Information matching method and device, computer device and storage medium
Patent Information
- Application Number
- CN202510487165.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-17
- Publication Date
- 2026-10-23
AI Technical Summary
采用传统的信息匹配方法,匹配得到的待推送信息与用户观看过的历史信息高度相似,无法满足用户兴趣的多样性需求
[0020]上述信息匹配方法、装置、计算机设备、计算机可读存储介质和计算机程序产品,获取目标对象针对多个历史信息分别产生的交互行为记录;对各历史信息各自的交互行为记录进行统计分析,确定各历史信息分别对应的等效交互次数;从各历史信息中,筛选感兴趣程度满足感兴趣条件的偏好信息。通过等效交互次数表征目标对象对历史信息的感兴趣程度,能够实现对感兴趣程度的量化分析,确保基于感兴趣程度所确定的偏好信息可以真实反映目标对象的兴趣。根据各偏好信息各自的质量指标数据,确定目标对象针对质量指标的偏好条件,并将质量指标数据与偏好条件适配的候选信息,确定为目标对象的待推送信息,相当于在进行信息匹配过程中,还考虑目标对象针对质量指标的偏好,能够基于多个维度的数据信息,确定目标对象的待推送信息,有利于进一步提高待推送信息与目标对象的适配度。因此,采用上述的信息匹配方法,能够提高匹配结果的准确性。
Smart Images

Figure CN122964344A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and in particular to an information matching method, apparatus, computer device, computer-readable storage medium, and computer program product. Background Technology
[0002] With the development of computer technology, more and more users are obtaining information through the internet. Driven by the strong information demands of internet users, information service platforms strive to prevent user churn by continuously providing massive amounts of information resources. However, faced with this vast amount of information, users struggle to quickly and effectively extract the information that meets their needs, leading to information overload. To alleviate this problem, information matching methods have emerged.
[0003] Traditional technologies use a user's viewed history as their preference information, then filter for information resources similar to that preference information to push to that user. However, traditional information matching methods often result in highly similar push notifications to the user's viewed history, failing to meet the diverse needs of users. Therefore, traditional information matching methods suffer from low accuracy. Summary of the Invention
[0004] Therefore, it is necessary to provide an information matching method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the accuracy of matching results in response to the above-mentioned technical problems.
[0005] Firstly, this application provides an information matching method. The method includes:
[0006] Acquire records of interaction behaviors of the target object in response to multiple historical information sources;
[0007] Statistical analysis is performed on the interaction behavior records of each of the historical information entries to determine the equivalent number of interactions corresponding to each of the historical information entries; the equivalent number of interactions is used to characterize the degree of interest of the target object in the historical information.
[0008] From the historical information provided, filter out preference information that meets the interest criteria;
[0009] Based on the quality index data of each of the aforementioned preference information, the preference conditions of the target object for the quality index are determined;
[0010] Candidate information that matches the quality indicator data with the preference conditions is determined as the information to be pushed to the target object.
[0011] Secondly, this application also provides an information matching device. The device includes:
[0012] The acquisition module is used to acquire records of the interaction behaviors of the target object in response to multiple historical information.
[0013] The equivalent interaction count determination module is used to perform statistical analysis on the interaction behavior records of each of the historical information to determine the equivalent interaction count corresponding to each of the historical information; the equivalent interaction count is used to characterize the degree of interest of the target object in the historical information;
[0014] The preference information filtering module is used to filter preference information that meets the interest criteria from the historical information.
[0015] The preference condition determination module is used to determine the preference conditions of the target object for the quality indicators based on the quality indicator data of each of the preference information.
[0016] The module for determining information to be pushed is used to determine the candidate information that matches the quality indicator data with the preference conditions as the information to be pushed to the target object.
[0017] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the aforementioned information matching method.
[0018] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the aforementioned information matching method.
[0019] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the aforementioned information matching method.
[0020] The aforementioned information matching method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire records of interaction behaviors generated by a target object in response to multiple historical information sources; statistically analyze the interaction behavior records for each historical information source to determine the equivalent number of interactions corresponding to each historical information source; and filter preference information whose degree of interest meets the interest criteria from each historical information source. By characterizing the target object's degree of interest in historical information through the equivalent number of interactions, a quantitative analysis of the degree of interest can be achieved, ensuring that the preference information determined based on the degree of interest can truly reflect the target object's interests. Based on the quality index data of each preference information source, the target object's preference conditions for the quality indicators are determined, and candidate information whose quality index data matches the preference conditions is identified as the information to be pushed to the target object. This is equivalent to considering the target object's preference for quality indicators during the information matching process, enabling the determination of the information to be pushed to the target object based on multi-dimensional data information, which is beneficial for further improving the fit between the information to be pushed and the target object. Therefore, adopting the above information matching method can improve the accuracy of the matching results. Attached Figure Description
[0021] Figure 1 This is a diagram illustrating the application environment of the information matching method in one embodiment;
[0022] Figure 2 This is a schematic diagram of the information matching process in one embodiment;
[0023] Figure 3 This is a flowchart illustrating an information matching method in one embodiment;
[0024] Figure 4 This is a schematic diagram illustrating the process of determining preference conditions for quality indicators in one embodiment;
[0025] Figure 5 This is a schematic diagram illustrating the process of determining preference conditions for quality indicators in another embodiment;
[0026] Figure 6 This is a schematic diagram of a page that pushes a video to a target object via a client in one embodiment;
[0027] Figure 7 This is a schematic diagram of a page pushing a video to a target object in one embodiment;
[0028] Figure 8 This is a schematic diagram of a page that pushes a video to a target object in another embodiment;
[0029] Figure 9 This is a flowchart illustrating the information matching method in another embodiment;
[0030] Figure 10This is a structural block diagram of an information matching device in one embodiment;
[0031] Figure 11 This is an internal structural diagram of a computer device in one embodiment;
[0032] Figure 12 This is a diagram of the internal structure of a computer device in another embodiment. Detailed Implementation
[0033] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0034] The information matching method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 110 communicates with server 120 via a network. This communication network can be a wired network or a wireless network. Therefore, terminal 110 and server 120 can be directly or indirectly connected via wired or wireless communication. For example, terminal 110 can be indirectly connected to server 120 via a wireless access point, or terminal 110 can be directly connected to server 120 via the Internet; this application does not impose any limitations on this.
[0035] The terminal 110 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. This application can be applied to information push scenarios. A client related to information push can be installed on the terminal 110. This client can include applications that require separate installation, or it can include mini-programs or web pages that can be used without downloading. The server 120 is the backend server corresponding to the client, or a server specifically used for information matching.
[0036] Furthermore, server 120 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, CDN (Content Delivery Network), and big data and artificial intelligence platforms. The data storage system can store the data that server 120 needs to process. The data storage system can be set up separately, integrated into server 120, or placed in the cloud or on other devices.
[0037] It should be noted that the information matching method in this embodiment can be executed by either terminal 110 or server 120 alone, or by both terminal 110 and server 120. For example... Figure 1 As shown, terminal 110 can be a terminal held by the target object. The target object can interact with multiple historical information displayed on terminal 110 through terminal 110, generating interaction behavior records. During the information matching process, server 120 can obtain the interaction behavior records generated by the target object using terminal 110 for multiple historical information, perform statistical analysis on the interaction behavior records of each historical information, determine the equivalent number of interactions corresponding to each historical information; filter the preference information that meets the interest conditions from each historical information; determine the target object's preference conditions for the quality indicators based on the quality index data of each preference information; and determine the candidate information that matches the quality index data with the preference conditions as the information to be pushed to the target object. Among them, the equivalent number of interactions is used to characterize the target object's degree of interest in the historical information. The information to be pushed can be further pushed to the target object through terminal 110 so that the target object can perform subsequent interactions.
[0038] For example, such as Figure 2 As shown, server 120 can determine the equivalent interaction count 1 of historical information 1, the equivalent interaction count 2 of historical information 2, ..., the equivalent interaction count M of historical information M based on the interaction behavior records. Then, based on these equivalent interaction counts, server 120 filters preference information from the M historical information pieces whose interest levels represented by the equivalent interaction counts meet the interest conditions. Next, server 120 can determine the target object's preference conditions for quality indicators based on the quality indicator data of each preference information, i.e., the target object's quality indicator preferences. Finally, the server can determine the information to be pushed from multiple candidate information pieces whose quality indicator data matches the preference conditions; this is the information to be pushed to the target object.
