An automatic evaluation method of network information based on artificial intelligence

Abstract

The invention belongs to the field of network information analysis, and specifically discloses an automatic network information evaluation method based on artificial intelligence. The invention realizes dynamic evaluation of network news quality by tracking the effects of horizontal and vertical dissemination after the target network news is released, and then dynamically assigning weights of multi-dimensional quality indicators based on the dissemination effect, so that the evaluation result is more in line with the actual dissemination situation, rather than relying on static and outdated data, and can achieve more comprehensive, accurate and timely network news quality evaluation, which is helpful to improve the overall quality management level of network news. At the same time, after the network news quality is evaluated based on the dissemination effect, a dissemination effect curve and a quality curve are constructed using the dissemination time period, and then the dissemination effect curve and the quality curve are correlated and compared, so that the relationship between the effect and quality of news in the dissemination process can be deeply understood, and the dynamic characteristics of news dissemination are revealed.

Description

Technical Field

[0001] The invention belongs to the field of network information analysis, and specifically discloses an automatic network information evaluation method based on artificial intelligence. Background Art

[0002] The emergence of the Internet has broken the time and space limitations of traditional media, allowing information to spread rapidly around the world, which has greatly promoted the development of online news. In order to ensure the accuracy and timeliness of information, it is necessary to judge the quality of online news.

[0003] Although the prior art with publication number CN115687628A proposes a news quality evaluation method, system, computer device and storage medium, by obtaining multiple texts to be evaluated from the news content to be evaluated; the multiple texts to be evaluated are vectorized and then input into the quality evaluation model to output multi-dimensional quality evaluation results, without using more computing resources, the correlation between multiple quality judgment attributes is considered to perform model training, thereby achieving high-precision and high-efficiency automatic evaluation of news quality.

[0004] Although the above scheme can use computer models to judge the quality of online news from multiple dimensions, and to a certain extent improve the efficiency and comprehensiveness of judging the quality of online news, its judging method is static, which ignores the dissemination process and effect of news. As we all know, there is a dissemination process after news is generated, and the dynamic changes of dissemination effect will appear in this process. Therefore, static quality judgment cannot track the quality changes of news in the dissemination process, and may underestimate the true value of news. For example, some news may require continuous attention to show its long-term social effect. Ignoring this may lead to distortion of the judgment results and affect the credibility of the judgment results. In this case, it is necessary to track and determine the dissemination effect of news, so as to The news quality is dynamically judged based on the broadcasting effect. As the prior art with announcement number CN104598477B proposes a method and system for determining the news broadcasting effect, a certain number of target websites are used as the statistical scope, and weighted values ​​are calculated for some target websites that have reprinted the source news among all the target websites, so as to determine the broadcasting effect value of the source news reprinted in all the target websites, thereby objectively and accurately judging the broadcasting effect of the source news. When tracking the news broadcasting effect, this scheme only performs broadcasting coverage analysis between websites, which belongs to horizontal tracking, and lacks vertical broadcasting link tracking within the website, resulting in a single broadcasting effect evaluation, resulting in an incomplete and incomplete understanding of the news broadcasting effect, and easily underestimating the actual news broadcasting effect. Summary of the invention

[0005] In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide an automatic network information evaluation method based on artificial intelligence to solve the problems existing in the prior art.

[0006] The purpose of the present invention can be achieved through the following technical solutions: an artificial intelligence-based automatic network information evaluation method, comprising the following steps: S1, obtaining the initial publishing platform and initial publishing time of the target network news, and determining the quality evaluation indicators and quality evaluation model of the target network news, wherein the quality evaluation indicators include content quality indicators and non-content quality indicators.

[0007] S2. Conducting the dissemination monitoring of the target network news at set time intervals starting from the initial release time, specifically including horizontal dissemination monitoring and vertical dissemination monitoring, and obtaining horizontal dissemination status data and vertical dissemination status data at different dissemination time periods.

[0008] S3. Evaluate the propagation effect value of each propagation period based on the horizontal propagation status data and the vertical propagation status data of each propagation period.

[0009] S4. Dynamically assign weights to quality evaluation indicators based on the communication effect value at each communication stage.

