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0078] Figure 9 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0079] Figure 10 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0080] Figure 11 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0081] Figure 12 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0082] Figure 13 This is a flowchart of another embodiment of the multimedia information push method in this application;

[0083] Figure 14 This is a schematic diagram illustrating the principle of a multimedia information push method in an embodiment of this application;

[0084] Figure 15 This is a schematic diagram illustrating the principle of constructing a risk database for a multimedia information push method in this application embodiment;

[0085] Figure 16 This is a schematic diagram illustrating the principle of constructing a risk database for another multimedia information push method in this application embodiment;

[0086] Figure 17 This is a schematic diagram illustrating the principle of constructing a risk database for another multimedia information push method in this application embodiment;

[0087] Figure 18(a) is a schematic diagram of a risk assessment display page of the multimedia information push method in an embodiment of this application;

[0088] Figure 18(b) is a schematic diagram of another risk assessment display page of the multimedia information push method in the embodiments of this application;

[0089] Figure 18(c) is another risk assessment display page of the multimedia information push method in the embodiments of this application;

[0090] Figure 19 This is a schematic diagram of one embodiment of the multimedia information push device in this application;

[0091] Figure 20 This is a schematic diagram of one embodiment of the computer device described in this application. Detailed Implementation

[0092] This application provides a method, apparatus, device, and storage medium for pushing multimedia information. It is used to construct a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database as multi-dimensional reference information to more comprehensively consider the target risk value corresponding to the multimedia information to be processed. This enables automated review and push of multimedia information based on the target risk value, which can not only reduce the possibility of human error in review, but also reduce the risk of the multimedia information push environment, thereby maintaining a healthy environment for multimedia information push.

[0093] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0094] It is understood that in the specific implementation of this application, data related to multimedia information to be processed and target risk values ​​are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0095] Understandably, the multimedia information push method disclosed in this application also involves Artificial Intelligence (AI) technology, which will be further introduced below. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling them to have the functions of perception, reasoning, and decision-making.

[0096] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0097] Secondly, Natural Language Processing (NLP) is an important field within computer science and artificial intelligence. It studies the theories and methods that enable effective communication between humans and computers using natural language. NLP is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field involves natural language—the language people use in daily life—and thus it has a close connection with linguistic research. NLP technologies typically include text processing, semantic understanding, machine translation, question answering, and knowledge graphs.

[0098] Secondly, Machine Learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0099] With the research and advancement of artificial intelligence (AI) technology, AI is being studied and applied in various fields, such as smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, autonomous driving, drones, robots, smart healthcare, and smart customer service. It is believed that with the development of technology, AI will be applied in more fields and play an increasingly important role.

[0100] It should be understood that the multimedia information push method provided in this application can be applied to various scenarios, including but not limited to artificial intelligence, cloud technology, maps, and smart transportation, to complete the delivery or push of multimedia by obtaining the risk value of multimedia information, and can be applied to scenarios such as advertising, film and television release or promotion, game and animation release or promotion, etc.

[0101] See Figure 1 , Figure 1 This is an optional structural diagram of the data sharing system 100 provided in this embodiment of the invention applied to a blockchain system. It consists of multiple nodes 101 (any form of computing device in the network, such as a server or user terminal) and clients. The nodes 101 form a peer-to-peer (P2P) network. The P2P protocol is an application layer protocol running on top of the Transmission Control Protocol (TCP). In the system, any machine, such as a server or terminal, can join and become a node 101. A node 101 includes a hardware layer, a middleware layer, an operating system layer, and an application layer.

[0102] It is understandable that in real-world scenarios, a wider variety of terminal devices can participate in the data processing as node 101. These terminal devices include, but are not limited to, mobile phones, computers, smart voice interaction devices, smart home appliances, and in-vehicle terminals. The specific number and types depend on the actual scenario and are not limited here. Additionally, in real-world scenarios, multiple servers can also participate as node 101. The number of servers depends on the actual scenario and is not limited here.

[0103] It should be noted that in this embodiment, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Terminal devices and servers can be directly or indirectly connected via wired or wireless communication, and terminal devices and servers can be connected to form a blockchain network; this application does not impose any limitations on this.

[0104] Understandably, see Figure 1 In the data sharing system 100, each node 101 stores an identical blockchain. The blockchain consists of multiple blocks, each containing the hash value of the transaction records stored in that block (i.e., a batch of multimedia information and the target risk value corresponding to each multimedia information), and the hash value of the previous block. Blocks are linked together through their hash values ​​to form the blockchain. Additionally, blocks may include a timestamp of block generation and the timestamp of each transaction object's bid for the resource. Essentially, a blockchain is a decentralized database, a chain of data blocks linked using cryptographic methods. Each data block contains relevant information used to verify the validity of the information (anti-counterfeiting) and to generate the next block. A blockchain can include an underlying platform, a platform product service layer, and an application service layer.

[0105] The underlying blockchain platform can include processing modules such as user management, basic services, smart contracts, and operation management. The user management module is responsible for managing the identity information of all blockchain participants (such as the original creators or publishers of multimedia information, the audience of multimedia information, and those subject to risk assessment of multimedia information). This includes maintaining public and private key generation (account management), key management, and maintaining the correspondence between user identities and blockchain addresses (access management). The basic services module is deployed on all blockchain node devices (such as the terminal devices of the original creators or publishers of multimedia information). It verifies the validity of business requests (such as the creation or publication of multimedia information) and records valid requests in storage after consensus is reached. For a new business request, the basic services first perform interface adaptation parsing and authentication (interface adaptation), then encrypt the business information using a consensus algorithm (consensus management), and finally transmit the encrypted information completely and consistently to the relevant department. The shared ledger (network communication) records and stores data. The smart contract module is responsible for registering and issuing contracts (such as multimedia information management contracts), as well as triggering and executing contracts. For example, when a resource transfer request corresponding to a resource to be traded is received, developers can define the contract logic (such as a multimedia information management contract) using a programming language, publish it to the blockchain (multimedia information management contract registration), and execute it by calling a key or other events according to the logic of the resource management contract terms to complete the resource management contract logic. It also provides the function of upgrading and canceling resource management contracts. The operation management module is mainly responsible for the deployment, configuration modification, resource management contract settings, cloud adaptation, and real-time visualization output of the status of the transaction task of the resource to be traded during the release process. For example, the bidding status of each object to be traded.

[0106] The platform's product service layer provides basic capabilities and implementation frameworks for typical applications. Developers can leverage these basic capabilities, along with the characteristics of specific business processes (such as multimedia information push services), to implement the blockchain logic for those processes. The application service layer provides blockchain-based application services to business participants (such as the original creators or publishers of multimedia information, the audience of multimedia information, and those responsible for risk assessment of multimedia information).

[0107] Based on the above introduction, the method for pushing multimedia information in this application will be described below. Please refer to [link / reference]. Figure 2 One embodiment of the multimedia information push method in this application includes:

[0108] In step S101, the multimedia information to be processed is obtained;

[0109] In this embodiment of the application, when delivering multimedia information to be processed to the audience of each multimedia information, i.e., the recipient of multimedia information, the multimedia information to be processed can be obtained first, so that subsequent analysis and processing can be performed based on the obtained multimedia information to be processed, thereby better maintaining the healthy delivery of multimedia information.

[0110] Specifically, the multimedia information to be processed can be advertising images and copy created by advertisers, or works created by self-media authors or bloggers, such as text and image works or audio and video works, or promotional images, articles or videos produced by film and television promoters, or other multimedia information, without specific restrictions here.

[0111] Specifically, before the multimedia information to be processed is delivered, for example, such as Figure 14 As shown, after the advertiser has created the advertising images and copy, they can first upload these materials—the multimedia information to be processed—to a platform such as... Figure 14 The advertising risk analysis engine in the illustrated system enables the system to acquire the multimedia information to be processed. This allows the system to analyze and process the acquired multimedia information to obtain its corresponding risk value, thus accurately determining whether the multimedia information can be delivered. Specifically, acquiring the multimedia information to be processed can be achieved through a risk assessment display page as shown in Figure 18(a), where the system receives the multimedia information, such as an advertising image uploaded by an advertiser.