[0039] In one embodiment, an information matching method is provided, which can be applied to a computer device, which can be a terminal or a server. This method is applied to... Figure 1 Taking server 120 as an example, in this embodiment, as... Figure 3 As shown, the method includes the following steps:
[0040] Step S302: Obtain the interaction behavior records generated by the target object in response to multiple historical information.
[0041] The target audience refers to the object that needs information recommendation based on the information matching results. Historical information refers to information that has been pushed to the target audience and that the target audience has interacted with. The specific content format and topic type of this historical information are not unique. Optionally, content formats include, but are not limited to, advertisements, articles, news, music, videos, and applications. Multiple historical information items with the same content format may belong to the same or different topic types. For example, for multiple historical information items with video content, their topic types may include drama, food, beauty, knowledge, etc.; for multiple historical information items with news content, their topic types may include social news, entertainment news, etc. In other words, the topic type reflects the semantics of the content displayed by the information, and the content format reflects the specific data format of the content displayed by the information, including format, media, etc.
[0042] Interaction behavior records are used to document the interaction behaviors of a target audience with historical information. Interaction behavior refers to the actions a target audience takes during their interaction with historical information. Classifying interaction behaviors based on preference dimensions determines the interaction category. This category can specifically include positive interactions (i.e., showing interest or liking) and negative interactions (i.e., showing disinterest or dislike). Classifying interaction behaviors based on interaction methods determines the behavior type, which can specifically include liking, saving, commenting, and playing. It's understandable that the same behavior type can belong to one interaction category; for example, disliking, unfollowing, deleting, and uninstalling are all negative interactions, while liking, saving, and downloading are all positive interactions. The same behavior type can also belong to multiple different interaction categories; for example, comments can include positive and negative comments, with positive comments being positive and negative comments being negative. In other words, interaction categories reflect the target audience's preference for historical information, while behavior types reflect the target audience's interaction methods with that information.
[0043] Specifically, the target user can use their terminal to interact with the information displayed on that terminal. During this process, the terminal or a server can record the interaction behavior, generating a record of the target user's interaction behavior with respect to historical information. The server generating the interaction behavior record can be the backend server of a content service platform that provides content services to the terminal. Thus, the server can obtain the target user's interaction behavior records for multiple pieces of historical information from the terminal or at least one server.
[0044] In one embodiment, the server can obtain the interaction behavior records of the target object in response to multiple historical information when the client used to display information on the terminal is open. Optionally, with the authorization of the target object, the server can obtain the interaction behavior records of the target object in response to multiple historical information on at least one client, and then determine the information to be pushed to the target object by analyzing the interaction behavior records. The at least one authorized client can include the client that is open and other clients. That is, the historical information and the candidate information to be matched can be information pushed through the same client or information pushed through different clients. In a specific implementation, when the target object uses client A for the first time, client A usually does not have the target object's interaction behavior records. At this time, the interaction behavior records of the target object generated on other clients can be obtained to match the candidate information of client A. From the candidate information, the information to be pushed to the target object through client A can be determined, thereby overcoming the problem that interaction behavior records cannot be obtained on the first launch and improving the accuracy of the information matching results.
[0045] Optionally, historical information and candidate information requiring matching can belong to the same content format. For example, both historical information and candidate information may be videos, or both may be music. Historical information and candidate information can also belong to different content formats, and both may contain information content corresponding to the same sensory medium. This sensory medium can include text, images, sound, etc. For example, if the candidate information is a video, the historical information can include audio and images.
[0046] Step S304: Perform statistical analysis on the interaction behavior records of each historical information to determine the equivalent number of interactions corresponding to each historical information.
[0047] The equivalent interaction count is used to characterize the target object's level of interest in historical information. The more equivalent interactions there are, the higher the target object's level of interest in that historical information. In other words, the equivalent interaction count is not the same as the physical interaction count; it should not be interpreted as the actual number of interactions, but rather as a quantitative indicator of the degree of interest. Therefore, the obtained equivalent interaction count does not have to be an integer, and can even be negative; the value of the equivalent interaction count only needs to be sufficient to quantify the degree of interest.
[0048] Specifically, the server can determine the interaction behavior of the target object for each piece of historical information and the time of each interaction behavior from the interaction behavior record. Then, it can further obtain the equivalent number of interactions corresponding to each piece of historical information by statistics.
[0049] In an optional embodiment, the server can count multiple interactions related to the same historical information within each predefined statistical period to determine the number of interactions corresponding to that historical information in each statistical period. Then, it can further count the number of interactions in each statistical period to determine the equivalent number of interactions for that historical information. The statistical period can be, for example, a week or a day; the specific method for counting multiple interactions can be, for example, calculating the average or the maximum value.
[0050] In an optional embodiment, the server can determine the behavior type of each interaction and the interaction category to which each interaction belongs, and perform weighted summation statistics on the interaction data of the same behavior type according to the weights matched with the interaction category and behavior type to determine the equivalent number of interactions corresponding to the historical information.
[0051] In this system, the direction of the weight corresponds to the interaction category, and the magnitude of the weight is positively correlated with the interaction duration corresponding to the behavior type. For example, positive interactions correspond to positive weights, and negative interactions correspond to negative weights. In positive interactions, the weight of a click is less than the weight of a positive review; in negative interactions, the weight of an uninstall is less than the weight of a negative review. Different behavior types can correspond to different interaction data. For example, interaction data for a "like" behavior can include the number of likes and the frequency of likes; interaction data for a "comment" behavior can include the number of comments and the number of words in the comments; interaction data for a "play" behavior can include the number of plays, the playback duration, and the playback speed. Optionally, interaction data of different dimensions can be normalized to obtain normalized data, which can then be further statistically analyzed according to weights to obtain the equivalent number of interactions.
[0052] Step S306: Filter the preference information from the historical information that meets the interest criteria.
[0053] The degree of interest meeting the interest condition can refer to the equivalent number of interactions being greater than or equal to a threshold. Specifically, given that the equivalent number of interactions for each piece of historical information has been determined, since the equivalent number of interactions can be used to characterize the target object's degree of interest in the historical information, the server can further filter preference information from each piece of historical information that meets the interest condition.
[0054] In an optional embodiment, the server can determine the number of times each historical information was pushed, and filter out preference information from each historical information that meets the interest criteria based on the number of pushes. For example, a relatively large number of pushes can be set for historical information with a high number of pushes, while a relatively small number of pushes can be set for historical information with a low number of pushes. This can help to uncover the true interests of the target object driven by internal factors and further improve the accuracy of the matching results.
[0055] Step S308: Determine the target object's preference conditions for the quality indicators based on the quality indicator data of each preference information.
[0056] Quality indicator data refers to data that reflects the quality of information. Quality indicators are metrics used to evaluate information quality. In practical applications, corresponding quality indicators can be set from multiple dimensions to evaluate information quality. Specifically, the server can obtain the quality indicator data for each preference information and determine the quality indicators that the obtained quality indicator data reflects. Therefore, based on this quality indicator data, the target object's preference conditions for that quality indicator can be determined.
[0057] In an optional embodiment, the quality indicator data may include parameter values corresponding to multiple information parameters. These information parameters are technical parameters reflecting information quality. The quality indicator data serves as the data foundation for quality evaluation results under the quality indicators.
[0058] Specifically, the value of each of the multiple information parameters can reflect the information quality of preference information under one or more quality indicators. For example, when the target object is audio, quality indicators can include sound quality and fidelity. Information parameters corresponding to sound quality can include signal-to-noise ratio, sampling frequency, quantization bits, etc., while information parameters corresponding to fidelity can include dynamic range, distortion, number of channels, etc. When the target object is video, quality indicators can include sharpness and smoothness. Information parameters corresponding to sharpness can include resolution, frame rate, noise, color space, etc., while information parameters corresponding to smoothness can include bitrate, latency, encoding format, etc. It can be understood that the same information parameter can be used to reflect one or more quality indicators; for example, frame rate can be used to reflect the sharpness and smoothness of a video.