[0010] S5. Collect quality evaluation indicators in each dissemination period and import them into the quality evaluation model in combination with the assigned quality evaluation indicator weights to obtain the quality value of the target network news in each dissemination period.

[0011] S6. Construct a communication effect curve and a quality curve based on the communication effect value and quality value of the target network news in each communication period, and then correlate and compare the communication effect curve and the quality curve to obtain the synchronization degree and the advantageous communication period.

[0012] In combination with all the above-mentioned technical solutions, the positive effects of the present invention are as follows: 1. The present invention embodies an objective and comprehensive analysis of the news dissemination effect by tracking the effects of horizontal and vertical dissemination after the target online news is released, and then dynamically assigns weights of multi-dimensional quality indicators based on the dissemination effect, thereby realizing dynamic evaluation of the quality of online news, making the evaluation results more in line with the actual dissemination situation, rather than relying on static and outdated data, and being able to achieve a more comprehensive, accurate and timely evaluation of the quality of online news, which helps to improve the overall quality management level of online news.

[0013] 2. After judging the quality of online news based on the dissemination effect, the present invention uses the dissemination period to construct a dissemination effect curve and a quality curve, and then correlates and compares the dissemination effect curve with the quality curve, which can deeply understand the relationship between the effect and quality of news in the dissemination process, reveal the dynamic characteristics of news dissemination, and at the same time determine the synchronization and advantageous dissemination period from the correlation comparison, providing strong support for the optimization of news content, the adjustment of dissemination strategy and the overall quality management. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0015] Figure 1 The present invention is a flowchart of the steps for implementing the method.

[0016] Figure 2 It is a propagation path diagram in the longitudinal propagation monitoring of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] See also Figure 1 As shown, the present invention proposes an automatic evaluation method for network information based on artificial intelligence, comprising the following steps: S1, obtaining the initial publishing platform and initial publishing time of the target network news, and determining the quality evaluation index and quality evaluation model of the target network news, wherein the quality evaluation index includes content quality index and non-content quality index, and the quality evaluation model is a weighted scoring model.

[0019] In the example implementation of the above scheme, the content quality indicators of online news are mainly aimed at news content, including but not limited to accuracy, originality, objectivity, completeness, reliability, etc., among which accuracy refers to whether the news content has been verified and whether it is accurate, originality refers to whether the news content is original and whether it avoids plagiarism and repeated reports, objectivity refers to whether the news content remains neutral and avoids personal bias and subjective comments, completeness refers to whether the news content is complete, and reliability refers to whether the news source is credible and whether the information comes from authoritative channels. Non-content quality indicators include but are not limited to page loading speed, content update timeliness, visit volume, audience coverage, etc., among which page loading speed refers to whether the loading time of the news page is fast enough, content update timeliness refers to the update timeliness of the subsequent development of the news, visit volume refers to the frequency of the news page being visited and interacted, and audience coverage refers to the number of audiences reached by the news content.

[0020] It is important to understand that the quality evaluation indicators of online news are divided into content quality indicators and non-content quality indicators because content quality indicators focus on the authenticity, comprehensiveness, and depth of the news, which directly affect the core value of the news and the audience's trust. Non-content quality indicators focus on the presentation of news, user experience, and other aspects, which affect the user experience and access efficiency. Therefore, by finely dividing quality indicators, on the one hand, the quality evaluation is more systematic and ensures the comprehensiveness of news quality evaluation. On the other hand, it can provide a clear and definite assignment subject for the dynamic weight assignment of content quality indicators and non-content quality indicators based on the communication effects tracked during the communication process, ensuring the dynamic evaluation of news quality.

[0021] S2. Conducting the dissemination monitoring of the target network news at set time intervals starting from the initial release time, specifically including horizontal dissemination monitoring and vertical dissemination monitoring, and obtaining horizontal dissemination status data and vertical dissemination status data at different dissemination time periods.

[0022] In one achievable method, the process of determining the time interval is as follows: historical release records are retrieved from the initial release platform of the target network news, and the news type and news subject are extracted from the target network news, and then historical release news with the same news type and news subject as the target network news are screened out from the historical release records as reference release news, and at the same time, the initial release time corresponding to the reference release news is extracted from the historical release records.