[0112] In step S102, the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database and the multimedia playback database is calculated respectively to obtain the first similarity set, the second similarity set and the third similarity set;

[0113] In this embodiment of the application, in order to better perform risk analysis based on the acquired multimedia information to be processed, a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database can be constructed first to calculate the first similarity set, the second similarity set, and the third similarity set between the multimedia information to be processed as multi-dimensional reference information. This allows for a more comprehensive consideration of the target risk value corresponding to the multimedia information to be processed based on the first similarity set, the second similarity set, and the third similarity set.

[0114] Specifically, before analyzing and processing the acquired multimedia information to be processed, for example, based on such Figure 14The advertising risk analysis engine in the system shown can first analyze and process multimedia information, such as an advertising image uploaded by an advertiser, before performing such analysis. Figure 14 The advertising risk analysis engine in the system shown constructs a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database. The multimedia risk database can be constructed (e.g., Figure 14 (e.g., advertising risk database, multimedia negative feedback database) Figure 14 The database of negative feedback from the intended advertising audience) and the multimedia playback database (e.g., Figure 14 (The database of advertising media playback as shown).

[0115] Furthermore, a multimedia risk database (e.g., Figure 14 The advertiser risk database (as indicated) could specifically be a database containing multimedia risk reporting data (e.g., Figure 15 Data with risk labels as shown) and multimedia risk capture data (e.g., Figure 15 Multimedia risk information (such as data captured from an online advertising database) can be obtained. Then, it can be determined whether the obtained multimedia risk information matches multimedia risk conditions (e.g., matching preset risk keywords or risk markers). If the multimedia risk information matches multimedia risk conditions, it can be understood that the multimedia risk information may pose a risk. The multimedia risk information can then be stored in a multimedia risk database to construct a multimedia risk database.

[0116] Furthermore, a multimedia negative feedback database is constructed (e.g., Figure 14 The intended negative feedback database for advertising audiences can specifically refer to audiences receiving multimedia information (such as...). Figure 15 The intended advertising audience), uploaded raw negative feedback data about multimedia information (e.g., reasons for and descriptions of an advertisement), and can also be viewed through multimedia information audiences (such as... Figure 15 The system generates a multimedia information database by receiving an ad request from the target audience (the intended audience), generating a historical playback log (e.g., ad media location information). Then, the original negative feedback data and the historical playback log are filtered and integrated, and the integrated multimedia negative feedback data is stored in a multimedia negative feedback database to construct the multimedia negative feedback database.

[0117] Furthermore, a multimedia playback database is constructed (e.g., Figure 14The advertised media playback database (as shown) can be constructed by crawling data based on a preset crawling strategy consisting of one or more different crawling conditions to obtain multimedia crawling data corresponding to the crawling strategy (for example, time period crawling conditions, crawling data according to different time periods, such as 10:00 AM to 12:00 PM and 2:00 PM to 4:00 PM respectively, to obtain the advertised crawling data corresponding to 10:00 AM to 12:00 PM and the advertised crawling data corresponding to 2:00 PM to 4:00 PM). Then, the multimedia crawled data can be stored in the multimedia playback database to construct the multimedia playback database.

[0118] Furthermore, after constructing the multimedia risk database, multimedia negative feedback database, and multimedia playback database, the multimedia information to be processed and the multimedia risk database (e.g., Figure 14 The first similarity between each multimedia risk information in the indicated advertising risk database is used to obtain the first similarity set. The specific method for calculating the similarity can be cosine similarity, Hamming distance or hash algorithm, etc., and other similarity calculation methods can also be used. No specific restrictions are imposed here.

[0119] Similarly, the similarity calculation method described above can also be used to calculate the similarity between the multimedia information to be processed and the multimedia negative feedback database (e.g., Figure 14 The second similarity between each multimedia negative feedback data in the indicated advertising audience negative feedback database is used to obtain a second similarity set. Alternatively, the above similarity calculation method can be used to calculate the similarity between the multimedia information to be processed and the multimedia playback database (e.g., ...). Figure 14 The third similarity between each multimedia crawled data in the indicated advertising media playback database is used to obtain a third similarity set.

[0120] In step S103, the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database are obtained respectively;

[0121] In this embodiment of the application, in order to better estimate the risk value of the multimedia information to be processed, after constructing the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database, the data in the multimedia risk database and the multimedia negative feedback database can be initialized with risk values. This allows for a better prediction of the risk value of the multimedia information to be processed based on the initial risk values ​​and the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database.

[0122] The initial risk value corresponding to the multimedia risk database is used to indicate the basic risk score of each risk case in the multimedia risk database, representing the degree of risk of that multimedia risk information. Similarly, the initial risk value corresponding to the multimedia negative feedback database is used to indicate the basic risk score of each multimedia negative feedback data point in the multimedia negative feedback database, representing the degree of risk of that multimedia negative feedback data point.

[0123] Specifically, after constructing the multimedia risk database, multimedia negative feedback database, and multimedia playback database, a risk value analysis engine (such as...) can be built. Figure 14 The advertising risk analysis engine shown in the image allows for the subsequent construction of a risk value analysis engine (such as...). Figure 14 The advertising risk analysis engine (as shown) analyzes and processes the multimedia information to be processed. Specifically, it can initialize risk values ​​by analyzing data from the multimedia risk database and the multimedia negative feedback database.

[0124] Understandably, since the risk cases in the multimedia playback database represent materials deemed to have low risk values ​​after assessment by other risk systems in the market—essentially materials permitted for playback—it is unnecessary to calculate the initial risk value for these risk cases. However, the risk cases or negative feedback data in the multimedia risk database and the multimedia negative feedback database represent materials considered to have high risk values, thus requiring initial risk value calculation for these databases. Specifically, for the multimedia risk database, the initial risk value can be calculated based on the exposure, clicks, and costs associated with each multimedia risk case. Since the generation of risk cases is strongly correlated with exposure, clicks, and costs, the initial risk value for each multimedia risk information can be obtained from the multimedia risk database.

[0125] Furthermore, for a multimedia negative feedback database, since the multimedia negative feedback data in the database can carry negative feedback quantity (such as the number of user complaints) and negative feedback height (such as the severity of the complaint reasons), and the negative feedback quantity and negative feedback height are strongly correlated with the risk level of the multimedia negative feedback data, the initial risk value corresponding to the multimedia negative feedback data can be calculated based on the negative feedback quantity and negative feedback height.

[0126] In step S104, the target risk value corresponding to the multimedia information to be processed is calculated based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set.

[0127] In this embodiment of the application, after obtaining the initial risk value and calculating the first similarity set, the second similarity set, and the third similarity set, the target risk value corresponding to the multimedia information to be processed can be calculated based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set. The target risk value can then be used to accurately reflect or present the risk level of the multimedia information to be processed, so as to better judge or determine whether the multimedia information to be processed is suitable for healthy delivery.

[0128] Specifically, for multimedia risk databases (e.g., Figure 14 (As shown in the advertising risk database), calculate the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database. Then, select the first similarity greater than the similarity threshold from the first similarity. Sum the first similarity greater than the similarity threshold and the initial risk value corresponding to the first similarity and take the average. Use the calculated sum as the first risk value.

[0129] Similarly, for multimedia negative feedback databases (e.g., Figure 14 (As shown in the advertising audience negative feedback database), calculate the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database. Then, select the second similarity greater than the similarity threshold from the second similarity. Sum the second similarity greater than the similarity threshold and the initial risk value corresponding to the second similarity and take the average. Use the calculated sum as the second risk value.

[0130] Similarly, for multimedia playback databases (e.g., Figure 14 Using the advertising media playback database shown, calculate the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database. Then, obtain the risk reduction weight based on the third similarity and use the risk reduction weight as the risk factor of the multimedia information to be processed. Then, sum the first risk value, the second risk value and the risk factor, and use the calculated sum as the target risk value.

[0131] For example, as shown in Figure 18(a), suppose there is an ad image to be placed, and the calculated target risk value (such as the image risk value) is more than 60; as shown in Figure 18(b), suppose there is an ad image and ad copy to be placed, and the calculated target risk value (such as the image risk value is 70, the ad copy risk value is 0, etc.); and as shown in Figure 18(c), suppose there is another ad image and ad copy to be placed, and the calculated target risk value (such as the risk value is 30).

[0132] In step S105, the review result of the multimedia information to be processed is determined based on the target risk value and the multimedia information review conditions.

[0133] In this embodiment of the application, after obtaining the target risk value, it is possible to better judge or determine whether the multimedia information to be processed is suitable for healthy delivery based on the target risk value and in combination with the multimedia information review conditions, so as to obtain the review result of the multimedia information to be processed.