[0059] In an optional embodiment, such as Figure 4As shown, the server can determine the target object's preference parameter range for each information parameter by statistically analyzing the parameter values corresponding to each information parameter, and further determine the preference conditions for the quality indicator represented by these preference parameter ranges. In other words, the target object's preference conditions for the quality indicator include the preference parameter ranges corresponding to multiple information parameters for the target object's preference information; these information parameters reflect the quality indicator. Taking the case where historical information is video, the quality indicator is resolution, and the quality indicator data includes resolution and frame rate as an example, the server can determine the preferred resolution range for each preference information by statistically analyzing the resolution of each preference information, and determine the preferred frame rate range for each preference information by statistically analyzing the frame rate of each preference information. Then, the preferred frame rate range and preferred resolution range are used as the target object's preference conditions for video resolution.
[0060] In an optional embodiment, such as Figure 5 As shown, the server can determine the index value of each preference information under the quality index reflected by each information parameter based on the parameter value of the preference information corresponding to each information parameter. Then, the server obtains the index preference of the target object for the quality index by statistically analyzing the index values of each preference information, and determines the preference conditions represented by the index preference.
[0061] The specific method for determining the index value of the preference information under the quality indicators reflected by each information parameter, based on the parameter values corresponding to the preference information, can be a quantitative analysis based on contribution. Optionally, the server can standardize the parameter values according to the contribution of each information parameter to the quality indicator, ensuring measurement consistency among the parameter values, and then further sum or average the standardized parameter values to determine the index value of the preference information under the quality indicators reflected by each information parameter.
[0062] Taking historical information as video, clarity as the quality metric, and resolution and frame rate as the quality metric data as an example, the server can determine the clarity of each preference information based on its resolution and frame rate. Then, by statistically analyzing the clarity of each preference information, the server can obtain the target object's clarity preference range for video clarity and further determine the preference conditions represented by the clarity preference range.
[0063] In an optional embodiment, the quality indicator data includes parameter values corresponding to multiple information parameters. In this embodiment, based on the quality indicator data of each preference information, the target object's preference conditions for the quality indicators are statistically obtained, including: obtaining the data range configured for each information parameter; for each information parameter, based on the parameter value corresponding to each preference information, statistically obtaining the target object's preference value for that information parameter; and determining the preference conditions for the quality indicators reflected by each information parameter based on the preference value and data range of each information parameter.
[0064] The data range can include directions indicated by positive and negative signs, and ranges represented by numerical values. Specifically, the server can obtain the data range configured for each information parameter. For example, the data range configured for resolution is "-X", and the data range configured for frame rate is "+Y". Then, for each information parameter, the server can statistically obtain the target object's preference value for that information parameter based on the parameter values corresponding to each preference information. This preference value can be, for example, the average, median, or cluster center. For example, the target object's preference value for resolution is A, and its preference value for frame rate is B. Finally, based on the preference values and data ranges of each information parameter, the server determines the preference parameter range for each information parameter and further determines the preference conditions that include the preference parameter ranges. These preference conditions are the preference conditions for the quality indicators reflected by each information parameter. For example, the target object's preference conditions for video clarity are a resolution within the range (AX, A) and a frame rate within the range (B, B+Y).
[0065] In the above embodiments, by combining the preference value with the numerical range, the target object's preference conditions for quality indicators can be determined, which can improve the robustness and adaptability of the preference conditions, thereby improving the accuracy of the information to be pushed based on the preference conditions.
[0066] Step S310: The candidate information that matches the quality indicator data with the preference conditions is determined as the information to be pushed to the target object.
[0067] Candidate information refers to information that needs to be determined whether it should be recommended to the target audience. The server can determine the information to be pushed to the target audience from multiple candidate information. Specifically, the server can obtain the quality indicator data of each candidate information and match each quality indicator data with the determined preference conditions. If the quality indicator data of the candidate information matches the preference conditions, then the candidate information is determined as the information to be pushed to the target audience.
[0068] Taking the scenario where historical information is video, the quality metric is resolution, and the quality metric data includes resolution and frame rate as an example, the server can obtain the resolution and frame rate of each candidate video. If the preference conditions include a preferred frame rate range and a preferred resolution range, the server can identify the candidate information as the target information to be pushed if both the candidate video's resolution and frame rate are within the preferred range. If the preference conditions include a resolution preference range, the server can calculate the resolution of the candidate information based on its resolution and frame rate, and identify the candidate information as the target information to be pushed if its resolution is within the resolution preference range.
[0069] In an optional embodiment, there are multiple quality indicators. For each quality indicator, the server can determine the target object's preference range for that quality indicator by statistically analyzing the parameter values of each information parameter corresponding to that quality indicator. Furthermore, by combining the preference ranges of each quality indicator, the server can determine the target object's preference conditions for that quality indicator. The server can then determine the candidate information that matches the quality indicator data with the preference conditions as the information to be pushed to the target object.
[0070] For example, after determining the target audience's preferred ranges for video clarity and video smoothness, the server can further determine preference conditions that encompass each preference range. Then, the server can identify candidate videos whose clarity and smoothness fall within the preferred ranges as matching the preference conditions, and further designate these candidate videos as the videos to be pushed to the target audience.
[0071] Optionally, the information to be pushed can be sent to the target object through that client if the target object initiates a triggering operation against the client. For example... Figure 6 As shown, when the information to be pushed is a video, the appropriate video can be pushed to the target audience through a video client.
[0072] Optionally, if the message to be pushed is a video, such as Figure 7As shown, adapted videos can also be pushed to target audiences based on the Over-The-Top (OTT) service model. OTT is a service model that delivers video, audio, messaging, and live streaming content directly to users via the internet. It bypasses traditional cable or satellite TV distribution channels, allowing viewers to watch content anytime, anywhere, on any device. This service model has developed rapidly in recent years and has become an important part of modern media consumption. The core of OTT is delivering content via the internet, allowing users to access video, audio, and messaging content without relying on traditional television networks or cable services.
[0073] Optional, such as Figure 8 As shown, the information to be pushed can be sent to the target object via a desktop computer.
[0074] The aforementioned information matching method acquires the interaction records of the target object in response to multiple historical information sources; statistically analyzes the interaction records of each historical information source to determine the equivalent number of interactions corresponding to each historical information source; and filters out preference information whose interest level meets the interest criteria from each historical information source. By characterizing the target object's interest level in historical information through the equivalent number of interactions, a quantitative analysis of interest level can be achieved, ensuring that the preference information determined based on interest level truly reflects the target object's interests. Based on the quality index data of each preference information source, the target object's preference conditions for quality indicators are determined, and candidate information whose quality index data matches the preference conditions is identified as the information to be pushed to the target object. This is equivalent to considering the target object's preference for quality indicators during the information matching process, enabling the determination of the information to be pushed to the target object based on multi-dimensional data information, which helps to further improve the fit between the information to be pushed and the target object. Therefore, adopting the above information matching method can improve the accuracy of the matching results.
[0075] In one embodiment, determining candidate information that matches quality indicator data with preference conditions as information to be pushed to the target object includes: determining selected information that matches quality indicator data with preference conditions from each candidate information; and determining the selected information as information to be pushed to the target object if the selected information has the same topic type as at least one preference information.
[0076] Specifically, the server can obtain the quality index data of each candidate information to determine the selected information whose quality index data matches the preference conditions, completing the first round of information filtering. Then, a second round of filtering is performed based on topic type. If the selected information has the same topic type as at least one preferred information, that selected information is designated as the target information to be pushed to. For example, if the preferred videos only include beauty videos and food videos, if the selected video that matches the preference conditions is a beauty video or a food video, then that selected video will be the target video to be pushed to.
[0077] In the above embodiments, by combining topic type and quality preference, the information to be pushed to the target object is selected from multiple candidate information. This can take into account the adaptability of both content and quality dimensions, and obtain the information to be pushed to the target object, which is conducive to further improving the accuracy of the information to be pushed.
[0078] In one embodiment, obtaining the interaction behavior records of the target object in response to multiple historical information includes: determining a historical time range in which the time difference with the current time is less than or equal to a time difference threshold; and obtaining the interaction behavior records of the target object in response to multiple historical information within the historical time range.
