Customer content evaluation method and device, intelligent equipment and storage medium

Abstract

The invention provides a broadcast content evaluation method and device, intelligent equipment and a storage medium, and belongs to the field of audio content evaluation.The method comprises the steps that a target evaluation standard is determined based on broadcast information and user information, and the target evaluation standard comprises multiple evaluation dimensions; evaluating the broadcast content based on a plurality of evaluation dimensions in the target evaluation standard to obtain an evaluation result corresponding to the broadcast content; and optimizing the broadcast recommendation content based on the evaluation result. According to the broadcast content evaluation method and device, the intelligent equipment and the storage medium provided by the invention, the accuracy and reliability of broadcast content evaluation can be improved.

Description

Technical Field

[0001] The present disclosure belongs to the field of audio content evaluation, and more specifically, to a podcast content evaluation method, apparatus, smart device and storage medium. Background Art

[0002] With the rapid development of Internet technology, podcasts, as an emerging form of audio media, have gradually become an important channel for people to obtain information and entertainment. However, the massive growth of podcast content has also brought about the problem of difficulty in selection. Users often find it difficult to find content that suits their interests and needs from a large number of podcasts. Therefore, how to effectively evaluate podcast content and optimize podcast playback content based on the evaluation results has become a problem that needs to be solved urgently.

[0003] Most existing podcast content evaluation methods rely on a single customer evaluation dimension, such as user text evaluation or rating. Although this method can reflect user preferences to a certain extent, it lacks unified standards and indicators, resulting in inaccurate and objective evaluation results.

[0004] This shows that there is an urgent need for an accurate and reliable method for evaluating podcast content. Summary of the invention

[0005] The purpose of the present disclosure is to provide a podcast content evaluation method, apparatus, intelligent device and storage medium to improve the accuracy and reliability of podcast content evaluation.

[0006] A first aspect of the embodiments of the present disclosure provides a podcast content evaluation method, comprising: Determine target evaluation criteria based on podcast information and user information, where the target evaluation criteria include multiple evaluation dimensions; Evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content; Optimize podcast recommendations based on evaluation results.

[0007] A second aspect of the embodiments of the present disclosure provides a podcast content evaluation device, including: A standard determination module, used to determine a target evaluation standard based on podcast information and user information, wherein the target evaluation standard includes multiple evaluation dimensions; A podcast evaluation module is used to evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content; The podcast optimization module is used to optimize the podcast recommendation content based on the evaluation results.

[0008] According to a third aspect of an embodiment of the present disclosure, a smart device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above-mentioned podcast content evaluation method when executing the computer program.

[0009] According to a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned podcast content evaluation method are implemented.

[0010] The podcast content evaluation method, apparatus, smart device, and storage medium provided by the embodiments of the present disclosure have the following beneficial effects: The present disclosure determines the target evaluation criteria through podcast information and user information, which can more accurately reflect the user's real needs and expectations for podcast content. The present disclosure evaluates podcast content through multiple evaluation dimensions, which can more comprehensively reflect the quality of podcast content, thereby prompting content creators to improve the quality of podcasts in multiple aspects. The present disclosure can more accurately recommend podcast content based on user preferences and needs through evaluation results, which can improve user experience and improve the accuracy and reliability of podcast content evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A flowchart of a podcast content evaluation method provided by an embodiment of the present disclosure; Figure 2 A structural block diagram of a podcast content evaluation device provided by an embodiment of the present disclosure; Figure 3 A schematic block diagram of a smart device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0013] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present disclosure. However, it should be clear to those skilled in the art that the present disclosure may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obstructing the description of the present disclosure with unnecessary details.

[0014] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0015] Please refer to Figure 1 , Figure 1 A flowchart of a podcast content evaluation method provided in an embodiment of the present disclosure, the method comprising: S101: Determine a target evaluation criterion based on podcast information and user information, where the target evaluation criterion includes multiple evaluation dimensions.

[0016] In this embodiment, podcast information refers to various attribute data related to the podcast itself, which may include: podcast category (such as technology, culture, entertainment, etc.), content length, update frequency, host's professionalism, sound quality, content richness (whether it covers multiple viewpoints, materials, etc.), etc. This information can reflect the characteristics and quality of the podcast itself from an objective perspective.

[0017] User information refers to data related to users who listen to podcasts, which may include: user listening time, user listening frequency, user behavior, etc.

[0018] The target evaluation criteria are the evaluation criteria that meet the podcast information and user information among multiple evaluation criteria. Each evaluation criterion contains different evaluation dimensions, which can be the podcast's sound quality, volume, attractiveness, information richness, content accuracy, content fun, interactivity, user behavior, user evaluation, etc.

[0019] Different podcast information and user information correspond to different evaluation criteria, which is specifically reflected in the differences in evaluation dimensions.

[0020] In this embodiment, the multiple evaluation dimensions may include: a podcast evaluation dimension and a user evaluation dimension. The podcast dimension may be determined through podcast information, and the user dimension may be determined through user information.

[0021] For example, a mapping table may be predetermined, as shown in Table 1.

[0022] Table 1 Mapping table of podcast information, user information and their evaluation dimensions

[0023] As shown in Table 1, a long content duration may be greater than 60 minutes, a high update frequency may be greater than once a week, and a high professional level of the host may be the host's years of experience or whether the host is in a pre-stored data table. The host information in the data table may be set based on experience and may be updated at any time via the Internet.

[0024] In response to the total listening time of the user being greater than the first listening time, it is defined as high listening time; In response to the user's listening frequency being greater than the first listening frequency, it is defined as a high listening frequency; In response to the user information being related to the podcast type, it is defined as high professionalism.