[0023] It should be pointed out that news type refers to the form or category of news, that is, the presentation method or content style of news reports. News type can determine the narrative angle, format and reporting method of news. For example, the news type can be breaking news: reporting important recent events, such as natural disasters and emergencies. Commentary article: providing personal insights and comments on an event or topic. Feature report: detailed introduction to the story of an event or person. News flash: reporting the latest events in a concise and clear manner. News theme refers to the core content or topic of a news report, that is, the main topic involved in the news. For example, the news theme can be "climate change" or "technological innovation."

[0024] It should be explained that when predicting the life cycle of current online news, historical news with the same news type and news theme are selected from the same publishing platform as a reference based on the news type and news theme of the current online news. This is because the life cycle of news is often affected by the news type and theme. News of the same type and theme have high similarities in terms of audience interests. Therefore, news with similar historical performance can provide more relevant prediction data. In addition, historical data provides empirical evidence. By analyzing the life cycle of similar news, the randomness and uncertainty in the prediction of current news can be reduced.

[0025] The access log of the reference published news is retrieved from the initial publishing platform, and the timestamp of each access log is extracted, and then the access log is grouped by time window. Exemplarily, the time window can be one hour or one day, and the access volume of each time window is obtained.

[0026] An access change curve is constructed based on the visits to the referenced news releases in each time window. Specifically, the curve can be constructed in a coordinate system with the time window as the horizontal axis and the visits as the vertical axis, and an access decay threshold is set. Then, the time window corresponding to the access decay threshold is intercepted from the constructed access change curve to extract the access decay time.

[0027] The access decay time of the reference release news combined with the initial release time constitutes the life cycle of the reference release news.

[0028] The number of reference news releases screened is counted. If there is only one screened number, the time interval is set according to the life cycle of the reference news release. If there is more than one screened number, the life cycle lengths of the reference news releases are averaged, where the life cycle length is the length between the access decay time and the initial release time, and the time interval is set according to the average life cycle length.

[0029] After constructing an access change curve based on the access log of the reference published news, the present invention obtains the access decay time according to the time window corresponding to the access decay threshold in the access change curve, and obtains the life cycle of the reference published news accordingly. This operation is reasonable and effective, mainly because first, the access change curve is generated based on the actual access log, reflecting the user's real access behavior, and the life cycle prediction analysis based on the access change curve is established in a real and reliable environment; second, the access change curve is a dynamic trend data, and the life cycle prediction based on the dynamic trend is more in line with the actual situation; third, the access decay threshold provides an objective standard to define when the access volume enters the decay stage. This quantitative evaluation method avoids subjective judgment and improves the scientific rationality of life cycle prediction.

[0030] Furthermore, after obtaining the life cycle duration or average life cycle of the referenced news release, the time interval for dividing the dissemination time periods is determined. The specific time interval may be based on the life cycle duration. If the life cycle duration is longer, the time interval may be set longer, which can reduce frequent analysis and reduce operating costs. If the life cycle duration is shorter, the time interval may be set shorter, which can improve the timeliness of dissemination data capture and provide more detailed dissemination data, thereby obtaining a reasonable number of dissemination time periods, which can reduce the noise caused by too frequent or too sparse dissemination data collection, thereby improving the accuracy and effectiveness of data analysis.

[0031] For example, assuming that the predicted life cycle duration is 10 days, the propagation period can be divided into daily time intervals. If the predicted life cycle duration is 3 days, the propagation period can be set to one time interval per hour.

[0032] What needs to be further emphasized is that when dividing the communication time periods according to the length of the life cycle, it is necessary to ensure that the number of divisions is not too small, so as to avoid the situation where too few divisions lead to the subsequent communication effect curve and quality curve being constructed too roughly, affecting the quality of subsequent correlation analysis.

[0033] As a preferred implementation of the above scheme, horizontal communication monitoring refers to the following process: monitoring the network platforms where the target network news has been disseminated during each dissemination period, recorded as diffusion network platforms, and at the same time monitoring the interaction volume of the target network news on the diffusion network platforms and the user accounts participating in the interaction.