[0079] In practical applications, a server can define a historical time range and further acquire records of the target object's interaction behavior with multiple pieces of historical information within that time range. Specifically, the server can determine a historical time range where the time difference from the current time is less than or equal to a time difference threshold, and then acquire records of the target object's interaction behavior with multiple pieces of historical information within that time range. For example, the server can acquire records of the target object's interaction behavior with multiple pieces of historical information over the past two months.
[0080] In the above embodiments, obtaining interaction behavior records within a historical time range can avoid the inability to accurately represent the current interests of the target object due to the interaction time being too long, which is conducive to further improving the accuracy of recommendation information determined based on interaction behavior records.
[0081] It should be noted that there is no single way to determine the time difference threshold.
[0082] In an optional embodiment, the information matching method further includes: determining the information categories to which the multiple historical information pushed to the target object belongs; and determining the time difference threshold for matching each historical information according to the content update frequency of each information category.
[0083] Information categories can be characterized by at least one of content format or topic type. Content format reflects the specific data form of the information, including format and medium, while topic type reflects the semantics of the information. Optionally, content formats include, but are not limited to, advertisements, articles, news, music, videos, and applications. Multiple historical pieces of information with the same content format may belong to different topic types. For example, video topic types may include drama, food, beauty, and knowledge, while news topic types may include social news and entertainment news.
[0084] It's understandable that different information categories have different content update frequencies. In terms of content format, news and short videos update more frequently, while articles and applications update more slowly. In terms of topic type, drama and knowledge-based content update more frequently, while food and beauty content update more slowly. The time difference threshold is inversely correlated with content update frequency. An inverse correlation between two variables means that, assuming other variables remain constant, both variables change in the same direction, but a change in one variable does not necessarily require a change in the other.
[0085] Specifically, the server can determine the information categories to which multiple historical information messages pushed to the target object belong, and then determine the time difference threshold for matching each historical information message according to the content update frequency of each information category. For example, if the historical information is video, the server can obtain the target object's interaction records for multiple TV series videos in the past Q1 days, and obtain the target object's interaction records for multiple beauty videos in the past Q2 days, where Q1 is greater than Q2.
[0086] In the above embodiments, for historical information with a relatively fast content update frequency, a smaller time difference threshold is matched to capture the short-term interests of the target object. For historical information with a relatively slow content update frequency, a longer time difference threshold is matched to capture the long-term interests of the target object. This achieves a balance between the amount of data recorded in the interaction behavior and the accuracy of the matching results, thereby improving the scientific nature of the information matching method.
[0087] In an optional embodiment, the information matching method further includes: determining a client for displaying candidate information; and determining a time difference threshold for matching the target object based on the frequency of the target object's use of the client.
[0088] In practical applications, servers can also determine the time difference threshold based on the client's usage frequency. Specifically, the server can identify the client used to display candidate information and its usage frequency, and further determine the time difference threshold for matching with the target object based on this usage frequency. The time difference threshold is inversely correlated with usage frequency. That is, for clients with low usage frequency, there may be insufficient historical data. By appropriately increasing the time difference threshold, a sufficient number of interaction records can be obtained to provide initial recommendation criteria.
[0089] It should be noted that for first-time users, the server, with authorization from the target object, can retrieve records of the target object's interactions with other clients on multiple historical data points. The time difference thresholds for each of these other clients can be determined based on usage frequency.
[0090] In the above embodiments, determining the time difference threshold based on the client's usage frequency ensures that a sufficient number of interaction behavior records are obtained, providing a data foundation for subsequent information matching.
[0091] In one embodiment, statistical analysis is performed on the interaction behavior records of each historical information to determine the equivalent number of interactions corresponding to each historical information. This includes: for each historical information, determining the various behavior types of the target object towards the historical information and the interaction data corresponding to each behavior type from the interaction behavior records of the historical information; performing statistical analysis on each interaction data to determine the equivalent number of interactions corresponding to each historical information.
[0092] Among them, behavior type is used to reflect the target object's interaction with historical information. Different behavior types can correspond to different interaction data. For example, the interaction data corresponding to liking can include the number of likes and the frequency of likes, the interaction data corresponding to commenting can include the number of words in the comment and the number of comments, and the interaction data corresponding to playing can include the number of plays and the playback duration.
[0093] Specifically, for each piece of historical information, the server can determine the target object's various behavior types and the corresponding interaction data from the interaction records of that historical information. Then, by statistically analyzing each interaction data, the equivalent number of interactions corresponding to each piece of historical information is determined. Optionally, the server can normalize interaction data of different dimensions to obtain normalized data, and then further statistically determine the equivalent number of interactions. Statistically determining the equivalent number of interactions for historical information based on behavior type allows for finer-grained analysis, resulting in more accurate equivalent interaction counts. This improves the ability of the equivalent interaction counts to represent the degree of interest, and further enhances the accuracy of information matching results.
[0094] In an optional embodiment, statistical analysis is performed on each interaction data to determine the equivalent number of interactions corresponding to each piece of historical information, including: determining the weights matched for each behavior type; and performing statistical analysis on the interaction data of each behavior type according to each weight to determine the equivalent number of interactions corresponding to the historical information.
[0095] The numerical value of the weight is positively correlated with the interaction duration corresponding to the behavior type. In other words, the longer the interaction duration, the higher the time cost borne by the target audience during the interaction, and the greater the impact of the interaction on the target audience's level of interest. For example, in positive interactions, the weight value of a click behavior is less than the weight value of a positive review behavior; in negative interactions, the weight value of an uninstall behavior is less than the weight value of a negative review behavior.
[0096] Specifically, the server can determine the weight assigned to each behavior type, and then statistically analyze the interaction data for each behavior type according to its weight to determine the equivalent number of interactions corresponding to the historical information. For example, if video A has 10 likes, 5 favorites, 1 comment, and 15 views, and the weight of the like behavior is 0.5, the weight of the favorite behavior is 0.6, the weight of the comment behavior is 0.8, and the weight of the view behavior is 1, then the equivalent number of interactions for video A is: 10*0.5 + 5*0.6 + 1*0.8 + 15*1 = 33.8.
[0097] In one possible implementation, the numerical direction of the weights can correspond to the interaction category of the interactive behavior. For example, positive interactions correspond to positive weights, and negative interactions correspond to negative weights. Therefore, in the statistical analysis of interaction data, the positive impact of positive interactions on the degree of interest and the negative impact of negative interactions on the degree of interest can be considered simultaneously, thereby improving the accuracy of the determined equivalent interaction count. Taking video A as an example, with 10 likes, 5 dislikes, 2 positive reviews, 1 negative review, and 15 views, if the weight of the like behavior is 0.5, the weight of the dislike behavior is -0.5, the weight of the positive behavior is 0.8, the weight of the negative behavior is -0.8, and the weight of the view behavior is 1, then the equivalent interaction count of video A is: 10*0.5 - 5*0.5 + 2*0.8 - 1*0.8 + 15*1 = 18.3.
[0098] In the above embodiments, statistical analysis is performed on the interaction data of each behavior type according to the weights that are positively correlated with the interaction duration, and the equivalent number of interactions corresponding to the historical information is determined, which can ensure the accuracy of the equivalent number of interactions.
[0099] In one embodiment, determining multiple behavior types of a target object towards historical information and corresponding interaction data for each behavior type from historical information interaction behavior records includes: determining multiple behavior types of a target object towards the historical information and at least one initial interaction data for each behavior type from historical information interaction behavior records; and determining interaction data corresponding to at least a portion of the behavior types based on each initial interaction data.
[0100] Initial interaction data refers to the data recorded in the interaction behavior log. Further interaction data is determined based on this initial interaction data. Behavior type reflects the target object's interaction methods with historical information. The interaction behavior log can record multiple interaction behaviors of the target object with historical information. For the same interaction type, the interaction behavior log can record multiple initial interaction data corresponding to that interaction type. Different initial interaction data can be recorded for different behavior types. For example, for the "like" behavior, the initial interaction data recorded in the interaction behavior log may include the number of likes and the frequency of likes; for the "comment" behavior, the initial interaction data may include the number of comment characters and the number of comments; for the "play" behavior, the initial interaction data may include the number of plays and the playback duration.