[0025] For example, the first listening time is 20 hours, the first listening frequency may be 4 times a week, the high professionalism of the user may be that the user is engaged in a job related to the podcast type or the user manually adds it when filling in the user information, and whether the user information is related to the podcast type can be judged through a pre-set related library, which may contain various fields and information related to the podcast type. When the same words appear in the user information, it can be defined that the user information is related to the podcast type.

[0026] The above-mentioned values ​​for judging length and height can be set according to actual needs.

[0027] S102: Evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content.

[0028] In this embodiment, after performing specific analysis, scoring or qualitative description on the podcast content according to various evaluation dimensions, a conclusion can be drawn that can overall reflect the quality of the podcast content and whether it meets the requirements. The conclusion can be a specific score value or a qualitative expression such as "excellent", "good", "qualified", "needs improvement", etc.

[0029] The podcast content can be evaluated by quantifying the various evaluation dimensions mentioned above and determining their corresponding weights.

[0030] For example, check whether the facts, data, events, etc. mentioned in the podcast are consistent with authoritative materials. For each content that does not match the authoritative materials, a certain score will be deducted, such as 2-5 points. The specific deduction standard can be set based on experience. Count the number of different knowledge points covered in the podcast and score according to the number, such as 1-2 points for less than 5 knowledge points, 3-4 points for 5-10 knowledge points, and 5 points for more than 10 knowledge points. Use audio analysis tools to determine whether the sound quality is clear and free of interference such as noise. 5 points for perfect sound quality, 1-2 points for minor flaws, and 3-5 points for obvious problems. Measure the average volume of the podcast through an audio measurement tool and compare it with the standard volume range. 3 points for volume within the standard range, 2 points for slightly higher or lower than the standard range, and 1 point or less for obviously too high or too low. Scoring is based on the ratio of the user's average listening time to the total podcast time. If the listening ratio reaches 80% or above, 5 points will be awarded, 60%-80% will get 3-4 points, 40%-60% will get 2 points, and below 40% will get less than 1 point.

[0031] Multiply the quantitative scores of each indicator by its corresponding weight, and then add them up to get the comprehensive evaluation result. For example, if the authenticity of a podcast is 4 points, with a weight of 0.1, the weighted score is 0.4; the professionalism of the content is 3 points, with a weight of 0.2, and the weighted score is 0.6; and so on. Finally, add up all the weighted scores to get the comprehensive evaluation result of the podcast. The weights corresponding to each indicator can be set based on experience.

[0032] S103: Optimizing podcast recommendation content based on evaluation results In this embodiment, the podcast recommendation content is the podcast-related information displayed by the platform to users, suggesting that users listen to it, such as the podcast programs displayed in the recommendation list on the homepage of the podcast application, the podcasts presented in the "Guess You Like" section based on user personalized recommendations, etc. The display order, whether it is recommended to specific users, etc. can all be adjusted and optimized.

[0033] For example, if user A obtains the rating of podcast B as 8.9 and the rating of podcast C as 6.2 through the above calculation, we can continue to recommend podcasts with similar ratings and fields to podcast B to user A, and reduce the number of podcasts with similar ratings and fields to podcast C. And appropriately lower the recommendation priority of podcast C in other podcast platforms that use the same podcast.

[0034] From the above, it can be concluded that the present disclosure can more accurately reflect the user's real needs and expectations for podcast content by determining the target evaluation criteria through podcast information and user information. The present disclosure can more comprehensively reflect the quality of podcast content by evaluating podcast content through multiple evaluation dimensions, thereby prompting content creators to improve the quality of podcasts in multiple aspects. The present disclosure can more accurately recommend podcast content based on user preferences and needs through evaluation results, which can improve user experience and improve the accuracy and reliability of podcast content evaluation.

[0035] In one embodiment of the present disclosure, the target evaluation criteria include: at least one podcast evaluation dimension and at least one user evaluation dimension; The podcast content is evaluated based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content, including: Evaluate the podcast content based on at least one podcast evaluation dimension to obtain a first evaluation result; Evaluate the podcast content based on at least one user evaluation dimension to obtain a second evaluation result; An evaluation result corresponding to the podcast content is determined based on the first evaluation result and the second evaluation result.

[0036] In this embodiment, the target evaluation criteria include at least one podcast evaluation dimension and at least one user evaluation dimension, otherwise it is impossible to perform a relatively objective and reliable content evaluation.

[0037] Podcast evaluation dimensions may include: audio quality, audio volume, content richness, content authenticity, etc. User evaluation dimensions may include: the proportion of listening time to the total podcast time, whether the volume is turned up during listening, whether loop playback is set during listening, etc.

[0038] Audio quality can be obtained by detecting the frequency response and signal-to-noise ratio of the podcast audio. The audio volume can be obtained by analyzing the volume and fluctuation of the podcast audio to see if it is within the preset appropriate volume range. It should be noted that for different types of podcasts, the corresponding appropriate volume ranges are different. The richness of the content can be calculated by the vocabulary and keyword density in the podcast, and the authenticity of the content can be evaluated by official authoritative information. The user's listening time, whether or not they have turned up the volume, and whether or not they have set a loop playback can all be recorded and evaluated by monitoring whether the user has taken corresponding actions.

[0039] The evaluation method can be a scoring system. For example, the actual performance of the audio can be scored. The evaluation can be based on the signal-to-noise ratio of the audio. A mapping table of the signal-to-noise ratio and its corresponding score can be set in advance. The audio volume, content richness, content authenticity, the proportion of listening time to the total podcast time, whether the volume is turned up during the listening process, and whether loop playback is set during the listening process are the same.