[0034] Applied to the above implementation plan, the interaction volume refers to the total amount of interactive behaviors such as likes, favorites, comments, shares and forwarding.

[0035] The horizontal communication status data obtained by adopting the above preferred implementation includes the number of diffusion network platforms, the total interaction volume and the distribution ratio of the audience population, wherein the total interaction volume is obtained by accumulating the interaction volume of the target network news on each diffusion network platform.

[0036] The process of obtaining the audience population distribution ratio is as follows: based on the user accounts that participate in the interaction of the target online news on the diffusion network platform, the user ages of the corresponding accounts are retrieved, and the population category to which each user account belongs is obtained based on the preset population category age group classification standard.

[0037] In the implementation example of the above scheme, the age group classification standards of population categories can refer to the existing commonly used age group classification standards.

[0038] The population categories to which each user account belongs are classified into the same population categories, the number of user accounts corresponding to each population category is summarized, and the proportion is calculated to obtain the audience population distribution ratio, specifically the distribution ratio of young people, middle-aged people and the elderly.

[0039] As a further preferred implementation of the above scheme, vertical communication monitoring refers to the following process: adding tracking parameters to the page link of the target online news before it is released, thereby tracking the click sources of the target online news from the users who participate in the interaction in the diffusion network platform, and screening out the user accounts whose click sources are shared as the receiving accounts, and at the same time, the user accounts of the sharing sources are used as the sharing accounts.

[0040] It should be pointed out that click sources include direct sources and sharing sources. Sharing sources refer to users being exposed to news through links shared and forwarded by third-party users.

[0041] The sharing account and the receiving account are each regarded as a node to form a propagation node chain, and all the propagation node chains formed are then relayed to form a propagation link, and each propagation link has a starting node and an ending node.

[0042] In the example of the above scheme, it is assumed that the propagation node chains formed are A→B, A→C, A→D, B→E, B→H, E→F, and E→G. At this time, the propagation links formed are A→B→E→F, A→B→E→G, A→B→H, A→C, and A→D.

[0043] In each dissemination period, the dissemination link chain formed by the target network news in the diffusion network platform is summarized, and the click timestamps of the start node and the end node in the dissemination chain corresponding to the click of the target network news are recorded to calculate the dissemination duration of the dissemination chain.

[0044] Compare the nodes in each propagation link, and then combine and merge the propagation links corresponding to the same nodes to form a propagation path graph, see Figure 2 shown.

[0045] The longitudinal propagation status data obtained by adopting the above-mentioned preferred implementation includes the number of propagation path graphs and the average propagation influence, wherein the average propagation influence is obtained as follows: counting the number of propagation links in the propagation path graph, and counting the number of nodes in each propagation link as the length of each propagation link.

[0046] Compare the nodes existing in the propagation path graph and count the number of propagation links involved in the same node.

[0047] The length of each propagation link is combined with the number of propagation links involved in each node and the propagation time of each propagation link into the formula , and obtain the corresponding communication influence of the communication path diagram , where Indicates the propagation path diagram. The length of the propagation link, Indicates the transmission link number, , Indicates the propagation path diagram. The number of propagation links involved in each node, Indicates the node number, , Indicates the propagation path diagram. The propagation time of the propagation link. From this formula, we can see that the longer the propagation link is, the more propagation links the node involves, the shorter the propagation time of the propagation link is, and the greater the propagation influence is.

[0048] The average propagation influence is obtained by calculating the mean of the propagation influence corresponding to each propagation path diagram.

[0049] It is important to know that when analyzing communication influence, the longest length of the communication link, the largest number of nodes involved in the communication link, and the shortest duration of the communication link are used as analysis parameters. This is because the longest length represents the deepest level of information dissemination, which means that the information is likely to affect a wider audience, thereby evaluating the potential influence of the information. The number of nodes involved means a stronger network effect, which helps to evaluate the social influence of information dissemination. The shortest propagation duration reflects the speed of information dissemination and the audience's ability to respond quickly to information. By using these parameters for communication influence analysis, a more comprehensive and accurate evaluation of communication influence can be made. These parameters can provide an in-depth understanding of the structure and dynamics of the communication network, while relying on average parameters may not accurately reflect the important characteristics and actual effects of the communication network.