[0101] In practical applications, the data recorded in interaction behavior logs can be filtered before use. Specifically, the server can determine multiple behavior types of the target object in relation to that historical information, as well as at least one initial interaction data corresponding to each behavior type, from the interaction behavior logs. Optionally, the server can extract the recorded behavior type from the interaction behavior log based on the behavior type field used to record the behavior type, and extract the initial interaction data corresponding to the behavior type from the interaction behavior log based on the interaction data field used to record the initial interaction data, and the correspondence between the interaction data field and the behavior type field.
[0102] Given at least one initial interaction data point corresponding to each behavior type, the server can determine the interaction data corresponding to at least a subset of behavior types based on the initial interaction data. For example, if the interaction behavior record includes records of liking and disliking video A, the interaction data corresponding to the liking behavior can be cleared to zero, i.e., it is considered that no liking behavior occurred.
[0103] In the above embodiments, initial interaction data is first extracted from the interaction behavior records, and then interaction data corresponding to at least a portion of the behavior types is determined based on each initial interaction data. This is equivalent to processing the recorded initial interaction data a second time to obtain the interaction data, which can ensure the accuracy of the interaction data and thus ensure the accuracy of the information to be pushed based on the interaction data.
[0104] In an optional embodiment, based on each initial interaction data, the interaction data corresponding to at least a portion of the behavior types is determined, including: from each behavior type, determining a first behavior type that corresponds to multiple initial interaction data under the same interaction parameter; for each first behavior type, determining the interaction data of the first behavior type under the interaction parameter by the maximum value of each initial interaction data corresponding to the first behavior type under the interaction parameter.
[0105] Interaction parameters refer to the dimensions for recording data on interactive behaviors, while initial interaction data refers to the specific parameter values recorded in the interaction behavior records for the interaction parameters. For example, interaction parameters for a "like" might include the number of likes and the frequency of likes; interaction parameters for a "comment" might include the number of words in the comment and the number of comments; and interaction parameters for a "play" might include the number of plays and the playback duration.
[0106] Specifically, if a behavior type corresponds to multiple initial interaction data points under the same interaction parameter, then that behavior type is designated as the first behavior type. For example, if the interaction behavior record contains multiple playback durations for the playback behavior, then the playback behavior can be designated as the first behavior type; if the interaction behavior record contains multiple comment character counts for the comment behavior, then the comment behavior can be designated as the first behavior type. Then, for each first behavior type, the server can determine the maximum value among the initial interaction data points corresponding to that first behavior type as the interaction data for that first behavior type under the interaction parameter. For example, the server can determine the longest duration among multiple playback durations as the interaction data for the playback behavior corresponding to the playback duration; the server can determine the maximum value among multiple comment character counts as the interaction data for the comment behavior corresponding to the comment character count.
[0107] In the above embodiments, for a first line type that corresponds to multiple initial interaction data under the same interaction parameter, the maximum value among the initial interaction data corresponding to the first line type under the interaction parameter is determined as the interaction data of the first line type under the interaction parameter. This can avoid data redundancy and improve work efficiency.
[0108] In an optional embodiment, based on each initial interaction data, determining the interaction data corresponding to at least a portion of the behavior types includes: determining a second behavior type configured with an interaction data processing method from each behavior type; and for each second behavior type, processing each initial interaction data corresponding to the second behavior type according to the interaction data processing method configured for the second behavior type to determine the interaction data of the second behavior type.
[0109] The interaction data processing method refers to the processing method for the initial interaction data. For example, if the playback duration corresponding to a playback action is less than or equal to a duration threshold, it can be assumed that the playback action occurred due to an erroneous operation. Therefore, a corresponding interaction data processing method can be configured for the playback action to reset the playback duration less than or equal to the duration threshold to zero.
[0110] In practical applications, the initial interactive data can also be further processed based on the interactive data processing method. Specifically, corresponding interactive data processing methods can be pre-configured for a subset of behavior types. Thus, the server can determine the second behavior type configured with the interactive data processing method from among the various behavior types. Then, for each second behavior type, the initial interactive data corresponding to that second behavior type is processed according to the interactive data processing method configured for that second behavior type to determine the interactive data for that second behavior type.
[0111] In an optional embodiment, interactive data processing can be implemented using a filter function. Specifically, the server can first initialize the filter function, which typically includes setting necessary parameters and configurations. For example, in Java's Filter, the `init` method is called after the filter is instantiated to initialize it. Then, when initial interactive data of the second behavior type corresponding to the filter function exists, the server can generate a corresponding data processing request, and the filter's `doFilter` method is called. In this method, the filter can respond to the data processing request and perform various processing actions, such as validation and logging. Optionally, the filter can respond to the request or simply pass it to the next filter or target resource. In the case of multi-level data processing, filters can form a chain of calls. Each filter processes the requested data and then passes the processed data to the next filter until the last filter has finished processing. In Java's Servlet filters, this can be implemented using a FilterChain. If errors occur during data processing, the filters can handle these errors. For example, in Java's Filter, if the `doFilter` method throws an exception, the exception is passed to the caller of the `doFilter` method. After processing all the data, the filter needs to perform resource release operations. This typically includes closing database connections and releasing resources such as memory.
[0112] In an optional embodiment, the filtering function is provided by the video engine, and therefore applicable to various types of videos. This filtering function is a logic function within the video engine layer. The logic function at the video engine layer is code developed based on the video engine; that is, any video using this video engine can have its filtering function configured in the video client.
[0113] In the above embodiments, for each second behavior type configured with an interactive data processing method, the initial interactive data corresponding to the second behavior type is processed according to the configured interactive data processing method to determine the interactive data of the second behavior type. This allows for flexible configuration of the corresponding interactive data processing method as needed, which is beneficial to improving the flexibility of the data processing process.
[0114] In one specific implementation, the historical information is video, and the second behavior type includes commenting and saving. In this embodiment, according to the interaction data processing method configured for the second behavior type, the initial interaction data corresponding to the second behavior type is processed to determine the interaction data for the second behavior type, including: when the target object publishes multiple semantically similar comment messages on the historical information, the sum of the word counts of each comment message is determined as the comment word count corresponding to the comment behavior; the first number of saves by the target object on the historical information and the second number of saves by the target object on the derivative videos of the historical information are counted to determine the number of saves corresponding to the saving behavior.
[0115] The comment word count is the interaction data for the commenting behavior, and the collection data is the interaction data for the collection behavior. Specifically, when a target object publishes multiple semantically similar comments on historical information, the server can determine the comment word count corresponding to the commenting behavior by summing the word counts of each comment. In other words, the interaction data processing method configured for commenting behavior is: to sum the comment word counts of semantically similar comments to obtain the comment word count for the commenting behavior.
[0116] The system calculates the number of collections corresponding to a collection action by statistically analyzing the first collection count of historical information and the second collection count of derivative videos based on that historical information. Derivative videos can include fan-made videos, edited videos, etc. For example, a fan-made video of a segment from a TV series can be considered a derivative video of that series. The specific methods for calculating the first and second collection counts can include summation, weighted summation, etc. Optionally, in the weighted summation process, the weight of derivative videos is less than the weight of historical information. In other words, the interactive data processing method configured for the collection action is: to calculate the number of collections of historical information and its derivative videos to obtain the number of collections corresponding to the collection action.
[0117] In the above embodiments, corresponding interaction data processing methods are configured for commenting and collecting behaviors, which can reduce the amount of interaction data while ensuring that the interaction behavior of the target object can be accurately represented, thus further improving work efficiency.
[0118] In one embodiment, such as Figure 9 As shown, an information matching method is provided, which can be executed by a computer device, which can be a terminal or a server. The computer device is... Figure 1 Taking a server as an example, in this embodiment, the method includes the following steps:
[0119] Step S901: Determine the information category to which the multiple historical information messages pushed to the target object belong.
[0120] Information categories can be characterized by at least one of content format or topic type. Content format reflects the specific data form of the information, including format and medium, while topic type reflects the semantics of the information. Specifically, the server can read the information push record file to determine multiple historical messages pushed to the target audience, and determine the information category of each historical message based on its content format or topic type.
[0121] For example, regarding historical videos pushed to a target audience, if the information category is represented by content format, it can include videos, short videos, animations, dramas, etc.; if the information category is represented by theme type, it can include drama, food, beauty, etc.; if the information category is represented by both content format and theme type, it can include drama videos, food videos, beauty videos, etc. Here, "drama," "food," and "beauty" are theme types, and "video" is the content format.