[0134] Specifically, after obtaining the target risk value, the review result of the multimedia information to be processed can be determined directly through the target risk value. For example, the target risk value can be compared with the risk threshold. If the target risk value is less than or equal to the risk threshold, it can be understood that the risk value of the multimedia information to be processed is not high and it is safe. The review result can then be "can be pushed". Conversely, if the target risk value is greater than the risk threshold, it can be understood that the risk value of the multimedia information to be processed is high and it is not safe. The review result can then be "cannot be pushed".

[0135] Furthermore, such as Figure 14 As shown, in order to more accurately determine whether the multimedia information to be processed is suitable for healthy delivery and to obtain the review result of the multimedia information to be processed, it is also possible to combine the multimedia information review conditions (such as... Figure 14 The intended advertising review rules, such as the number of risky keywords in the multimedia information to be processed being less than or equal to a certain threshold, are used to further obtain the review results of the multimedia information to be processed (e.g., Figure 14 (The ad review process is as shown). For example, if the target risk value is less than or equal to the risk threshold, and the number of risky keywords hit by the multimedia information to be processed is less than or equal to the quantity threshold, it can be understood that the risk value of the multimedia information to be processed is not high and it is safe. In this case, the review result is that it can be pushed. Conversely, if the target risk value is greater than the risk threshold, or the number of risky keywords hit by the multimedia information to be processed is greater than the quantity threshold, it can be understood that the risk value of the multimedia information to be processed is high and it is not safe. In this case, the review result is that it cannot be pushed.

[0136] For example, as shown in Figure 18(c), suppose an advertisement image and advertisement copy to be placed are calculated to have a target risk value (e.g., a risk value of 30). Suppose the risk threshold is 40, that is, the target risk value is less than the risk threshold. However, suppose the number of risk keywords hit by the multimedia information to be processed is greater than the number threshold (e.g., the advertisement copy in Figure 18(c) says "Don't spend 2xx yuan, you can get it for 0 (red envelope) yuan". For example, the word "red envelope" hits the risk keyword and there is a risk mark). It can be understood that the risk value of the advertisement image and advertisement copy to be placed is high and it is unsafe. The review result corresponding to the advertisement image and advertisement copy to be placed is that it cannot be pushed.

[0137] In step S106, if the review result is that the multimedia information to be processed can be pushed, the multimedia information to be processed will be pushed to the multimedia information receiving object.

[0138] In this embodiment of the application, after obtaining the review result of the multimedia information to be processed, if the review result is pushable, the multimedia information to be processed can be pushed to the multimedia information receiving object.

[0139] Specifically, such as Figure 14 As shown, after obtaining the review results of the multimedia information to be processed (such as...) Figure 14 After the ad review (as indicated), if the review result is pushable, the multimedia information to be processed can be pushed to the multimedia information receiving objects (e.g., Figure 14 If the indicated advertisement is approved for push, then the multimedia information to be processed, such as the advertisement to be delivered, can be played to the target audience.

[0140] In this embodiment, a method for pushing multimedia information is provided. Through the aforementioned approach, a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database are constructed as multi-dimensional reference information to calculate the first similarity set, the second similarity set, and the third similarity set between multimedia information to be processed. Simultaneously, by considering the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, the target risk value corresponding to the multimedia information to be processed can be more comprehensively considered using the initial risk value, the first similarity set, the second similarity set, and the third similarity set. This enables automated review and push of multimedia information based on the target risk value. Furthermore, automated review of multimedia information can also be performed on top of manual review, combined with the target risk value, to push multimedia information. This not only reduces the possibility of human error in review but also lowers the risk of the multimedia information push environment, thereby maintaining a healthy environment for multimedia information push.

[0141] Optionally, in the above Figure 2Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 3 As shown, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes:

[0142] In step S301, multimedia risk information is obtained, which includes multimedia risk reporting data and multimedia risk capture data.

[0143] In step S302, if the multimedia risk information matches the multimedia risk condition, the multimedia risk information is stored in the multimedia risk database.

[0144] In this embodiment of the application, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database and the multimedia playback database to obtain the first similarity set, the second similarity set and the third similarity set, the multimedia risk database can be constructed first. By obtaining multimedia risk information, the multimedia risk information can be matched with multimedia risk conditions. If the multimedia risk information matches the multimedia risk conditions, the multimedia risk information can be stored in the multimedia risk database.

[0145] Multimedia risk information includes multimedia risk reporting data (such as...) Figure 15 Advertising data reported by auditors or product operations staff (as indicated) and multimedia risk capture data (such as... Figure 15 (As shown, advertising data retrieved from an online advertising database via a scheduled task).

[0146] Specifically, when constructing a multimedia risk database, multimedia risk information can be obtained first. This can be achieved by analyzing reported multimedia data (such as...). Figure 15 The system iterates through advertising data reported by auditors or product operations staff (as indicated), and identifies risk markers (such as...) Figure 15 Multimedia data (such as risk labels, risk icons, or risk keywords) is used as multimedia risk reporting data. Simultaneously, a scheduled task periodically scrapes data from the multimedia information publishing platform to obtain multimedia risk data (such as...). Figure 15(As shown, advertising data is captured from the online advertising database through a scheduled task.) Then, the acquired multimedia risk reporting data or multimedia risk capture data can be matched with preset multimedia risk keywords. If a multimedia risk keyword is matched in the multimedia risk reporting data or multimedia risk capture data, the multimedia risk reporting data or multimedia risk capture data that matches the multimedia risk keyword can be used as a risk case and stored in the multimedia risk database.

[0147] Optionally, in the above Figure 3 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 4 As shown, step S301, which obtains multimedia risk information, includes steps S401 and S402; step S302 includes step S403.

[0148] In step S401, the reported multimedia data is traversed, and the multimedia data marked with risk is used as multimedia risk reporting data.

[0149] In step S402, data is periodically captured from the multimedia information publishing platform to obtain multimedia risk capture data;

[0150] In step S403, if multimedia risk keywords exist in the multimedia risk reporting data or multimedia risk capture data, the multimedia risk reporting data or multimedia risk capture data is stored in the multimedia risk database.

[0151] In this embodiment of the application, when constructing the multimedia risk database, the reported multimedia data can be traversed, and the multimedia data marked with risks can be used as multimedia risk reporting data. At the same time, a scheduled task can be used to periodically crawl data from the multimedia information publishing platform to obtain multimedia risk crawling data. This allows for a more comprehensive acquisition of multimedia risk cases, enabling better judgment or determination of the risk level of the multimedia information to be processed based on the multimedia risk cases in the multimedia risk database.

[0152] Specifically, when constructing a multimedia risk database, multimedia risk information can be obtained from staff (e.g., such as...). Figure 15 The auditor or product operations staff (as indicated) use the work platform (e.g., ... Figure 15The reviewers, as shown, report the collected multimedia data through advertising review platforms, or product operations through advertising material retrieval platforms, etc. It is understandable that if the reviewers find any risks such as illegality or non-compliance in the multimedia data (as shown in Figure 18(a)), they can mark the multimedia data as risky (e.g., ...). Figure 15 (Illustrated advertising labeling, such as risk symbols, risk icons, or risk keywords, etc.)

[0153] Furthermore, the system can process the reported multimedia data (such as...) Figure 15 The system iterates through advertising data reported by auditors or product operations staff, and uses the multimedia data marked with risk as multimedia risk reporting data.

[0154] Furthermore, in order to obtain more comprehensive multimedia risk information, a scheduled task can be used to periodically scrape data from the multimedia information publishing platform to acquire multimedia risk data (such as...). Figure 15 (As shown, this refers to advertising data captured from an online advertising database via a scheduled task.) For example, by checking the online advertising database, a scheduled inspection task can be set up to pull online advertising data every hour, capture data from landing pages, and parse the captured data and the content of the advertising materials themselves to obtain parsed data, i.e., multimedia risk capture data. Specifically, this can be done through video frame extraction or image OCR, or other parsing methods, which are not specifically limited here.