[0050] S3. Evaluate the communication effect value of each communication period based on the horizontal communication status data and vertical communication status data of each communication period. The specific evaluation is as follows: compare the horizontal communication status data of each communication period with the previous communication period to obtain the increase in the number of diffusion network platforms, the increase in the total interaction volume and the increase in the proportion of audience population distribution.

[0051] The specific increment of the audience population distribution ratio in the above is obtained as follows: Assuming that the audience population distribution ratio in the previous communication period is , the audience distribution ratio in the current communication period is , the audience population distribution ratio increment at this time is , , , which is specifically reflected in the following communication effect values .

[0052] Compare the longitudinal propagation status data of each propagation period with that of the previous propagation period to obtain the increment in the number of propagation path graphs and the increment in the average propagation influence.

[0053] Using evaluative , get the propagation effect value of each propagation period , Indicates the transmission period number, , where , , Respectively represent The increase in the number of diffusion network platforms, the increase in the total interaction volume, and the increase in the proportion of audience population distribution during the dissemination period compared with the previous dissemination period. , , Respectively represent The number of diffusion network platforms, total interaction volume, and audience distribution ratio during the communication period. , Respectively represent The increase in the number of propagation path diagrams and the increase in the average propagation influence between the propagation period and the previous propagation period, , Respectively represent The number of propagation path diagrams and average propagation influence during the propagation period, Indicates that, represents a positive integer, represents a set constant, and , illustratively, .

[0054] S4. The dynamic assignment of weights of quality judgment indicators based on the communication effect values ​​of each communication stage is implemented as follows: the communication effect value of each communication period is compared with the set effective threshold. For example, the effective threshold is 0.6. If the communication effect value of a certain communication period does not reach the effective threshold, it means that the target network news has not been effectively spread during the communication period. At this time, the content quality indicators and non-content quality indicators are assigned with equal weights. This can comprehensively evaluate the communication effect without emphasizing one aspect, ensuring that the comprehensive evaluation is more objective.

[0055] If the communication effect value of a certain communication period reaches the effective threshold, the difference between the communication effect value of the communication period and the effective threshold is divided by the effective threshold to obtain the communication effect super-effectiveness ratio, where the communication effect super-effectiveness ratio can quantify the degree to which the communication effect exceeds expectations, and then the equal weight is combined with the communication effect super-effectiveness ratio to perform weight appreciation calculation, and the calculation result is used as the weight value of the non-content quality index of the communication period, and the equal weight is combined with the communication effect super-effectiveness ratio to perform weight depreciation calculation, and the calculation result is used as the weight value of the content quality index of the communication period. In this way, when the communication effect exceeds expectations, it means that the overall effect of the communication period is good. At this time, the equal weight is combined with the super-effectiveness ratio to perform weight appreciation, and the weight value of the non-content quality index is increased to better reflect its contribution to the communication effect, while reducing the weight of the content quality index. This is because the communication effect is already very high, and further improving the content quality may contribute less to the improvement of the effect. The weight depreciation can avoid over-emphasizing the role of content quality in the communication effect that has reached the standard.

[0056] The present invention combines equal weight and super-efficiency ratio by assigning weights to quality indicators in judging the quality of online news, so that both content quality and non-content quality indicators can be appropriately adjusted according to actual results. This approach helps to dynamically balance the influence of content and dissemination process, making the quality evaluation results closer to the actual situation.

[0057] S5. Collect quality evaluation indicators in each dissemination period and import them into the quality evaluation model in combination with the assigned quality evaluation indicator weights to obtain the quality value of the target network news in each dissemination period.

[0058] It should be noted that the collection of quality evaluation indicators and the specific calculation of quality values ​​in each transmission period are not the focus of the implementation of the present invention and will not be elaborated in detail here.

[0059] S6. Construct a communication effect curve and a quality curve based on the communication effect value and quality value of the target network news in each communication period. Specifically, the coordinate system is established with the communication period as the horizontal axis and the communication effect value and quality value as the vertical axis. Then, the communication effect curve and the quality curve are correlated and compared to obtain the synchronization degree and the advantageous communication period. The synchronization degree is obtained as follows: mark the extreme value points on the communication effect curve and the quality curve respectively, where the extreme value refers to the local maximum or minimum value of the curve value at a certain point relative to the surrounding points, and the marked extreme value points are numbered in order from front to back on the curve.