[0122] Step S902: Determine the time difference threshold for matching each piece of historical information according to the content update frequency of each information category.
[0123] The time difference threshold is inversely correlated with the content update frequency. This is because different information categories have different content update frequencies. In terms of content format, news and short videos update more frequently, while articles and applications update less frequently. In terms of theme, drama and knowledge-based content update more frequently, while food and beauty-related content update less frequently. Based on this, the server can determine the time difference threshold for matching each piece of historical information according to its respective content update frequency.
[0124] Step S903: For each information category, determine the historical time range in which the time difference with the current time is less than or equal to the time difference threshold corresponding to that information type.
[0125] Specifically, once the time difference thresholds for each information category are determined, the server can further determine the historical time range for each information category where the time difference from the current time is less than or equal to the time difference threshold corresponding to that information type, so that interaction behavior records can be obtained based on the historical time range.
[0126] Step S904: For each information category, obtain the interaction behavior records of the target object for multiple historical information belonging to that information category within the historical time range corresponding to that information category.
[0127] Interaction behavior records are used to record the interaction behavior of the target object with historical information. Interaction behavior refers to the actions generated by the target object during its interaction with historical information. Specifically, the server can obtain interaction behavior records of the target object with multiple pieces of historical information belonging to that information category for each information type, according to different historical time ranges. Since a smaller time difference threshold is used to match historical information with a faster update frequency to obtain interaction behavior records generated within a relatively short historical time range, the short-term interests of the target object can be captured. Conversely, a longer time difference threshold is used to match historical information with a slower update frequency to obtain interaction behavior records generated within a relatively long historical time range, the long-term interests of the target object can be captured. This achieves a balance between the amount of interaction behavior records and the accuracy of the matching results, improving the scientific nature of the information matching method.
[0128] Step S905: For each piece of historical information, determine the target object's various behavior types for the historical information and at least one initial interaction data corresponding to each behavior type from the interaction behavior records of the historical information.
[0129] Initial interaction data refers to the data recorded in the interaction behavior log. Behavior type reflects the target object's interaction methods with historical information. The interaction behavior log can record multiple interaction behaviors of the target object with historical information. For the same interaction type, the interaction behavior log can record multiple initial interaction data corresponding to that interaction type. Different initial interaction data can be recorded for different behavior types. For example, for the "like" behavior, the initial interaction data recorded in the interaction behavior log may include the number of likes and the frequency of likes; for the "comment" behavior, the initial interaction data may include the number of comment characters and the number of comments; for the "play" behavior, the initial interaction data may include the number of plays and the playback duration.
[0130] Step S906: From each behavior type, determine the first behavior type that corresponds to multiple initial interaction data under the same interaction parameters.
[0131] Interaction parameters refer to the dimensions for recording data on interactive behaviors, while initial interaction data refers to the specific parameter values recorded in the interaction behavior records for the interaction parameters. For example, interaction parameters for a "like" might include the number of likes and the frequency of likes; interaction parameters for a "comment" might include the number of words in the comment and the number of comments; and interaction parameters for a "play" might include the number of plays and the playback duration.
[0132] Specifically, if a behavior type corresponds to multiple initial interaction data points under the same interaction parameter, then that behavior type is designated as the first behavior type. For example, if the interaction behavior record contains multiple playback durations for the playback behavior, then the playback behavior can be designated as the first behavior type; similarly, if the interaction behavior record contains multiple comment word counts for the comment behavior, then the comment behavior can be designated as the first behavior type.
[0133] Step S907: For each first behavior type, determine the maximum value among the initial interaction data corresponding to the first behavior type under the interaction parameters as the interaction data of the first behavior type under the interaction parameters.
[0134] Specifically, for each first behavior type, the server can determine the maximum value among the initial interaction data corresponding to that first behavior type as the interaction data for that first behavior type under the interaction parameters. For example, the server can determine the longest duration among multiple playback durations as the interaction data corresponding to the playback duration for the playback behavior; the server can determine the maximum value among multiple comment character counts as the interaction data corresponding to the comment character count for the comment behavior.
[0135] Step S908: Determine the second behavior type configured with interactive data processing method from among the behavior types.
[0136] The interaction data processing method refers to the processing method for the initial interaction data. In practical applications, the initial interaction data can also be further processed based on the interaction data processing method. Specifically, corresponding interaction data processing methods can be pre-configured for certain behavior types. Thus, the server can determine the second behavior type configured with the interaction data processing method from among the various behavior types.
[0137] Step S909: For each second behavior type, process the initial interaction data corresponding to the second behavior type according to the interaction data processing method configured for the second behavior type to determine the interaction data of the second behavior type.
[0138] Optionally, if the playback duration corresponding to the playback action is less than or equal to the duration threshold, it can be assumed that the playback action occurred under erroneous operation. Therefore, a corresponding interactive data processing method can be configured for the playback action to clear the playback duration less than or equal to the duration threshold to zero.
[0139] Optionally, when the target object publishes multiple semantically similar comments based on historical information, the server can sum the word counts of each comment to determine the comment word count corresponding to the comment action. This completes the processing of the initial interaction data (i.e., the word counts of each comment) to obtain the interaction data corresponding to the comment action (i.e., the summed comment word count). In other words, the interaction data processing method configured for the comment action is: summing the comment word counts of semantically similar comments to obtain the comment word count for the comment action.
[0140] Optionally, the server can count the first number of favorites for historical information and the second number of favorites for derivative videos of historical information to determine the number of favorites corresponding to the favorite action. Derivative videos of historical information can include fan-made videos, edited videos, etc. In other words, the interactive data processing method configured for the favorite action is: count the number of favorites for historical information and its derivative videos to obtain the number of favorites corresponding to the favorite action.
[0141] It should be noted that for behavior types other than the first and second behavior types, the initial interaction data obtained can be directly used as the interaction data for subsequent processing.
[0142] Step S910: Determine the weight of each interaction behavior according to its behavior type and interaction category recorded in the interaction behavior record.
[0143] The numerical value of the weight is positively correlated with the interaction duration corresponding to the behavior type; the direction of the weight can correspond to the interaction category. In other words, the longer the interaction duration, the higher the time cost borne by the target object during the interaction, and the greater the impact of the interaction on the target object's level of interest. For example, positive interactions correspond to positive weights, and negative interactions correspond to negative weights; in positive interactions, the weight of a click is less than the weight of a positive review; in negative interactions, the weight of an uninstall is less than the weight of a negative review.
[0144] Step S911: Statistical analysis is performed on each interaction data according to its respective weight to determine the equivalent number of interactions corresponding to the historical information.
[0145] The equivalent interaction count is used to characterize the target audience's interest in historical information. A higher equivalent interaction count indicates a greater level of interest. In other words, this equivalent interaction count is not the same as the physical interaction count and should not be interpreted as the actual number of interactions, but rather as a quantitative indicator of interest. Therefore, the equivalent interaction count does not have to be an integer; it can even be negative. The value of the equivalent interaction count only needs to quantify the degree of interest. During the statistical analysis of interaction data, both the positive impact of positive interactions and the negative impact of negative interactions on the degree of interest are considered to improve the accuracy of the determined equivalent interaction count.
[0146] Step S912: Filter the preference information from the historical information that meets the interest criteria.
[0147] The degree of interest meeting the interest condition can refer to the equivalent number of interactions being greater than or equal to a threshold. Specifically, given that the equivalent number of interactions for each piece of historical information has been determined, since the equivalent number of interactions can be used to characterize the target object's degree of interest in the historical information, the server can further filter preference information from each piece of historical information that meets the interest condition.
[0148] Step S913: Obtain the data range configured for each of the multiple information parameters included in the quality indicator data.
[0149] The data range can include directions indicated by positive and negative signs, as well as ranges represented by numerical values. Specifically, the server can obtain the data range configured for each information parameter. For example, the data range configured for resolution is "-X", and the data range configured for frame rate is "+Y".
[0150] Step S914: For each information parameter, based on the parameter values corresponding to each preference information, the preference value of the target object for the information parameter is statistically obtained.
[0151] The preference value can be, for example, the average, median, or cluster center. For instance, the preference value for the target object is A for resolution, and the preference value for frame rate is B.
[0152] Step S915: Based on the preference values and data ranges of each information parameter, determine the preference conditions of the quality indicators reflected by each information parameter.