[0155] Furthermore, the acquired multimedia risk reporting data or multimedia risk capture data can be matched with preset multimedia risk keywords. For example, the multimedia risk reporting data with risk tags that have been traversed can be matched with multimedia risk keywords. Assuming that the text content of the risk tags or multimedia risk reporting data matches the multimedia risk keywords, it can be considered that the acquired multimedia risk reporting data (such as the advertising image material shown in Figure 18(a)) is risky. Then, the multimedia risk reporting data that matches the multimedia risk keywords can be stored in the multimedia risk database (e.g., Figure 15The interface indicates that multimedia risk reporting data matching multimedia risk keywords can be uploaded to the multimedia risk database (i.e., the multimedia risk database) as risk cases. For example, risk detection capabilities such as black words, inappropriate phrases, or clickbait can be invoked to identify risks in the parsed multimedia risk data. If the parsed data, such as landing pages or advertising materials, matches the aforementioned detection capabilities, the captured advertisement can be considered risky. This means the multimedia risk data matches multimedia risk keywords, or multimedia risk data matching multimedia risk keywords is used as a risk case and stored in the multimedia risk database (e.g., ...). Figure 15 As shown, the risk case reporting interface can be used to upload multimedia risk data that matches multimedia risk keywords as risk cases to the risk case library, i.e., the multimedia risk database.

[0156] Optionally, in the above Figure 2 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 5 As shown, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes:

[0157] In step S501, raw negative feedback data corresponding to the multimedia information is received, wherein the raw negative feedback data includes historical exposure data of the multimedia information;

[0158] In step S502, the historical playback log corresponding to the multimedia information is obtained, wherein the historical playback log includes at least the playback position information and the group object information corresponding to the multimedia information;

[0159] In step S503, historical exposure data, playback location information, and group object information are integrated, and the integrated multimedia negative feedback data is stored in the multimedia negative feedback database.

[0160] In this embodiment of the application, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, a multimedia negative feedback database can be constructed first. This is done by receiving the original negative feedback data corresponding to the multimedia information and obtaining the historical playback logs corresponding to the multimedia information. Then, the historical exposure data, playback location information, and group object information are integrated, and the integrated multimedia negative feedback data is stored in the multimedia negative feedback database. This allows for a better assessment or determination of the risk level of the multimedia information to be processed based on the multimedia negative feedback data in the multimedia negative feedback database.

[0161] The original negative feedback data refers to negative feedback information (such as the reason for the complaint and a detailed description of the complaint content) collected through negative feedback pages (such as complaint pages). This data includes historical exposure data for multimedia information, which can specifically represent, for example, the number of times an advertisement was displayed to the target audience through the advertising playback system within a historical time period, such as the past week. Historical playback logs refer to the playback logs recorded by the system when playing multimedia information within a historical time period, such as the past week. These logs include at least the playback location information and the target audience information. The playback location information refers to the geographical location of the multimedia information, while the target audience information refers to the personal information of the recipients or viewers of the multimedia information, such as their interests and preferences, and their address information.

[0162] Specifically, when constructing a multimedia negative feedback database, this can be achieved by obtaining historical playback logs corresponding to multimedia information, for example... Figure 16 As shown, when the advertising playback system displays an ad to the audience within a historical time period, it can generate a viewid to represent the exposure data for this time. At the same time, it records the playback log context, i.e., the historical playback log.

[0163] Simultaneously, it receives the raw negative feedback data corresponding to the multimedia information, specifically, for example, such as... Figure 16 As shown, when an audience sees an ad and discovers issues such as counterfeiting other brands, false advertising, or fraud, they can provide negative feedback and complain about these problems. This can be done through a pre-built negative feedback page (e.g., ...). Figure 16The H5 page (shown as an example) is used to collect negative feedback information uploaded by the advertising audience. This information can include: the reason for the complaint, a description, and the complainant's information (such as a mobile phone number). When submitting data on the page, the viewid from when the ad was displayed is also included. This allows the historical exposure data to be retrieved from the received raw negative feedback data by extracting the viewid field, which can then be used to record more detailed information. Furthermore, a negative feedback CGI can be pre-built to receive the negative feedback data submitted by the advertising audience. After receiving the data submitted by the advertising audience, the negative feedback CGI does not need to perform much logic and directly saves the data to TDbank.

[0164] Furthermore, a logjoin service can be built to detect data changes in the tdbank and obtain newly added data in the tdbank. That is, when new historical exposure data, playback location information, and group object information are collected, that is, when new data is added to the tdbank, the historical exposure data, playback location information, and group object information can be integrated, and the integrated multimedia negative feedback data can be stored as newly added data in the multimedia negative feedback database.

[0165] Optionally, in the above Figure 5 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 6 As shown, step S503 integrates historical exposure data, playback location information, and group object information, and stores the integrated multimedia negative feedback data in the multimedia negative feedback database, including:

[0166] In step S601, historical exposure data is integrated into playback location information and group object information to obtain multimedia scene data;

[0167] In step S602, the original negative feedback data and multimedia scene data are recorded as multimedia negative feedback data and stored in the multimedia negative feedback database.

[0168] In this embodiment of the application, after obtaining historical exposure data, playback location information and group object information, the historical exposure data can be integrated into the playback location information and group object information to obtain multimedia scene data. Then, the original negative feedback data and multimedia scene data can be recorded as multimedia negative feedback data and stored in the multimedia negative feedback database.

[0169] Specifically, historical exposure data, playback location information, and target audience information are integrated, which can be achieved through... Figure 16The diagram illustrates a pre-built logjoin service used to detect data changes in the TDbank and retrieve newly added data. Specifically, when new historical exposure data, playback location information, and audience information are collected, indicating new data has been added to the TDbank, this can be achieved by extracting the `viewid` field from the original negative feedback data as historical exposure data, and obtaining playback location and audience information from historical playback logs. This historical exposure data, playback location information, and audience information can then be integrated to establish more detailed data relationships. Specifically, the `viewid` field can be used to join the ad scenario data from the ad playback logs. Finally, the integrated multimedia negative feedback data, i.e., the joined data, can be stored in a multimedia negative feedback database for downstream analysis.

[0170] Optionally, in the above Figure 2 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 7 As shown, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes:

[0171] In step S701, data is crawled based on the crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy. The crawling strategy includes one or more different crawling conditions.

[0172] In step S702, the multimedia crawled data is stored in the multimedia playback database.

[0173] In this embodiment of the application, before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, when constructing the multimedia playback database, data can be crawled based on the crawling strategy to obtain the multimedia crawling data corresponding to the crawling strategy. The multimedia crawling data can then be stored in the multimedia playback database so that the risk level of the multimedia information to be processed can be better judged or determined based on the multimedia crawling data in the multimedia playback database.

[0174] Specifically, when constructing a multimedia playback database, such as Figure 17As shown, when constructing a multimedia playback database such as a market advertising database, in order to obtain more advertising data from various media outlets in the advertising market, a web crawler task can be set up. Based on a strategy control module, different combinations of strategy conditions can be used to crawl the multimedia data, obtaining data under different strategy condition combinations. For example, crawling based on time period conditions (such as...) Figure 17 (As indicated by the time period strategy), data is crawled separately according to different time periods to obtain multimedia crawled data corresponding to each time period; and / or crawling conditions based on group object information (such as... Figure 17 (As indicated by the group object strategy), data is crawled separately according to different group attributes to obtain multimedia crawled data corresponding to each group attribute; and / or crawling conditions based on geographic distribution information (such as... Figure 17 (As indicated by the geographic strategy), data is crawled separately according to different geographic attributes to obtain multimedia crawled data corresponding to each geographic attribute; and / or crawling based on device coverage conditions (such as... Figure 17 The device segment strategy (as shown) is used to crawl data according to different device coverage rates to obtain multimedia crawl data corresponding to each device coverage rate.

[0175] Optionally, in the above Figure 7 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 8 As shown, when the crawling strategy includes one or more of the following: time period crawling conditions, group object information crawling conditions, geographical distribution information crawling conditions, and device coverage crawling conditions; step S701 performs data crawling based on the crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy, including:

[0176] In step S801, based on the time-period crawling conditions, data is crawled separately according to different time periods to obtain multimedia crawled data corresponding to each time period; and / or

[0177] In step S802, based on the group object information crawling conditions, data is crawled separately according to different group attributes to obtain multimedia crawled data corresponding to each group attribute; and / or

[0178] In step S803, based on the geographic distribution information crawling conditions, data is crawled separately according to different geographic attributes to obtain multimedia crawled data corresponding to each geographic attribute; and / or

[0179] In step S804, based on the device coverage crawling conditions, data is crawled separately according to different device coverage to obtain multimedia crawled data corresponding to each device coverage.