[0060] Based on the same numbering principle, the extreme points marked on the propagation effect curve and the quality curve are formed into associated groups.

[0061] The propagation time periods of each associated group on the propagation effect curve and the quality curve are obtained respectively, and the propagation time periods are subtracted to obtain the time differences corresponding to each associated group.

[0062] The synchronization degree is calculated based on the time difference statistics corresponding to each association group: , where Indicates The time difference corresponding to the associated group, Indicates the association group number, , Indicates the number of associated groups. It indicates the life cycle length obtained by publishing news based on reference. The smaller the time difference corresponding to each associated group, the greater the synchronization. Represents a natural constant.

[0063] The reason for using the extreme points on the curve as analysis parameters when analyzing the synchronization between the propagation effect curve and the quality curve is that the extreme points can effectively identify important change points of the curve. These points often mark the turning points of the curve. Using these points to analyze the synchronization can make the analysis more accurate and efficient.

[0064] It is important to understand that synchronization can reflect the relationship and consistency between communication effect and quality. Specifically, it shows whether the quality of communication matches the change in quality during different communication periods. If the curves are highly synchronized, it means that the impact of quality on communication effect is relatively stable and significant; if they are not synchronized, it may indicate that the communication effect is greatly affected by other factors. The degree of synchronization helps to evaluate the actual contribution of quality to communication effect.

[0065] Furthermore, the advantageous propagation period is obtained as follows: based on the extreme value points marked in the propagation effect curve and the quality curve, the propagation period corresponding to the maximum value point is selected as the effect advantage period and the quality advantage period.

[0066] Compare the effect advantage period with the quality advantage period. If they are consistent, the corresponding communication period will be regarded as the advantageous communication period. This consistency indicates that the communication effect and quality have reached a high level during this period, with significant communication advantages. If they are inconsistent, the communication effect value and quality value of each communication period will be superimposed to obtain the comprehensive advantage value of each communication period, and then compared. The communication period corresponding to the maximum comprehensive advantage value will be selected as the advantageous communication period. This method of selecting the best period by calculating the comprehensive advantage value can more comprehensively identify the optimal communication period and avoid the limitations of a single indicator.

[0067] It is important to understand that identifying advantageous communication periods can help optimize resource allocation and invest more resources in the best performing periods to achieve higher communication effects and benefits.