[0153] Among these, the preference conditions refer to the preference conditions of the quality indicators reflected by each information parameter. For example, the preference conditions for video clarity for a target object are that the resolution is within the range of (AX,A) and the frame rate is within the range of (B,B+Y).
[0154] Step S916: From the candidate information, determine the selected information that matches the quality indicator data with the preference conditions.
[0155] Specifically, the server can obtain the quality index data of each candidate information in order to determine the selected information whose quality index data matches the preference conditions from the candidate information, thus completing the first round of information filtering.
[0156] Step S917: If the selected information has the same topic type as at least one preferred information, the selected information is determined as the information to be pushed to the target object.
[0157] Specifically, the server can perform a second round of filtering based on topic type. If the selected information has the same topic type as at least one preference item, the selected information will be designated as the target content to be pushed to. For example, if the preferred videos only include beauty videos and food videos, and the selected video that matches the preference conditions is a beauty video or a food video, then the selected video will be used as the target video to be pushed to.
[0158] In the above embodiments, the equivalent number of interactions characterizes the target object's interest in historical information, enabling quantitative analysis of the degree of interest and ensuring that the preference information determined based on the degree of interest accurately reflects the target object's interests. Furthermore, during the information matching process, the target object's preferences for quality indicators are also considered, allowing for the determination of the information to be pushed to the target object based on multi-dimensional data, which helps to further improve the fit between the information to be pushed and the target object. Therefore, the above information matching method can improve the accuracy of the matching results.
[0159] In one embodiment, this application also provides an application scenario for audio push, in which the above-described information matching method is applied. In this application scenario, the target user opens an audio client via a control terminal to search for audio content of interest. During the process of the server determining the audio to be pushed to the target user, it acquires the interaction records of the target user's terminal for multiple historically pushed audio clips, performs statistical analysis on the interaction records of each historical audio clip, and determines the equivalent number of interactions corresponding to each historical audio clip. From the historical audio clips, it filters out preferred audio clips whose level of interest meets the interest criteria. Based on the quality index data of each preferred audio clip, it determines the target user's preference conditions for the quality index. Candidate audio clips whose quality index data matches the preference conditions are determined as the audio to be pushed to the target user. The equivalent number of interactions is used to characterize the target user's level of interest in the historical audio clips. The audio to be pushed can be further pushed to the target user via the audio client so that the target user can perform subsequent interactions.
[0160] In one embodiment, this application also provides a video push application scenario, in which the server can determine the video to be pushed to the target object by applying the above-described information matching method.
[0161] In practical applications, the target audience can watch videos through a video client. This video client can record the target audience's interactive behavior across M historical videos watched over a period of time (e.g., two months). This interactive behavior record documents the target audience's interactions with the videos while watching them, along with the interaction data. This interaction data may include, for example, the number of likes, favorites, comments, and play counts.
[0162] Optionally, the server can process the interaction data to obtain the equivalent number of interactions that characterize the target object's level of interest in the historical video. For example, the server can set weights for different interaction behaviors and perform a weighted sum of the interaction data corresponding to the historical video according to the weights to obtain the equivalent number of interactions of the target object with respect to that historical video.
[0163] Optionally, if the interaction parameters for the same interaction behavior correspond to multiple parameter values, the server can retain the largest parameter value, which is equivalent to obtaining the number of interactions.
[0164] Optionally, given the equivalent interaction counts for each of the M historical videos, the server can filter out K videos with relatively high equivalent interaction counts from the M videos. Based on the video parameters of these K videos, the server can determine the parameter range typically satisfied by videos of interest to the user, and then filter out N videos from all videos on the client that satisfy the aforementioned parameter range. For example, if the resolution is A and the frame rate is B obtained from the statistics of the K videos, the parameter range can include a resolution range of (AX, A) and a frame rate range of (B, B+Y).
[0165] Optionally, the server can select P videos from N videos that match the theme type of M videos as the target videos to be pushed. Once the P videos are obtained, the video client can display these P videos if the target video client performs a trigger operation.
[0166] In the above embodiments, selecting suitable videos for display from the video client based on interaction data and video quality improves the user experience. Classifying interactive behaviors according to the target audience's level of interest in videos allows for more accurate capture of videos that the target audience is interested in, thus making the displayed videos more suitable for the target audience. Determining the final filtered videos based on their video parameters eliminates low-quality videos and provides higher-quality videos for the target audience, further enhancing the video viewing experience.
[0167] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0168] Based on the same inventive concept, this application also provides an information matching apparatus for implementing the information matching method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more information matching apparatus embodiments provided below can be found in the limitations of the information matching method described above, and will not be repeated here.
[0169] In one embodiment, such as Figure 10 As shown, an information matching device is provided, including: an acquisition module 1001, an equivalent interaction count determination module 1002, a preference information filtering module 1003, a preference condition determination module 1004, and a push information determination module 1005, wherein:
[0170] The acquisition module 1001 is used to acquire the interaction behavior records of the target object in response to multiple historical information.
[0171] The equivalent interaction count determination module 1002 is used to perform statistical analysis on the interaction behavior records of each historical information to determine the equivalent interaction count corresponding to each historical information; the equivalent interaction count is used to characterize the target object's interest in the historical information;
[0172] The preference information filtering module 1003 is used to filter preference information that meets the interest criteria from various historical information.
[0173] The preference condition determination module 1004 is used to determine the preference conditions of the target object for the quality indicators based on the quality indicator data of each preference information.
[0174] The push information determination module 1005 is used to determine the candidate information that matches the quality indicator data with the preference conditions as the push information of the target object.
[0175] In one embodiment, the equivalent interaction count determination module 1002 includes: an interaction data determination unit, used to determine, for each piece of historical information, multiple behavior types of the target object towards the historical information from the interaction behavior records of the historical information, and the interaction data corresponding to each behavior type; and a data statistics unit, used to perform statistical analysis on each interaction data to determine the equivalent interaction count corresponding to each piece of historical information.
[0176] In one embodiment, the interaction data determination unit includes: an initial interaction data determination component, used to determine, from the interaction behavior records of historical information, multiple behavior types of the target object in relation to historical information, and at least one initial interaction data corresponding to each behavior type; and an interaction data determination component, used to determine, based on each initial interaction data, interaction data corresponding to at least a portion of the behavior types.
[0177] In one embodiment, the interaction data determination component is specifically used to: determine a first behavior type from each behavior type that corresponds to multiple initial interaction data under the same interaction parameter; and for each first behavior type, determine the maximum value among the initial interaction data corresponding to the first behavior type under the interaction parameter as the interaction data of the first behavior type under the interaction parameter.
[0178] In one embodiment, the interaction data determination component is specifically used to: determine a second behavior type configured with an interaction data processing method from each behavior type; and for each second behavior type, process each initial interaction data corresponding to the second behavior type according to the interaction data processing method configured for the second behavior type to determine the interaction data of the second behavior type.
[0179] In one embodiment, the historical information is video; the second behavior type includes commenting behavior and favorites behavior. In this embodiment, the interaction data determination component is specifically used to: determine the comment word count corresponding to the comment behavior by summing the word counts of each comment message when the target object publishes multiple semantically similar comments on the historical information; and to determine the favorites count corresponding to the favorites behavior by statistically analyzing the first favorites count of the target object on the historical information and the second favorites count of the target object on the derivative videos of the historical information.
[0180] In one embodiment, the data statistics unit is specifically used to: determine the weights matched for each behavior type; the magnitude of the weights is positively correlated with the interaction duration corresponding to the behavior type; and perform statistical analysis on the interaction data of each behavior type according to each weight to determine the equivalent number of interactions corresponding to the historical information.
[0181] In one embodiment, the acquisition module 1001 is specifically used to: determine a historical time range whose time difference with the current time is less than or equal to a time difference threshold; and acquire the interactive behavior records of the target object in response to multiple historical information within the historical time range.
[0182] In one embodiment, the information matching device further includes a time difference threshold determination module, used to: determine the information categories to which the multiple historical information pushed to the target object belongs; determine the time difference threshold for matching each historical information according to the content update frequency of each information category; the time difference threshold is inversely correlated with the content update frequency.