[0040] The evaluation results of each evaluation dimension can be added together or performed through weighted calculation. It should be noted that since the evaluation criteria corresponding to different podcast information and different user information are different, the corresponding evaluation dimensions are also different, which may be reflected in the different types of evaluation dimensions or the different numbers of evaluation dimensions. However, when performing weighted calculation, the sum of the weights corresponding to each evaluation dimension should be 1.

[0041] It can be concluded from the above that the present disclosure achieves a more comprehensive, objective and reliable evaluation of podcast content by introducing a target evaluation standard including at least one podcast evaluation dimension and at least one user evaluation dimension, which not only takes into account the characteristics of the podcast content itself, but also fully integrates the user's usage behavior and feedback, thereby significantly improving the accuracy and practicality of the evaluation. This embodiment quantifies the podcast evaluation dimensions through specific evaluation methods and indicators (such as frequency response, signal-to-noise ratio, vocabulary, keyword density, etc.), making the evaluation results more objective and measurable.

[0042] In this embodiment, the podcast content evaluation method further includes: In response to the recording year corresponding to the audio source of the podcast being earlier than the first year, reducing the weight threshold corresponding to the audio quality by a step size of the first year; In response to the type of the podcast being a strong music class, the upper limit and the lower limit of the volume range are expanded by a first volume step.

[0043] In this embodiment, the recording year corresponding to the audio source of the podcast can be obtained through the podcast application. The present disclosure takes into account that the audio quality of the podcast may be affected by many factors, among which the influence of the year is unavoidable. Even if it is repaired through technology, it is difficult to achieve the current requirements of clear audio quality. Therefore, when the recording year of the audio source is earlier than the preset first year, the impact of the audio quality on its evaluation results should be appropriately reduced.

[0044] Among them, the first year and the first year step are both set through experience.

[0045] The types of podcasts are not limited to being differentiated based on the content and plot, but can also be differentiated based on the need for musical ups and downs. For example, for news, bedtime stories, and educational lecture podcasts, there is no need for too much musical ups and downs, while for horror stories and music appreciation podcasts, musical ups and downs are needed to create an atmosphere or they are music themselves, so the corresponding volume ranges are also different.

[0046] For example, horror story podcasts require a sudden increase or decrease in volume to create a dangerous atmosphere, so the corresponding audio range for horror story podcasts should be larger than the audio range of official podcasts such as news.

[0047] The volume interval refers to a range of values ​​between the upper and lower limits of the podcast volume. It should be noted that when evaluating the audio volume, the podcast playback volume should be unified. The playback volume refers to the volume displayed on the terminal device, while the audio volume refers to the actual playback volume. For example, when the audio interval is set, the corresponding playback volume is 3, then when evaluating the volume, the corresponding playback volume should also be 3.

[0048] The first volume step and the volume interval are both set based on experience.

[0049] From the above, it can be concluded that the present disclosure can more accurately reflect the actual status of podcast audio quality in different eras by considering the recording year of the podcast audio source and adjusting the weight threshold corresponding to the audio quality accordingly. Improve the accuracy and reliability of the podcast content evaluation method. The present disclosure is targeted at different types of podcasts. The method adjusts the upper and lower limits of the volume range to accommodate content with different musical fluctuation requirements. This personalized processing method makes the evaluation closer to the actual characteristics of the podcast content, improving the pertinence and effectiveness of the podcast content evaluation.

[0050] In one embodiment of the present disclosure, determining an evaluation result corresponding to the podcast content based on the first evaluation result and the second evaluation result includes: The first evaluation result and the second evaluation result are weightedly calculated to obtain the evaluation result corresponding to the podcast content; wherein the weight corresponding to the first evaluation result is the first weight, and the weight corresponding to the second evaluation result is the second weight, and both the first weight and the second weight are determined based on user information.

[0051] In one embodiment of the present disclosure, the podcast content evaluation method further includes: determining a first initial weight and a second initial weight; Adjusting the first initial weight based on the user's listening time and the user's listening frequency to obtain a first weight; The second initial weight is adjusted based on the user's listening time and the user's listening frequency to obtain a second weight.

[0052] In this embodiment, the first evaluation result is a comprehensive result obtained by evaluating the podcast content based on the podcast evaluation dimension, which can represent the quality of the podcast from the perspective of its own attributes. It is a quantitative summary of the performance of the podcast in terms of content professionalism, information richness, audio quality, etc., and is used to determine the final evaluation result of the podcast content together with the second evaluation result obtained from the user evaluation dimension.

[0053] The second evaluation result is a comprehensive result obtained by evaluating the podcast content based on the user evaluation dimension, which can reflect the user's views and feelings about the podcast content. It can integrate user comments, listening behavior and other information, and reflect the user's recognition and preference for the podcast. Together with the first evaluation result, it serves as an important basis for determining the final evaluation result of the podcast.

[0054] The first weight is a coefficient for weighting the first evaluation result, which represents the relative importance of the podcast evaluation dimension in the final evaluation result. The second weight is a coefficient for weighting the second evaluation result, which represents the relative importance of the user evaluation dimension in the final evaluation result. Similarly, the second weight is also determined based on user information, and its value will be adjusted accordingly based on the user's different behaviors and matching with the podcast, so that the final evaluation result can better reflect the importance of user feedback to the podcast evaluation.