[0180] In this embodiment of the application, when constructing a multimedia playback database, data can be crawled based on a crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy. The crawling strategy may include one or more of the following: time period crawling conditions, group object information crawling conditions, geographical distribution information crawling conditions, and device coverage crawling conditions, thus forming a combination of different condition strategies.

[0181] Specifically, such as Figure 17 As shown, in order to obtain as much advertising data as possible from various media outlets, the main capabilities that need to be supported when setting up a web crawler task are (such as...). Figure 17 (Indicates the underlying capabilities of the crawler):

[0182] (1) Time period control refers to configuring different execution times of crawler tasks through a crontab-like method, so that multimedia crawling data corresponding to different execution times can be crawled based on the different execution times of the crawler tasks.

[0183] (2) Simulation of different users (e.g., simulation of different groups), for example, simulating different gender groups such as minors or women to crawl multimedia data corresponding to different groups. Specifically, simulating different gender groups such as minors or women can be achieved by constructing multimedia data crawling requests through different mobile phone numbers or other user accounts.

[0184] (3) Regional simulation refers to crawling different regions by constructing different client IPs, such as multimedia crawling data of first-level cities and multimedia crawling data of second-level cities.

[0185] (4) Device simulation refers to data crawling by constructing different referers and User-Agents to simulate crawling requests from different devices.

[0186] Furthermore, such as Figure 17 As shown, after acquiring various capabilities for web scraping tasks, different scraping strategies can be designed by configuring a strategy control module to coordinate with the scraping tasks, thereby obtaining diverse data. Among these, the data scraping based on the scraping strategy corresponds to the scraping capabilities:

[0187] (1) Crawling based on time period conditions, such as Figure 17The time-segment strategy shown allows for data crawling according to different time periods to obtain multimedia crawling data corresponding to each time period. For example, advertisers in different industries prefer to use different time periods for advertising, but due to limited machine resources and data interface call frequency limitations, the time-segment strategy control module can control the multimedia data crawling of different media in different time periods, such as calling the data crawler 10 times every 10 minutes for each media.

[0188] (2) Crawling based on group object information conditions, such as Figure 17 The intended target group strategy involves crawling data according to different group attributes to obtain multimedia crawling data corresponding to each group attribute. For example, since different groups receive different advertisements, some advertisements may only be suitable for the elderly, some may only be suitable for women, and some may not be suitable for minors. Therefore, different group information such as age, gender, education, or interests are used to crawl advertisements for different groups.

[0189] (3) Crawling conditions based on geographical distribution information, such as Figure 17 The intended geographic strategy involves crawling data according to different geographic attributes to obtain multimedia data corresponding to each geographic attribute. For example, in terms of geographic distribution, the geographic areas can be divided according to first-tier cities, second-tier cities, third-tier cities, or other distribution forms. Other distribution methods can also be used, without specific restrictions here. The goal is to randomly select values ​​for each media to crawl multimedia data from different geographic distributions each time.

[0190] (4) Crawling based on device coverage conditions, such as Figure 17 The device strategy shown involves crawling data according to different device coverage rates to obtain multimedia crawling data corresponding to each device coverage rate. For example, different devices have different market share rates, so the device market share rates can be sorted from high to low and taken out one by one as the settings to be covered when crawling data, until the total share rate of all devices reaches a share rate threshold such as 90%, and data crawling is performed based on the share rate of the different devices taken out. Among them, the higher the share rate, the greater the probability of the device being used.

[0191] Understandably, in order to obtain more diverse data, data can be crawled by randomly combining the values ​​of the above four strategies.

[0192] Optionally, in the above Figure 2 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 9As shown, step S103 obtains the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, including:

[0193] In step S901, the exposure, clicks and consumption corresponding to each multimedia risk information in the multimedia risk database are obtained;

[0194] In step S902, the initial risk value corresponding to the multimedia risk information is determined based on the exposure, clicks, and consumption.

[0195] In step S903, the negative feedback amount and negative feedback height corresponding to each multimedia negative feedback data in the multimedia negative feedback database are obtained;

[0196] In step S904, the initial risk value corresponding to the multimedia negative feedback data is determined based on the negative feedback quantity and the negative feedback height.

[0197] In this embodiment, to better estimate the risk value of the multimedia information to be processed, after constructing the multimedia risk database, multimedia negative feedback database, and multimedia playback database, risk values ​​can be initialized on the data in these databases. This allows for further estimation of the risk value of the multimedia information to be processed based on the initial risk values ​​and the similarity between the multimedia information to be processed and the data in these databases. Specifically, the exposure, clicks, and consumption corresponding to each multimedia risk information in the multimedia risk database are obtained, and the initial risk value corresponding to the multimedia risk information is determined based on these values. Simultaneously, the negative feedback quantity and negative feedback height corresponding to each multimedia negative feedback data in the multimedia negative feedback database are obtained, and the initial risk value corresponding to the multimedia negative feedback data is determined based on these values.

[0198] Exposure refers to the number of times each multimedia risk message is exposed within a specified time period. For example, if there is a game advertisement, the number of people who view the game advertisement each day is equivalent to traffic, or exposure. Clicks refer to the number of times each multimedia risk message is clicked. For example, if there is a game advertisement, the number of people who view the game advertisement each day is equivalent to traffic, or exposure. The number of people who clicked through to the game advertisement's details page or landing page from among those who viewed the advertisement is equivalent to the click count. Cost refers to the cost per click for each multimedia risk message within a set time period. For example, the cost per click incurred after the game advertisement is clicked within a set time period.

[0199] Negative feedback volume refers to the number of complaints made by multimedia information recipients (such as advertising audiences) through a complaint page due to inappropriate or illegal information when receiving or browsing multimedia information. For example, a game advertisement might be complained about 20 times in one day. Negative feedback intensity refers to the severity of the complaint reasons described by multimedia information recipients (such as advertising audiences) when they complain about multimedia information, reaching a preset complaint index. For example, if a game advertisement is complained about because it contains a violent word, the complaint verification intensity is 0.3 degrees, which is the preset complaint index.

[0200] Specifically, after constructing the multimedia risk database, multimedia negative feedback database, and multimedia playback database, a risk value analysis engine (such as...) can be built. Figure 14 The advertising risk analysis engine shown in the image allows for the subsequent construction of a risk value analysis engine (such as...). Figure 14 The advertising risk analysis engine (as shown) analyzes and processes the multimedia information to be processed. Since the risk cases or negative feedback data in the multimedia risk database and the multimedia negative feedback database are materials that are considered to have risks and high risk values, the risk value initialization needs to be performed on the multimedia risk database and the multimedia negative feedback database. Therefore, the risk value initialization can be performed on the data in the multimedia risk database and the multimedia negative feedback database first.

[0201] Understandably, the multimedia risk database and the multimedia negative feedback database require initial risk values. For the multimedia risk database, since the occurrence of risk cases is strongly correlated with exposure, clicks, and spending, the initial risk value can be calculated based on the exposure (a), clicks (b), and spending (c) corresponding to each multimedia risk information in the database. Then, using preset weights for exposure, clicks, and spending, the initial risk value can be calculated using a formula such as: a*60% + b*30% + c*10%.

[0202] Furthermore, for a multimedia negative feedback database, it is understood that since the multimedia negative feedback data in the database can carry negative feedback quantity (such as user complaint quantity a) and negative feedback height (such as the severity of complaint reasons b), and the negative feedback quantity and negative feedback height are strongly correlated with the risk level of the multimedia negative feedback data, the initial risk value corresponding to the multimedia negative feedback data can be calculated using a formula such as a*b.

[0203] Optionally, in the above Figure 9Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 10 As shown, the similarity includes one or more of the first similarity, second similarity, and third similarity; step S102 calculates the similarity between the multimedia information to be processed and the data in the multimedia risk database, multimedia negative feedback database, and multimedia playback database, respectively, to obtain the first similarity set, the second similarity set, and the third similarity set, including:

[0204] In step S1001, the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database is calculated to obtain the first similarity set;

[0205] In step S1002, the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database is calculated to obtain the second similarity set;

[0206] In step S1003, the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database is calculated to obtain a third similarity set.