[0183] In one embodiment, the information matching device further includes a time difference threshold determination module, configured to: determine a client for displaying candidate information; determine a time difference threshold for matching the target object based on the frequency of use of the target object on the client; the time difference threshold is inversely correlated with the frequency of use.
[0184] In one embodiment, the quality indicator data includes parameter values corresponding to multiple information parameters. In this embodiment, the preference condition determination module 1004 is specifically used to: obtain the data range configured for each information parameter; for each information parameter, statistically obtain the preference value of the target object for the information parameter based on the parameter value corresponding to each preference information; and determine the preference condition of the quality indicator reflected by each information parameter based on the preference value and data range of each information parameter.
[0185] In one embodiment, the push information determination module 1005 is specifically used to: determine selected information from each candidate information whose quality index data matches the preference conditions; and determine the selected information as the push information of the target object when the selected information has the same topic type as at least one preference information.
[0186] The aforementioned information matching device, during the information matching process, acquires records of interaction behaviors generated by the target object in response to multiple historical information items; performs statistical analysis on the interaction behavior records of each historical information item to determine the equivalent number of interactions corresponding to each historical information item; and filters preference information whose interest level meets the interest criteria from each historical information item. By characterizing the target object's interest level in historical information through the equivalent number of interactions, a quantitative analysis of interest level can be achieved, ensuring that the preference information determined based on interest level truly reflects the target object's interests. Based on the quality index data of each preference information item, the target object's preference conditions for the quality indicators are determined, and candidate information whose quality index data matches the preference conditions is identified as the information to be pushed to the target object. This is equivalent to considering the target object's preference for quality indicators during the information matching process, enabling the determination of the information to be pushed to the target object based on multi-dimensional data information, which helps to further improve the fit between the information to be pushed and the target object. Therefore, adopting the above information matching method can improve the accuracy of the matching results.
[0187] Each module in the aforementioned information matching device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.
[0188] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 11 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data involved in the information matching process. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements an information matching method.
[0189] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 12As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements an information matching method. The display unit of the computer device is used to form a visually visible image. It can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.
[0190] Those skilled in the art will understand that Figure 11 The structure shown in Figure 12 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0191] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method embodiment described above.
[0192] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method embodiment described above.
[0193] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method embodiments described above.
[0194] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant regions and areas. Furthermore, the recipient may choose not to authorize the use of their information and related data, or may refuse or conveniently refuse to receive push notifications.
[0195] In this application, during the actual application of relevant data collection and processing, the informed consent or separate consent of the personal information subject should be obtained in strict accordance with the requirements of relevant local laws and regulations, and subsequent data use and processing should be carried out within the scope of laws and regulations and the authorization of the personal information subject.
[0196] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.
[0197] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0198] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. An information matching method, characterized in that, The method includes: Acquire records of interaction behaviors of the target object in response to multiple historical information sources; Statistical analysis is performed on the interaction behavior records of each of the historical information entries to determine the equivalent number of interactions corresponding to each of the historical information entries; the equivalent number of interactions is used to characterize the degree of interest of the target object in the historical information. From the historical information provided, filter out preference information that meets the interest criteria; Based on the quality index data of each of the aforementioned preference information, the preference conditions of the target object for the quality index are determined; Candidate information that matches the quality indicator data with the preference conditions is determined as the information to be pushed to the target object.
2. The method according to claim 1, characterized in that, The step of statistically analyzing the interaction behavior records of each of the historical information entries to determine the equivalent number of interactions corresponding to each of the historical information entries includes: For each piece of historical information, the target object's various behavior types and the corresponding interaction data for each behavior type are determined from the interaction behavior records of the historical information. Statistical analysis is performed on each of the interaction data to determine the equivalent number of interactions corresponding to each of the historical information.
3. The method according to claim 2, characterized in that, The step of determining, from the interaction behavior records of the historical information, multiple behavior types of the target object in relation to the historical information, and the corresponding interaction data for each behavior type, includes: From the interaction behavior records of the historical information, determine the various behavior types of the target object in relation to the historical information, as well as at least one initial interaction data corresponding to each behavior type; Based on the initial interaction data, determine the interaction data corresponding to at least a portion of the behavior types.
4. The method according to claim 3, characterized in that, The step of determining at least a portion of the interaction data corresponding to each of the initial interaction data includes: From the various behavior types, determine the first behavior type that corresponds to multiple initial interaction data under the same interaction parameters; For each of the first behavior types, the maximum value among the initial interaction data corresponding to the first behavior type under the interaction parameters is determined as the interaction data of the first behavior type under the interaction parameters.
5. The method according to claim 3, characterized in that, The step of determining at least a portion of the interaction data corresponding to each of the initial interaction data includes: From each of the described behavior types, determine the second behavior type configured with an interactive data processing method; For each of the second behavior types, the initial interaction data corresponding to the second behavior type is processed according to the interaction data processing method configured for the second behavior type to determine the interaction data of the second behavior type.
6. The method according to claim 5, characterized in that, The historical information is video; the second behavior type includes commenting behavior and saving behavior; The step of processing each initial interaction data corresponding to the second behavior type according to the interaction data processing method configured for the second behavior type to determine the interaction data of the second behavior type includes: When the target object publishes multiple semantically similar comments on the historical information, the sum of the word counts of each comment is determined as the word count corresponding to the comment behavior; The number of collections corresponding to the collection behavior is determined by statistically analyzing the first number of collections of the target object for the historical information and the second number of collections of the target object for the derivative videos of the historical information.
7. The method according to claim 2, characterized in that, The step of statistically analyzing each of the interaction data to determine the equivalent number of interactions corresponding to each of the historical information includes: Determine the weight assigned to each of the aforementioned behavior types; the magnitude of the weight is positively correlated with the interaction duration corresponding to the behavior type. According to the weights described, statistical analysis is performed on the interaction data of each behavior type to determine the equivalent number of interactions corresponding to the historical information.
8. The method according to claim 1, characterized in that, Retrieve the interaction records of the target object in response to multiple historical information sources, including: Determine the historical time range from which the time difference with the current time is less than or equal to the time difference threshold; Obtain records of the target object's interactive behaviors in response to multiple historical information within the historical time range.
9. The method according to claim 8, characterized in that, The method further includes: Determine the information category to which each of the multiple historical messages pushed to the target object belongs; Based on the content update frequency of each of the aforementioned information categories, a time difference threshold is determined for each of the aforementioned historical information categories; the time difference threshold is inversely correlated with the content update frequency.
10. The method according to claim 8, characterized in that, The method further includes: Determine the client to be used to display the candidate information; Based on the frequency of the target object's use on the client, a time difference threshold is determined to match the target object; the time difference threshold is inversely correlated with the usage frequency.
11. The method according to any one of claims 1 to 10, characterized in that, The quality indicator data includes parameter values corresponding to multiple information parameters; the step of statistically obtaining the target object's preference conditions for the quality indicators based on the quality indicator data of each preference information includes: Obtain the data range configured for each of the aforementioned information parameters; For each of the aforementioned information parameters, based on the parameter values corresponding to each of the aforementioned preference information, the preference value of the target object for the aforementioned information parameter is statistically obtained; Based on the preference values and data ranges of each of the aforementioned information parameters, the preference conditions for the quality indicators reflected by each of the aforementioned information parameters are determined.
12. The method according to any one of claims 1 to 10, characterized in that, The step of determining the candidate information that matches the quality indicator data with the preference conditions as the information to be pushed to the target object includes: From the candidate information, select information that matches the quality index data with the preference conditions; If the selected information has the same topic type as at least one preference information, the selected information is determined as the information to be pushed to the target object.
13. An information matching device, characterized in that, The device includes: The acquisition module is used to acquire records of the interaction behaviors of the target object in response to multiple historical information. The equivalent interaction count determination module is used to perform statistical analysis on the interaction behavior records of each of the historical information to determine the equivalent interaction count corresponding to each of the historical information; the equivalent interaction count is used to characterize the degree of interest of the target object in the historical information; The preference information filtering module is used to filter preference information that meets the interest criteria from the historical information. The preference condition determination module is used to determine the preference conditions of the target object for the quality indicators based on the quality indicator data of each of the preference information. The module for determining information to be pushed is used to determine the candidate information that matches the quality indicator data with the preference conditions as the information to be pushed to the target object.
14. A computer device comprising a memory and a processor, 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.
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.