[0055] The first initial weight is a basic weight value that is pre-set before the weight is adjusted based on detailed user information such as the specific user listening time and listening frequency. It is used to indicate the relative importance of the first evaluation result (the objective quality evaluation result of the podcast itself) in the overall evaluation. It is the starting reference point for subsequent weight adjustments based on actual user information. The setting of this initial value can be determined based on experience, the platform's general principles for podcast evaluation, or a general understanding of the target user group.

[0056] The second initial weight corresponds to the first initial weight. It is a pre-given value that represents the initial relative importance of the second evaluation result (user's subjective evaluation result) in the overall evaluation before adjustment based on specific user information. Together with the first initial weight, it constitutes the initial weight distribution scheme. With the in-depth analysis and consideration of user information, these two initial weights will be changed accordingly, so that the final weight distribution is more in line with the actual situation of different users and podcasts, thus making the comprehensive evaluation result more reasonable.

[0057] In this embodiment, the first initial weight and the second initial weight may be adjusted based on the user's listening time and the user's listening frequency in the following manner: In response to the user listening time being greater than or equal to the first time and the user listening frequency being less than the first frequency, the first initial weight is reduced by the first step length to obtain the first weight, and the second initial weight is increased by the first step length to obtain the second weight; In response to the user listening frequency being greater than or equal to the first frequency and the user listening duration being less than the first duration, reducing the first initial weight by a second step length to obtain the first weight, and increasing the second initial weight by a second step length to obtain the second weight; In response to the user listening time being greater than or equal to the first time and the user listening frequency being greater than or equal to the first frequency, reducing the first initial weight by a third step length to obtain a first weight, and increasing the second initial weight by a third step length to obtain a second weight; The first weight is the weight corresponding to the first evaluation result, and the second weight is the weight corresponding to the second evaluation result; The third step length is the sum of the first step length and the second step length.

[0058] In this embodiment, the user's listening time refers to the user's total listening time rather than the listening time for a podcast. This embodiment takes into account that if a user is a new user who has not been exposed to podcasts, the reference value of such user's behavior when listening to podcasts will be correspondingly lower than the reference value of the behavior of old users.

[0059] For example, user D is encountering a podcast for the first time and is not accustomed to the expression form of podcasts. He frequently changes podcast channels, volume, and operation settings. At this time, the reference value of user D's behavior is not high, so the first initial weight can be 0.7 and the second initial weight can be 0.3. When user D's listening time reaches the first time or the listening frequency is greater than the first frequency, user D can be determined to be an old user to a certain extent. The weight corresponding to the user evaluation result (second evaluation result) can be increased accordingly, and the weight corresponding to the podcast evaluation result (first evaluation result) can be reduced, so that the overall evaluation result is more accurate.

[0060] When the listening time of user D is greater than or equal to the first time and the listening frequency of user D is less than the first frequency, the first initial weight can be reduced with the first step to obtain the first weight, and the second initial weight can be increased with the first step to obtain the second weight.

[0061] When the listening time of user D is less than the first time and the listening frequency of user D is greater than or equal to the first frequency, the first initial weight can be reduced by a second step to obtain the first weight, and the second initial weight can be increased by a second step to obtain the second weight.

[0062] When user D's listening time is greater than the first time and the listening frequency is greater than the first frequency, user D is an old user who frequently and continuously listens to podcasts. The first initial weight can be reduced by the third step to obtain the first weight, and the second initial weight can be increased by the third step to obtain the second weight.

[0063] The first step length and the second step length are both set based on experience. The third step length is the sum of the first step length and the second step length.

[0064] It can be concluded from the above that this embodiment adjusts the initial weight according to the user's listening time and listening frequency, so as to obtain a weight distribution that is more in line with the user's personal preferences and needs. This personalized processing method makes the evaluation results closer to the user's actual experience, and improves the pertinence and effectiveness of the evaluation. This embodiment comprehensively considers the podcast evaluation dimension (first evaluation result) and the user evaluation dimension (second evaluation result), and dynamically adjusts their weights based on user information. This method can more accurately reflect the actual quality of the podcast content and the user's recognition of it. The combination of this dual evaluation dimension and the dynamic adjustment of the weights jointly improve the accuracy and reliability of the evaluation.

[0065] In one embodiment of the present disclosure, the podcast content evaluation method further includes: In response to the number of podcast types listened to by the user being less than a first number and the target podcast type not satisfying a user evaluation condition, the target podcast content is evaluated with a first initial weight and a second initial weight.

[0066] In this embodiment, the target podcast is the podcast for which the podcast content is evaluated, and the first number is manually set based on experience. The user evaluation condition is that the type of the target podcast belongs to one of the types of podcasts that the user has listened to in the past.

[0067] Taking into account that some users are only interested in podcasts in a certain field or fields, and do not listen to them universally, the above-mentioned users are still new users for other types of podcasts, and the reference value of their behavior is still low, so the podcast content is still evaluated with the first initial weight and the second initial weight.

[0068] From the above, it can be concluded that this embodiment further refines the evaluation strategy and improves the accuracy and applicability of the evaluation by introducing considerations of the number of podcast types listened to by users and the degree of match between the target podcast type and the user's historical listening records. This embodiment, through a more refined evaluation strategy, can provide users with evaluation results that are more in line with their current status and needs. When a user is interested in a new type of podcast but lacks sufficient experience, the present disclosure can give a relatively objective evaluation based on the initial weight, helping users to better understand the content and quality of this type of podcast, thereby optimizing the user's listening experience.

[0069] In one embodiment of the present disclosure, the podcast content evaluation method further includes: determining a first initial weight and a second initial weight; Determine matches between users and podcasts; In response to the matching result being a match: Lowering the first initial weight according to the first matching step length to obtain a first weight; and increasing the second initial weight according to the first matching step length to obtain a second weight; In response to the matching result being no match: The first initial weight is increased according to the second matching step length to obtain a first weight; and the second initial weight is decreased according to the second matching step length to obtain a second weight.