[0207] In this embodiment, after constructing the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database, a first similarity can be calculated between the multimedia information to be processed and each multimedia risk information in the multimedia risk database to obtain a first similarity set; a second similarity can be calculated between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database to obtain a second similarity set; and a third similarity can be calculated between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database to obtain a third similarity set, so that the risk value of the multimedia information to be processed can be better calculated or determined based on the first similarity set, the second similarity set, and the third similarity set.

[0208] Specifically, after constructing the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database, for the multimedia information to be processed, the multimedia information to be processed can be compared with the multimedia risk database (e.g., Figure 14 (e.g., advertising risk database, multimedia negative feedback database) Figure 14 The database of negative feedback from the intended advertising audience) and the multimedia playback database (e.g., Figure 14The system performs a similarity comparison calculation on data such as risk cases or risk materials in the advertising media playback database (e.g., market advertising database). Specifically, it calculates the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database to obtain a first similarity set; calculates the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database to obtain a second similarity set; and calculates the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database to obtain a third similarity set.

[0209] Understandably, in order to more accurately calculate the multimedia information to be processed and compare it with the multimedia risk database (e.g., Figure 14 (e.g., advertising risk database, multimedia negative feedback database) Figure 14 The database of negative feedback from the intended advertising audience) and the multimedia playback database (e.g., Figure 14 The similarity of data in a target advertising media database (such as a market advertising database), such as risk cases or risk materials, can be calculated as follows: For text, the Hamming distance algorithm can be used for similarity calculation; for images, common image mean hashing, image perceptual hashing, or image difference hashing algorithms can be used to calculate similarity x, y, and z for each, and then the average calculation formula (x+y+z) / 3 can be used as the similarity; for audio and video, frames can be randomly sampled first, and then the similarity calculation method with the images mentioned above can be used to calculate the similarity a1, a2, a3, ..., an for each frame. Then, (a1+a2+a3+...+an) / n is used as the similarity to obtain the first similarity set, the second similarity set, and the third similarity set.

[0210] Optionally, in the above Figure 10 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 11 As shown, step S104 calculates the target risk value corresponding to the multimedia information to be processed based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set, including:

[0211] In step S1101, the first similarity values ​​greater than the similarity threshold and the initial risk values ​​corresponding to the first similarity values ​​are summed and averaged to obtain the first risk value;

[0212] In step S1102, the second similarity values ​​greater than the similarity threshold and the initial risk values ​​corresponding to the second similarity values ​​are summed and averaged to obtain the second risk value;

[0213] In step S1103, the risk reduction weight is obtained based on the third similarity, and the risk reduction weight is used as the risk factor of the multimedia information to be processed;

[0214] In step S1104, the target risk value is determined based on the first risk value, the second risk value, and the risk factor.

[0215] In this embodiment, after obtaining the first similarity set, the second similarity set, and the third similarity set, as well as the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, it is understood that a higher similarity indicates a higher risk level for the multimedia information to be processed. Therefore, similarity screening can be performed using a similarity threshold. Specifically, the first similarity greater than the similarity threshold is screened, and the first similarity greater than the similarity threshold and the initial risk value corresponding to the first similarity are summed and averaged, with the calculated average value serving as the first risk value. Simultaneously, the second similarity greater than the similarity threshold is screened, and the second similarity greater than the similarity threshold and the initial risk value corresponding to the second similarity are summed and averaged, with the calculated average value serving as the second risk value. Furthermore, a risk reduction weight is obtained based on the third similarity, and this risk reduction weight is used as a risk factor for the multimedia information to be processed. Then, the target risk value of the multimedia information to be processed can be calculated based on the obtained first risk value, second risk value, and risk factor.

[0216] Specifically, after obtaining the first similarity set, the second similarity set, and the third similarity set, and after obtaining the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, the target risk value corresponding to the multimedia information to be processed can be calculated. For example, for the multimedia risk database (e.g., ... Figure 14 (e.g., advertising risk database, multimedia negative feedback database) Figure 14 The database of negative feedback from the intended advertising audience) and the multimedia playback database (e.g., Figure 14The intended advertising media playback database (e.g., a market advertising database) includes a multimedia risk database, a multimedia negative feedback database, and a multimedia playback database, each containing 100 data points. Assume the initial risk values ​​for the multimedia risk database, corresponding to the first similarity, are a1, a2, a3, ..., a100; the initial risk values ​​for the multimedia negative feedback database, corresponding to the second similarity, are b1, b2, b3, ..., b100; and assume the first similarity between the multimedia information to be processed and the 100 data points in the multimedia risk database is sa1, sa2, sa3, ..., sa100; the second similarity with the 100 data points in the multimedia negative feedback database is sb1, sb2, sb3, ..., sb100; and the third similarity with the 100 data points in the multimedia playback database is sc1, sc2, sc3, ..., sc100.

[0217] Furthermore, since the first similarity can be used to represent the degree of similarity between the multimedia information to be processed and each risk data in the 100 data in the multimedia risk database, the risk score of the multimedia information to be processed can be represented by calculating the product between the first similarity and the initial risk value based on the initial risk value corresponding to each risk data. Therefore, the first risk value can be calculated by multiplying and averaging the first similarity greater than the similarity threshold and the initial risk value corresponding to the first similarity. Specifically, the first 10 first similarities, such as sa1, sa2, sa3, ..., sa10, and the initial risk values ​​corresponding to the first 10 first similarities, such as a1, a2, a3, ..., a10, can be extracted, and then the summation and averaging can be performed to obtain the first risk value, i.e., (sa1a1)+(sa2a2)+(sa3a3)+(sa3a3)+...+(sa10a10).

[0218] Furthermore, similarly, since the second similarity can be used to represent the degree of similarity between the multimedia information to be processed and each risk data in the 100 data in the multimedia negative feedback database, the risk score of the multimedia information to be processed can be represented by calculating the product between the second similarity and the initial risk value based on the initial risk value corresponding to each risk data. Therefore, the second risk value can be calculated by multiplying and averaging the second similarity greater than the similarity threshold and the initial risk value corresponding to the second similarity. Specifically, the top 10 second similarities, such as sb1, sb2, sb3, ..., sb10, and the initial risk values ​​corresponding to the top 10 second similarities, such as b1, b2, b3, ..., b10, can be extracted, and then the summation and averaging can be performed to obtain the second risk value, i.e., (sb1b1) + (sb2b2) + (sb3b3) + (sb3b3) + ... + (sb10b10).

[0219] Furthermore, for multimedia playback databases (e.g., Figure 14 The data already existing in the advertising media playback database (such as the market advertising database) represents materials that have been assessed by other risk systems in the market and are considered to have low risk values. These can be understood as materials permitted for playback in the market. Therefore, it is not necessary to calculate the initial risk value of risk cases in the multimedia playback database. Instead, by assigning a certain confidence level, the risk proportion (i.e., risk factor) of risk cases in the multimedia playback database can be determined, thereby reducing the risk value of the multimedia information to be processed. The risk reduction weight is obtained based on third-degree similarity. First, the average similarity of 100 third-degree similarities can be calculated, such as (sc1+sc2+sc3+...+sc100) / 100. Then, the weight for reducing the risk value (i.e., risk reduction weight) can be calculated based on the average similarity, such as 1-(sc1+sc2+sc3+...+sc100) / 100. It can be understood that the higher the similarity, the greater the reduction.

[0220] Furthermore, the target risk value can be obtained by calculating the sum of the first risk value, the second risk value, and the risk factors.

[0221] Optionally, in the above Figure 11 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 12 As shown, step S1104 determines the target risk value based on the first risk value, the second risk value, and the risk factor, including:

[0222] In step S1201, the sum of the first risk value and the second risk value is obtained;

[0223] In step S1202, the product between the sum and the risk factor is calculated, and the product is used as the target risk value.

[0224] Specifically, based on the first risk value, the second risk value, and the risk factor, the target risk value is determined, which can be achieved by calculating the sum of the first risk value and the second risk value:

[0225] ((sa1a1)+(sa2a2)+(sa3a3)+...+(sa10a10)+(sb1b1)+(sb2b2)+(sb3b3)+(sb3b3)+...+(sb10b10));

[0226] Furthermore, the sum of the risk factors is calculated as a product, and this product is used as the target risk value.

[0227] ((sa1a1)+(sa2a2)+(sa3a3)+...+(sa10a10)+(sb1b1)+(sb2b2)+(sb3b3)+(sb3b3)+...+(sb10b10))*(1-(sc1+sc2+sc3+...+sc100) / 100).