[0070] In this embodiment, whether the user and the podcast match can be determined by comparing the information filled in by the user when registering the account with the type information of the podcast.

[0071] Take a history and culture podcast as an example. If the user is a history student or a history enthusiast, then the user may be more compatible with the podcast; if the user is more interested in technology, then the degree of matching is relatively low. The matching results between users and podcasts can also be determined by analyzing the user's historical listening records, subscription preferences and other information.

[0072] The present disclosure takes into account that when a user matches with a podcast, it means that the user is more in agreement with and interested in the theme, style, etc. of the podcast. At this time, the user's subjective evaluation can better reflect the actual value and attractiveness of the podcast. Therefore, the first initial weight is reduced according to the first matching step, and the second initial weight is increased at the same time, so that the evaluation result is more inclined to the user's subjective feelings, highlighting the importance of user evaluation in the comprehensive evaluation, and better reflecting the actual significance and value of the podcast to such matching users. On the contrary, when the user does not match with the podcast, the user may not be satisfied or interested in certain aspects of the podcast. At this time, the objective quality of the podcast itself becomes more important because it is an inherent attribute of the podcast and does not depend on the user's subjective preference. Therefore, the first initial weight is increased according to the second matching step, and the second initial weight is reduced, so that the podcast's own professional content, production level and other objective factors occupy a larger proportion in the evaluation, which can more objectively reflect the quality of the podcast itself, and is not too affected by the user's subjective mismatch factors.

[0073] The first matching step length and the second matching step length are both set based on experience and actual needs.

[0074] Specifically, assuming that for the above-mentioned history and culture podcast, the first matching step is set to 0.05, and the second matching step is 0.03. When it is determined that a certain history enthusiast user matches the podcast, the original first weight is 0.6 and the second weight is 0.4. After adjustment, the first weight becomes 0.6-0.05=0.55, and the second weight becomes 0.4+0.05=0.45. When it is determined that a certain technology enthusiast user does not match the podcast, the initial first weight of 0.6 and the second weight of 0.4 are adjusted, and the first weight becomes 0.6+0.03=0.63, and the second weight becomes 0.4-0.03=0.37.

[0075] From the above, it can be concluded that, in this embodiment, when the user and the podcast have a high degree of match, the first initial weight (i.e., the weight of the podcast's own attributes) is reduced and the second initial weight (i.e., the weight of the user's evaluation) is increased, so that the evaluation results are more focused on the user's subjective feelings. This adjustment can accurately capture the user's recognition and interest preferences for the podcast content, thereby providing more personalized evaluation feedback. For mismatched user and podcast combinations, the first initial weight is increased and the second initial weight is reduced, ensuring that the objective quality of the podcast content itself dominates the evaluation. This adjustment helps to avoid users making one-sided or unfair evaluations of the podcast content due to personal preferences or interests, and enhances the objectivity and fairness of the evaluation.

[0076] In one embodiment of the present disclosure, the podcast content is evaluated based on at least one podcast evaluation dimension to obtain a first evaluation result, including: Evaluate the podcast content based on at least one fixed evaluation dimension to obtain a first podcast evaluation result; Evaluate the podcast content based on at least one non-fixed evaluation dimension to obtain a second podcast evaluation result; The first podcast evaluation result, the second evaluation result and the podcast information are input into the first neural network to obtain the first evaluation result.

[0077] In this embodiment, fixed evaluation dimensions refer to dimensions that do not change with different podcast information, as shown in Table 1, where "audio quality" and "audio volume" are both fixed evaluation dimensions, that is, dimensions that change regardless of changes in information such as the type, duration, and update frequency of the podcast.

[0078] Non-fixed evaluation dimensions refer to dimensions that change with different podcast information, as shown in Table 1, where the podcast evaluation dimensions corresponding to the "technology podcast type" are "content authenticity" and "content professionalism", while the podcast evaluation dimension corresponding to the "entertainment podcast type" is "content fun". The above-mentioned "content authenticity", "content professionalism" and "content fun" will change with the change of podcast type information, that is, technology podcast type podcasts will not judge the fun of their content, and entertainment podcasts will not judge the authenticity and professionalism of their content.

[0079] The first neural network is trained on a data set consisting of a large amount of labeled podcast data (i.e., the accurate comprehensive evaluation results of each podcast are known), the first podcast evaluation results, and the second podcast evaluation results.

[0080] In this embodiment, the same first podcast evaluation result and the same second podcast evaluation result are input into the first neural network, and the obtained first evaluation result will also be different due to the different input podcast information. This is because different podcast information determines the non-fixed evaluation dimension of the podcast. For example, the non-fixed evaluation criteria corresponding to the science and education type podcast and the non-fixed evaluation criteria corresponding to the music type podcast are different. Therefore, even if the first podcast evaluation results and the second podcast evaluation results of the science and education type podcast and the music type podcast are consistent, their output first evaluation results may also be different.

[0081] For example, in this embodiment, the fixed evaluation dimensions include audio quality and audio volume, for the above-mentioned science and education type podcasts and music type podcasts, the audio quality and audio volume requirements of the music type podcasts are higher, so when judging the first evaluation results, the fixed evaluation dimensions should be considered as the focus, and the science and education type podcasts should pay more attention to the authenticity of their content (one of the non-fixed evaluation criteria).

[0082] For podcast information, it can be converted into a numerical vector form suitable for neural network input through one-hot encoding and other methods. For example, if there are three types of podcasts: technology, entertainment, and culture, the "technology" podcast can be encoded as [1,0,0], the "entertainment" as [0,1,0], and the "culture" as [0,0,1]. The first podcast evaluation result, the second podcast evaluation result, and the processed podcast information are combined in the set order to construct a complete input vector. For example, if the first podcast evaluation result is 0.8 (normalized), the second podcast evaluation result is 0.7, the podcast category is one-hot encoded as [1,0,0] (indicating technology), the duration is 0.6 (assumed) after normalization, etc., these values ​​are arranged in order to form an input vector as the input data of the neural network.

[0083] In this embodiment, when the user and the podcast have a high degree of match, the first initial weight (i.e., the weight of the podcast's own attributes) is reduced and the second initial weight (i.e., the weight of the user's evaluation) is increased, so that the evaluation results are more focused on the user's subjective feelings, and can accurately capture the user's recognition and interest preferences for the podcast content, thereby providing more personalized evaluation feedback. For mismatched user and podcast combinations, the first initial weight is increased and the second initial weight is reduced, ensuring that the objective quality of the podcast content itself dominates the evaluation. This adjustment helps to avoid users making one-sided or unfair evaluations of podcast content due to personal preferences or interests, enhances the objectivity and fairness of the evaluation, and improves the accuracy and reliability of the podcast evaluation method.

[0084] In one embodiment of the present disclosure, the podcast content is evaluated based on at least one user evaluation dimension to obtain a second evaluation result, including: Evaluate the podcast content based on the behaviors of multiple users during the listening period to obtain multiple first user evaluation results; Evaluating the podcast content based on the behaviors of multiple users during non-listening periods to obtain multiple second user evaluation results; Determining weights corresponding to the plurality of first user evaluation results and weights corresponding to the plurality of second user evaluation results based on the user information of the plurality of users, the first condition, and the second condition; A weighted calculation is performed based on the weights corresponding to the multiple first user evaluation results and the weights corresponding to the multiple second user evaluation results to obtain the second evaluation result.

[0085] In one embodiment of the present disclosure, a first user evaluation result when the user's behavior during the listening period meets the first condition is better than a first user evaluation result when the user's behavior during the listening period does not meet the first condition.

[0086] In one embodiment of the present disclosure, the first condition includes: Set up loop playback; and / or, the difference between the actual volume of the podcast and the volume of the user's environment is greater than a first volume threshold, and the user increases the volume; The second condition mentioned above includes: User downloads; and / or user sharing.

[0087] In this embodiment, the first volume threshold refers to the user's habit of turning up the actual volume of the podcast to a decibel value greater than the ambient volume. The actual volume of the podcast refers to the actual decibel number heard by the user. When the difference between it and the ambient volume of the user reaches a certain level (i.e., greater than the first volume threshold), and the user actively increases the volume, it means that the user wants to hear the podcast content more clearly, reflecting that the user pays more attention to the content and does not want to miss any details. Therefore, the evaluation results corresponding to this type of behavior should also occupy a relatively important position in the overall evaluation and be given a higher weight.

[0088] The first volume threshold is set according to the user's habits. For example, user E is accustomed to raising the decibel level of the actual podcast sound to 5 decibels louder than the surrounding environment volume, while user F is accustomed to raising the decibel level of the actual podcast sound to 10 decibels louder than the surrounding environment volume. In this case, the first volume threshold corresponding to user E may be 5 decibels, and the first volume threshold corresponding to user F may be 10 decibels.

[0089] Listening session refers to the time period when users are listening to podcasts. During this period, users' behavior can intuitively reflect their current feelings and reactions to the podcast content. For example, whether users paused, skipped part of the content, adjusted the volume, etc., all imply the user's interest in the content, whether it is easy to understand, etc. By collecting and analyzing these behaviors of many users during the listening period, the podcast content is quantitatively or qualitatively evaluated from different angles, thereby obtaining multiple first-user evaluation results, each of which represents a feedback on the podcast given by the corresponding user based on the listening behavior.

[0090] Non-listening time refers to other time periods when users are not actually listening to podcasts. The behavior at this time reflects the user's subsequent attitude towards the podcast content and their willingness to spread the content after listening. For example, whether the user actively downloads the podcast for subsequent listening, whether the podcast is shared with others, whether the podcast is commented on on the platform, etc., all of which indirectly reflect the user's recognition of the podcast content and whether they think it is valuable and worth recommending to others. Based on these behavioral analyses, the podcast content can be evaluated to obtain multiple second user evaluation results.

[0091] In this embodiment, the weights corresponding to the multiple first user evaluation results and the weights corresponding to the multiple second user evaluation results may be determined in the following manner: The first user evaluation result corresponding to the user information that meets the first condition and the second condition in the user information of multiple users is assigned a first user weight, and the corresponding second user evaluation result is assigned a second user weight; The first user evaluation result corresponding to the user information that meets the first condition and does not meet the second condition among the user information of multiple users is assigned a third user weight, and the corresponding second user evaluation result is assigned a fourth user weight; The first user evaluation result corresponding to the user information that does not meet the first condition but meets the second condition among the user information of multiple users is assigned a fifth user weight, and the corresponding second user evaluation result is assigned a sixth user weight; The first user evaluation result corresponding to the user information that does not meet the first condition and the second condition among the user information of multiple users is assigned to the seventh user weight, and the corresponding second user evaluation result is assigned to the eighth user weight.