[0228] Optionally, in the above Figure 2 Based on the corresponding embodiments, in another optional embodiment of the multimedia information push method provided in this application, such as... Figure 13 As shown, the review criteria for multimedia information include hitting risky review keywords; based on the target risk value and the review criteria for multimedia information, the review result for the multimedia information to be processed is determined, including:

[0229] In step S1301, if the target risk value is greater than the risk threshold, or the number of times the risk audit keywords are hit is greater than the quantity threshold, then the audit result corresponding to the multimedia information to be processed is not pushable.

[0230] In step S1302, if the target risk value is less than or equal to the risk threshold, and the number of times the risk audit keywords are hit is less than or equal to the quantity threshold, then the audit result that can be pushed to the multimedia information to be processed is obtained.

[0231] In this embodiment of the application, after obtaining the target risk value, it is possible to better judge or determine whether the multimedia information to be processed is suitable for healthy delivery based on the target risk value and in combination with the multimedia information review conditions. If the target risk value is greater than the risk threshold, or the number of times the risk review keywords are hit is greater than the quantity threshold, the review result corresponding to the multimedia information to be processed that cannot be pushed can be obtained. Conversely, if the target risk value is less than or equal to the risk threshold, and the number of times the risk review keywords are hit is less than or equal to the quantity threshold, the review result corresponding to the multimedia information to be processed that can be pushed can be obtained.

[0232] Specifically, after obtaining the target risk value, such as Figure 14 As shown, in order to more accurately determine whether the multimedia information to be processed is suitable for healthy delivery and to obtain the review result of the multimedia information to be processed, it is also possible to combine the multimedia information review conditions (such as... Figure 14 The intended advertising review rules, such as the number of risky keywords in the multimedia information to be processed being less than or equal to a certain threshold, are used to further obtain the review results of the multimedia information to be processed (e.g., Figure 14 (The ad review process is as shown). For example, if the target risk value is less than or equal to the risk threshold, and the number of risky keywords hit by the multimedia information to be processed is less than or equal to the quantity threshold, it can be understood that the risk value of the multimedia information to be processed is not high and it is safe. In this case, the review result is that it can be pushed. Conversely, if the target risk value is greater than the risk threshold, or the number of risky keywords hit by the multimedia information to be processed is greater than the quantity threshold, it can be understood that the risk value of the multimedia information to be processed is high and it is not safe. In this case, the review result is that it cannot be pushed.

[0233] For example, as shown in Figure 18(c), suppose an advertisement image and advertisement copy to be placed are calculated to have a target risk value (e.g., a risk value of 30). Suppose the risk threshold is 40, that is, the target risk value is less than the risk threshold. However, suppose the number of risk keywords hit by the multimedia information to be processed is greater than the number threshold (e.g., the advertisement copy in Figure 18(c) says "Don't spend 2xx yuan, you can get it for 0 (red envelope) yuan". For example, the word "red envelope" hits the risk keyword and there is a risk mark). It can be understood that the risk value of the advertisement image and advertisement copy to be placed is high and it is unsafe. The review result corresponding to the advertisement image and advertisement copy to be placed is that it cannot be pushed.

[0234] The multimedia information push device in this application is described in detail below. Please refer to [link / reference]. Figure 19 , Figure 19 This is a schematic diagram of one embodiment of the multimedia information push device in this application. The multimedia information push device 20 includes:

[0235] Acquisition unit 201 is used to acquire multimedia information to be processed;

[0236] The processing unit 202 is used to calculate the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database and the multimedia playback database, respectively, to obtain a first similarity set, a second similarity set and a third similarity set;

[0237] The acquisition unit 201 is also used to acquire the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database, respectively.

[0238] The processing unit 202 is also used to calculate the target risk value corresponding to the multimedia information to be processed based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set;

[0239] The determining unit 203 is used to determine the review result of the multimedia information to be processed based on the target risk value and the multimedia information review conditions;

[0240] The processing unit 202 is also used to push the multimedia information to be processed to the multimedia information receiving object if the review result is pushable.

[0241] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application,

[0242] The acquisition unit 201 is also used to acquire multimedia risk information, which includes multimedia risk reporting data and multimedia risk capture data.

[0243] The processing unit 202 is also used to store the multimedia risk information to the multimedia risk database if the multimedia risk information matches the multimedia risk condition.

[0244] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the acquisition unit 201 can specifically be used for:

[0245] The reported multimedia data is traversed, and the multimedia data marked with risk is used as the multimedia risk reporting data.

[0246] Regularly crawl data from multimedia information publishing platforms to obtain multimedia risk capture data;

[0247] The processing unit 202 can be specifically used to: if there are multimedia risk keywords in the multimedia risk reporting data or multimedia risk capture data, store the multimedia risk reporting data or multimedia risk capture data into the multimedia risk database.

[0248] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application,

[0249] The acquisition unit 201 is also used to receive raw negative feedback data corresponding to multimedia information, wherein the raw negative feedback data includes historical exposure data of multimedia information;

[0250] The acquisition unit 201 is also used to acquire the historical playback log corresponding to the multimedia information, wherein the historical playback log includes at least the playback position information and the group object information corresponding to the multimedia information.

[0251] The processing unit 202 is also used to integrate historical exposure data, playback location information and group object information, and store the integrated multimedia negative feedback data in the multimedia negative feedback database.

[0252] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the processing unit 202 may specifically be used for:

[0253] Historical exposure data is integrated with playback location information and group object information to obtain multimedia scene data;

[0254] The original negative feedback data and multimedia scene data are recorded as multimedia negative feedback data and stored in the multimedia negative feedback database.

[0255] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application,

[0256] The processing unit 202 is also used to crawl data based on the crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy, wherein the crawling strategy includes one or more different crawling conditions.

[0257] The processing unit 202 is also used to store the multimedia crawled data into the multimedia playback database.

[0258] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the processing unit 202 may specifically be used for:

[0259] Based on time-time crawling criteria, data is crawled separately according to different time periods to obtain multimedia crawled data corresponding to each time period; and / or

[0260] Based on the group object information crawling conditions, data is crawled separately according to different group attributes to obtain multimedia crawled data corresponding to each group attribute; and / or

[0261] Based on geographic distribution information crawling conditions, data is crawled separately according to different geographic attributes to obtain multimedia crawled data corresponding to each geographic attribute; and / or

[0262] Based on the device coverage crawling condition, data is crawled separately according to different device coverage, and multimedia crawled data corresponding to each device coverage is obtained.

[0263] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the acquisition unit 201 can specifically be used for:

[0264] Obtain the exposure, clicks, and cost corresponding to each multimedia risk information in the multimedia risk database;

[0265] Based on exposure, clicks, and consumption, determine the initial risk value corresponding to multimedia risk information;

[0266] Obtain the negative feedback quantity and negative feedback height corresponding to each multimedia negative feedback data in the multimedia negative feedback database;

[0267] Based on the amount and height of negative feedback, the initial risk value corresponding to the multimedia negative feedback data is determined.

[0268] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the processing unit 202 may specifically be used for:

[0269] Calculate the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database to obtain the first similarity set;

[0270] Calculate the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database to obtain the second similarity set;

[0271] Calculate the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database to obtain the third similarity set.

[0272] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the processing unit 202 may specifically be used for:

[0273] The first risk value is obtained by summing and averaging the first similarity scores that are greater than the similarity threshold and the initial risk values ​​corresponding to the first similarity scores.

[0274] The second risk value is obtained by summing and averaging the second similarity values ​​that are greater than the similarity threshold and the initial risk values ​​corresponding to the second similarity values.

[0275] Risk reduction weights are obtained based on third-degree similarity and used as risk factors for the multimedia information to be processed.

[0276] The target risk value is determined based on the first risk value, the second risk value, and the risk factors.

[0277] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the processing unit 202 may specifically be used for:

[0278] Obtain the sum between the first risk value and the second risk value;

[0279] Calculate the product between the sum and the risk factor, and use the product as the target risk value.

[0280] Optionally, in the above Figure 19 Based on the corresponding embodiments, in another embodiment of the multimedia information push device provided in this application, the determining unit 203 can specifically be used for:

[0281] If the target risk value is greater than the risk threshold, or the number of times the risk audit keywords are hit is greater than the quantity threshold, then the audit result corresponding to the multimedia information to be processed is not pushable.