[0092] In this embodiment, it is taken into account that different behaviors of users express their different evaluations and interests in the podcast content. For example, setting loop playback when listening to a podcast represents recognition of the podcast content. When not listening to the podcast, sharing it with others or listening to the same podcast multiple times is different. It should be noted that the above-mentioned listening to the same podcast multiple times and setting loop playback are different. Setting loop playback refers to the action performed while listening, while listening to the same podcast multiple times can be listening to it again after a month, which belongs to the non-listening period, that is, the process of multiple listening is the listening period, but the action of multiple listening is the non-listening period.

[0093] The first user weight, the second user weight, the third user weight, the fourth user weight, the fifth user weight, the sixth user weight, the seventh user weight and the eighth user weight are all set according to experiments.

[0094] From the above, it can be concluded that the present disclosure can cover all aspects of the podcast content more comprehensively by subdividing the evaluation dimensions into fixed evaluation dimensions and non-fixed evaluation dimensions. The introduction of non-fixed evaluation dimensions enables the evaluation method to flexibly adjust the evaluation criteria according to different podcast information, which not only adapts to the characteristics of diverse podcast content, but also improves the applicability and generalization ability of the evaluation method. Whether it is for science and technology, entertainment or other types of podcasts, this embodiment can give evaluation results that meet their characteristics and needs. This embodiment uses a first neural network to comprehensively process the first podcast evaluation results, the second podcast evaluation results and the podcast information, and this method realizes the intelligent evaluation. Neural networks can automatically learn and extract features and rules in data, so as to give more objective and accurate evaluation results, thereby improving the accuracy and reliability of the podcast evaluation method.

[0095] In one embodiment of the present disclosure, optimizing the podcast recommendation content based on the evaluation results includes: In response to the evaluation result satisfying the optimization criteria, optimizing the podcast recommendation content based on the evaluation result; In response to the evaluation result not satisfying the optimization criteria, no optimization is performed.

[0096] In this embodiment, the optimization criteria are a series of pre-set conditions or thresholds used to determine whether the evaluation results of the podcast have reached a level that requires optimization of its recommended content. For example, the optimization criteria may be that the evaluation result is less than 8 points. If the evaluation result obtained through the above steps and methods is less than 8 points, the podcast recommended content is optimized.

[0097] You can optimize content by adjusting recommendation priorities, accurately matching recommendation audiences, or increasing recommendation channels.

[0098] For example, when the evaluation result is 7 points, the optimization conditions are met. The priority of this type of podcast can be lowered, the recommendation of similar types of podcasts can be reduced, and the podcasts that are related to the user information and have higher evaluation results from other users with similar user information can be increased.

[0099] From the above, it can be concluded that by setting optimization criteria, the present disclosure can accurately identify which podcast recommendations need to be improved, so as to optimize them in a targeted manner. This helps to improve the quality of podcast content received by users, making it more in line with their interests and needs, thereby improving user satisfaction and loyalty. The optimization strategy avoids indiscriminately processing all podcast recommendations, but selectively optimizes based on whether the evaluation results meet the optimization criteria, significantly improving the efficiency of the recommendation system and reducing unnecessary calculations and resource consumption.

[0100] Corresponding to the podcast content evaluation method in the above embodiment, Figure 2This is a structural block diagram of a podcast content evaluation device provided by an embodiment of the present disclosure. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 2 The podcast content evaluation device 20 includes: a standard determination module 21, a podcast evaluation module 22 and a podcast optimization module 23.

[0101] Wherein, the standard determination module 21 is used to determine the target evaluation standard based on the podcast information and the user information, and the target evaluation standard includes multiple evaluation dimensions; A podcast evaluation module 22, used to evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria, and obtain evaluation results corresponding to the podcast content; The podcast optimization module 23 is used to optimize the podcast recommendation content based on the evaluation results.

[0102] In one embodiment of the present disclosure, the target evaluation criteria include: at least one podcast evaluation dimension and at least one user evaluation dimension; The podcast evaluation module 22 is specifically configured to evaluate the podcast content based on at least one podcast evaluation dimension to obtain a first evaluation result; Evaluate the podcast content based on at least one user evaluation dimension to obtain a second evaluation result; An evaluation result corresponding to the podcast content is determined based on the first evaluation result and the second evaluation result.

[0103] In one embodiment of the present disclosure, the podcast evaluation module 22 is specifically used to perform weighted calculation on the first evaluation result and the second evaluation result to obtain an evaluation result corresponding to the podcast content; wherein the weight corresponding to the first evaluation result is the first weight, and the weight corresponding to the second evaluation result is the second weight, and both the first weight and the second weight are determined based on user information.

[0104] In one embodiment of the present disclosure, the podcast content evaluation device 20 further includes: a first weight determination module; A first weight determination module, used to determine a first initial weight and a second initial weight; Adjusting the first initial weight based on the user's listening time and the user's listening frequency to obtain a first weight; The second initial weight is adjusted based on the user's listening time and the user's listening frequency to obtain a second weight.

[0105] In one embodiment of the present disclosure, the podcast content evaluation device 20 further includes: a second weight determination module; A second weight determination module, used to determine a first initial weight and a second initial weight; Determine matches between users and podcasts; In response to the matching result being a match: Lowering the first initial weight according to the first matching step length to obtain a first weight; and increasing the second initial weight according to the first matching step length to obtain a second weight; In response to the matching result being no match: The first initial weight is increased according to the second matching step length to obtain a first weight; and the second initial weight is decreased according to the second matching step length to obtain a second weight.

[0106] In one embodiment of the present disclosure, the podcast evaluation module 22 is further configured to evaluate the podcast content based on at least one fixed evaluation dimension to obtain a first podcast evaluation result; Evaluate the podcast content based on at least one non-fixed evaluation dimension to obtain a second podcast evaluation result; The first podcast evaluation result, the second evaluation result and the podcast information are input into the first neural network to obtain the first evaluation result.

[0107] In one embodiment of the present disclosure, the podcast optimization module 23 is specifically configured to optimize the podcast recommendation content based on the evaluation result in response to the evaluation result satisfying the optimization criteria; In response to the evaluation result not satisfying the optimization criteria, no optimization is performed.

[0108] See also Figure 3 , Figure 3 This is a schematic block diagram of a smart device provided by an embodiment of the present disclosure. Figure 3 The smart device 300 in the embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303 and one or more memories 304. The processors 301, input devices 302, output devices 303 and memories 304 communicate with each other via a communication bus 305. The memory 304 is used to store computer programs, which include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. The processor 301 is configured to call the program instructions to execute the functions of each module / unit in the above-mentioned device embodiments, such as Figure 2 The functions of modules 21 to 23 are shown.

[0109] It should be understood that in the embodiment of the present disclosure, the processor 301 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0110] The input device 302 may include a touch panel, a fingerprint collection sensor (for collecting the user's fingerprint information and fingerprint direction information), a microphone, etc., and the output device 303 may include a display (LCD, etc.), a speaker, etc.

[0111] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0112] In a specific implementation, the processor 301, input device 302, and output device 303 described in the embodiments of the present disclosure may execute the implementation methods described in the first and second embodiments of the podcast content evaluation method provided in the embodiments of the present disclosure, and may also execute the implementation methods of the smart device described in the embodiments of the present disclosure, which will not be repeated here.

[0113] In another embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by the processor, all or part of the processes in the above-mentioned embodiment method are implemented, and the computer program can also be completed by instructing the relevant hardware through the computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0114] The computer-readable storage medium may be an internal storage unit of the smart device of any of the aforementioned embodiments, such as a hard disk or memory of the smart device. The computer-readable storage medium may also be an external storage device of the smart device, such as a plug-in hard disk equipped on the smart device, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer-readable storage medium may also include both an internal storage unit of the smart device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the smart device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0115] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0116] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the smart devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0117] In the several embodiments provided in the present application, it should be understood that the disclosed intelligent devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces or units, or it can be an electrical, mechanical or other form of connection.

[0118] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present disclosure.

[0119] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0120] The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A podcast content evaluation method, characterized in that: include: Determine a target evaluation criterion based on the podcast information and the user information, wherein the target evaluation criterion includes multiple evaluation dimensions; Evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content; The podcast recommendation content is optimized based on the evaluation results.

2. The podcast content evaluation method according to claim 1, characterized in that: The target evaluation criteria include: at least one podcast evaluation dimension and at least one user evaluation dimension; The step of evaluating the podcast content based on the multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content includes: Evaluate the podcast content based on at least one podcast evaluation dimension to obtain a first evaluation result; Evaluate the podcast content based on at least one user evaluation dimension to obtain a second evaluation result; An evaluation result corresponding to the podcast content is determined based on the first evaluation result and the second evaluation result.

3. The podcast content evaluation method according to claim 2, characterized in that: Determining the evaluation result corresponding to the podcast content based on the first evaluation result and the second evaluation result includes: The first evaluation result and the second evaluation result are weightedly calculated to obtain an evaluation result corresponding to the podcast content; wherein the weight corresponding to the first evaluation result is a first weight, and the weight corresponding to the second evaluation result is a second weight, and both the first weight and the second weight are determined based on the user information.

4. The podcast content evaluation method according to claim 3, characterized in that: Also includes: Determining a first initial weight and a second initial weight; Adjusting the first initial weight based on the user's listening time and the user's listening frequency to obtain a first weight; The second initial weight is adjusted based on the user's listening time and the user's listening frequency to obtain a second weight.

5. The podcast content evaluation method according to claim 3, characterized in that: Also includes: Determining a first initial weight and a second initial weight; Determining a match between the user and the podcast; In response to the matching result being a match: reducing the first initial weight according to a first matching step length to obtain the first weight; and increasing the second initial weight according to the first matching step length to obtain the second weight; In response to the matching result being a mismatch: The first initial weight is increased according to a second matching step length to obtain the first weight; and the second initial weight is decreased according to the second matching step length to obtain the second weight.

6. The podcast content evaluation method according to claim 2, characterized in that: The step of evaluating the podcast content based on at least one podcast evaluation dimension to obtain a first evaluation result includes: Evaluate the podcast content based on at least one fixed evaluation dimension to obtain a first podcast evaluation result; Evaluate the podcast content based on at least one non-fixed evaluation dimension to obtain a second podcast evaluation result; The first podcast evaluation result, the second evaluation result and the podcast information are input into a first neural network to obtain the first evaluation result.

7. The podcast content evaluation method according to claim 1, characterized in that: The optimizing the podcast recommendation content based on the evaluation result includes: In response to the evaluation result satisfying the optimization criteria, optimizing the podcast recommendation content based on the evaluation result; In response to the evaluation result not satisfying the optimization criterion, no optimization is performed.

8. A podcast content evaluation device, characterized in that: include: A standard determination module, used to determine a target evaluation standard based on podcast information and user information, wherein the target evaluation standard includes multiple evaluation dimensions; A podcast evaluation module, used to evaluate the podcast content based on multiple evaluation dimensions in the target evaluation criteria to obtain evaluation results corresponding to the podcast content; The podcast optimization module is used to optimize the podcast recommendation content based on the evaluation results.

9. An intelligent device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images