[0282] If the target risk value is less than or equal to the risk threshold, and the number of times the risk audit keywords are hit is less than or equal to the quantity threshold, then the audit result corresponding to the multimedia information to be processed can be pushed.

[0283] This application also provides a schematic diagram of another computer device, such as... Figure 20 As shown, Figure 20This is a schematic diagram of a computer device structure provided in an embodiment of this application. The computer device 300 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 331 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the computer device 300. Furthermore, the CPU 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the computer device 300.

[0284] Computer device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 333, such as Windows Server. TM Mac OS X TM Unix TM Linux TM FreeBSD TM etc.

[0285] The aforementioned computer device 300 is also used to perform, for example Figures 2 to 14 The steps in the corresponding embodiments.

[0286] Another aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements... Figures 2 to 14 The steps in the method described in the illustrated embodiment.

[0287] Another aspect of this application provides a computer program product comprising a computer program, which, when executed by a processor, implements as follows: Figures 2 to 14 The steps in the method described in the illustrated embodiment.

[0288] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0289] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0290] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0291] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0292] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for pushing multimedia information, characterized in that, include: Acquire multimedia information to be processed; The similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database is calculated respectively to obtain a first similarity set, a second similarity set, and a third similarity set; The initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database are obtained respectively. Specifically, the initial risk value corresponding to the multimedia risk information is calculated by obtaining the exposure, clicks and consumption corresponding to each multimedia risk information in the multimedia risk database. The initial risk value corresponding to the multimedia negative feedback data is calculated based on the negative feedback amount and negative feedback height of the multimedia negative feedback data. Based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set, the target risk value corresponding to the multimedia information to be processed is calculated; Based on the target risk value and the multimedia information review conditions, the review result of the multimedia information to be processed is determined; If the review result indicates that the multimedia information to be processed is pushable, then the multimedia information to be processed will be pushed to the multimedia information receiving object.

2. The method according to claim 1, characterized in that, Before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes: Obtain multimedia risk information, wherein the multimedia risk information includes multimedia risk reporting data and multimedia risk capture data; If the multimedia risk information matches the multimedia risk condition, then the multimedia risk information is stored in the multimedia risk database.

3. The method according to claim 2, characterized in that, The acquisition of multimedia risk information includes: The reported multimedia data is traversed, and the multimedia data marked with risk is used as the multimedia risk reporting data. Regularly crawl data from multimedia information publishing platforms to obtain multimedia risk capture data; If the multimedia risk information matches the multimedia risk condition, then storing the multimedia risk information in the multimedia risk database includes: If multimedia risk keywords are present in the multimedia risk reporting data or the multimedia risk capture data, then the multimedia risk reporting data or the multimedia risk capture data will be stored in the multimedia risk database.

4. The method according to claim 1, characterized in that, Before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes: Receive raw negative feedback data corresponding to multimedia information, wherein the raw negative feedback data includes historical exposure data of the multimedia information; Obtain the historical playback log corresponding to the multimedia information, wherein the historical playback log includes at least the playback location information and the group object information corresponding to the multimedia information; The historical exposure data, the playback location information, and the group object information are integrated, and the integrated multimedia negative feedback data is stored in the multimedia negative feedback database.

5. The method according to claim 4, characterized in that, The step of integrating the historical exposure data, the playback location information, and the group object information, and storing the integrated multimedia negative feedback data in the multimedia negative feedback database, includes: The historical exposure data is integrated with the playback location information and the group object information to obtain multimedia scene data; The original negative feedback data and the multimedia scene data are recorded as multimedia negative feedback data and stored in the multimedia negative feedback database.

6. The method according to claim 1, characterized in that, Before calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain the first similarity set, the second similarity set, and the third similarity set, the method further includes: Data is crawled based on a crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy, wherein the crawling strategy includes one or more different crawling conditions. The multimedia crawled data is stored in the multimedia playback database.

7. The method according to claim 6, characterized in that, When the crawling strategy includes one or more of the following: time period crawling conditions, group object information crawling conditions, geographical distribution information crawling conditions, and device coverage crawling conditions; The process of crawling data based on a crawling strategy to obtain multimedia crawled data corresponding to the crawling strategy includes: Based on the time-period crawling conditions, data is crawled separately according to different time periods to obtain multimedia crawled data corresponding to each time period; and / or Based on the aforementioned group object information crawling conditions, data is crawled separately according to different group attributes to obtain multimedia crawled data corresponding to each group attribute; and / or Based on the aforementioned geographic distribution information crawling conditions, data is crawled separately according to different geographic attributes to obtain multimedia crawled data corresponding to each geographic attribute; and / or Based on the device coverage crawling conditions, data is crawled separately according to different device coverage rates to obtain multimedia crawled data corresponding to each device coverage rate.

8. The method according to claim 1, characterized in that, The step of obtaining the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database respectively includes: Obtain the exposure, clicks, and consumption corresponding to each multimedia risk information in the multimedia risk database; Based on the exposure, the number of clicks, and the consumption, the initial risk value corresponding to the multimedia risk information is determined; Obtain the negative feedback quantity and negative feedback height corresponding to each multimedia negative feedback data in the multimedia negative feedback database; Based on the negative feedback quantity and the negative feedback height, the initial risk value corresponding to the multimedia negative feedback data is determined.

9. The method according to claim 8, characterized in that, The step of calculating the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database, and the multimedia playback database to obtain a first similarity set, a second similarity set, and a third similarity set includes: Calculate the first similarity between the multimedia information to be processed and each multimedia risk information in the multimedia risk database to obtain the first similarity set; Calculate the second similarity between the multimedia information to be processed and each multimedia negative feedback data in the multimedia negative feedback database to obtain the second similarity set; Calculate the third similarity between the multimedia information to be processed and each multimedia crawled data in the multimedia playback database to obtain the third similarity set.

10. The method according to claim 9, characterized in that, The step of calculating the target risk value corresponding to the multimedia information to be processed based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set includes: The first risk value is obtained by summing and averaging the first similarity values ​​that are greater than the similarity threshold and the initial risk values ​​corresponding to the first similarity values. The second risk value is obtained by summing and averaging the second similarity values ​​that are greater than the similarity threshold and the initial risk values ​​corresponding to the second similarity values. Risk reduction weights are obtained based on the third similarity, and these risk reduction weights are used as risk factors for the multimedia information to be processed. The target risk value is determined based on the first risk value, the second risk value, and the risk factor.

11. The method according to claim 10, characterized in that, Determining the target risk value based on the first risk value, the second risk value, and the risk factor includes: Obtain the sum of the first risk value and the second risk value; Calculate the product between the sum and the risk factor, and use the product as the target risk value.

12. The method according to claim 1, characterized in that, The multimedia information review criteria include hitting risk review keywords; determining the review result of the multimedia information to be processed based on the target risk value and the multimedia information review criteria includes: If the target risk value is greater than the risk threshold, or the number of times the risk audit keywords are hit is greater than the quantity threshold, then the audit result corresponding to the multimedia information to be processed is not pushable. If the target risk value is less than or equal to the risk threshold, and the number of times the risk audit keywords are hit is less than or equal to the quantity threshold, then the audit result corresponding to the multimedia information to be processed can be pushed.

13. A multimedia information push device, characterized in that, include: The acquisition unit is used to acquire multimedia information to be processed. The processing unit is used to calculate the similarity between the multimedia information to be processed and the data in the multimedia risk database, the multimedia negative feedback database and the multimedia playback database, respectively, to obtain a first similarity set, a second similarity set and a third similarity set; The acquisition unit is further configured to acquire the initial risk values ​​corresponding to the multimedia risk database and the multimedia negative feedback database respectively. The initial risk value corresponding to the multimedia risk information is calculated by acquiring the exposure, clicks and consumption corresponding to each multimedia risk information in the multimedia risk database. The initial risk value corresponding to the multimedia negative feedback data is calculated based on the negative feedback amount and negative feedback height of the multimedia negative feedback data. The processing unit is further configured to calculate the target risk value corresponding to the multimedia information to be processed based on the initial risk value, the first similarity set, the second similarity set, and the third similarity set; The determining unit is used to determine the review result of the multimedia information to be processed based on the target risk value and the multimedia information review conditions; The processing unit is further configured to push the multimedia information to be processed to a multimedia information receiving object if the review result is pushable.

14. A computer device comprising a memory, a processor, and a bus system, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 12; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 12.

16. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 12.

Images