[0068] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A method for automatic evaluation of network information based on artificial intelligence, characterized in that: The following steps are involved: S1. Obtain the initial release platform and initial release time of the target online news, and determine the quality evaluation index and quality evaluation model of the target online news, where the quality evaluation index includes content quality index and non-content quality index; S2. Conducting the dissemination monitoring of the target network news at set time intervals starting from the initial release time, specifically including horizontal dissemination monitoring and vertical dissemination monitoring, and obtaining horizontal dissemination status data and vertical dissemination status data at different dissemination time periods; S3, evaluating the propagation effect value of each propagation period based on the horizontal propagation status data and the vertical propagation status data of each propagation period; S4. Dynamically assign weights of quality evaluation indicators based on the communication effect values ​​at each communication stage; S5, collecting quality evaluation indicators in each dissemination period and importing them into the quality evaluation model in combination with the assigned quality evaluation indicator weights to obtain the quality value of the target network news in each dissemination period; The lateral transmission monitoring is described in the following process: Monitor the network platforms where the target network news has been spread during each dissemination period, record them as diffusion network platforms, and monitor the amount of interaction of the target network news on the diffusion network platforms and the user accounts participating in the interaction; The vertical transmission monitoring is described in the following process: Before the target online news is published, tracking parameters are added to its page link to track the click sources of the target online news in the diffusion network platform by participating users, and the user accounts whose click sources are shared are selected as the receiving accounts, and the user accounts of the sharing sources are used as the sharing accounts; The sharing account and the receiving account are each regarded as a node to form a transmission node chain, and all the formed transmission node chains are relayed to form a transmission link, and each transmission link has a starting node and an ending node; In each dissemination period, the dissemination link chain formed by the target network news in the diffusion network platform is summarized, and the click timestamps of the start node and the end node in the dissemination link corresponding to the click of the target network news are recorded, and the dissemination duration of the dissemination link is calculated; Compare the nodes existing in each propagation link, and then combine and merge the propagation links corresponding to the same nodes to form a propagation path graph; The evaluation of the communication effect value of each communication period is as follows: Compare the horizontal communication status data of each communication period with the previous communication period to obtain the increment of the number of diffusion network platforms, the increment of the total interaction volume and the increment of the distribution ratio of the audience population; Compare the longitudinal propagation status data of each propagation period with that of the previous propagation period to obtain the increment of the propagation path diagram and the increment of the average propagation influence; Using evaluative , get the propagation effect value of each propagation period , Indicates the transmission period number, , where , , Respectively represent The increase in the number of diffusion network platforms, the increase in the total interaction volume, and the increase in the proportion of audience population distribution during the dissemination period compared with the previous dissemination period. , , Respectively represent The number of diffusion network platforms, total interaction volume, and audience distribution ratio during the communication period. , Respectively represent The increase in the number of propagation path diagrams and the increase in the average propagation influence between the propagation period and the previous propagation period, , Respectively represent The number of propagation path diagrams and average propagation influence during the propagation period, Indicates that, represents a positive integer, represents a set constant, and ; The dynamic assignment of quality evaluation index weights according to the communication effect values ​​of each communication stage is implemented as follows: The communication effect value of each communication period is compared with the set effective threshold. If the communication effect value of a certain communication period does not reach the effective threshold, the equal weights of the content quality index and the non-content quality index are assigned. If the communication effect value of a certain communication period reaches the effective threshold, the comparison difference between the communication effect value of the communication period and the effective threshold is divided by the effective threshold to obtain the communication effect excess ratio. Then, the equal weight is combined with the communication effect excess ratio to perform weight appreciation calculation, and the calculation result is used as the weight value of the non-content quality index of the communication period. At the same time, the equal weight is combined with the communication effect excess ratio to perform weight depreciation calculation, and the calculation result is used as the weight value of the content quality index of the communication period.

2. The method for automatic network information evaluation based on artificial intelligence as claimed in claim 1, characterized in that: The process of determining the set time interval is as follows: Retrieving historical release records from the initial release platform of the target online news, extracting the news type and news subject from the target online news, and then selecting historical release news with the same news type and news subject as the target online news from the historical release records as reference release news, and at the same time extracting the initial release time corresponding to the reference release news from the historical release records; Retrieve the access log of the reference news release from the initial release platform, extract the timestamp of each access log, and then group the access logs by time window to obtain the number of visits in each time window; An access change curve is constructed based on the access volume of reference published news in each time window, and an access decay threshold is set. Then, a time window corresponding to the access decay threshold is intercepted from the constructed access change curve to extract the access decay time; The visit decay time of the reference release news combined with the initial release time constitutes the life cycle of the reference release news; The number of reference news releases screened is counted. If there is only one screened number, the time interval is set according to the life cycle of the reference news release. If there is more than one screened number, the life cycle lengths of the reference news releases are averaged and the time interval is set according to the average life cycle length.

3. The method for automatic network information evaluation based on artificial intelligence as claimed in claim 1, characterized in that: The horizontal communication status data includes the number of diffusion network platforms, the total interaction volume and the audience population distribution ratio. The total interaction volume is obtained by accumulating the interaction volume of the target network news on each diffusion network platform. The process of obtaining the audience population distribution ratio is as follows: Based on the user accounts that participate in the interaction of the target online news on the diffusion network platform, the user ages of the corresponding accounts are retrieved, and the population category to which each user account belongs is obtained according to the preset population category age group classification standard; The population categories to which each user account belongs are classified into the same population category, the number of user accounts corresponding to each population category is summarized, and the proportion is calculated to obtain the audience population distribution ratio.

4. The method for automatic network information evaluation based on artificial intelligence as claimed in claim 1, characterized in that: The vertical propagation status data includes the number of propagation path graphs and the average propagation influence, wherein the average propagation influence is obtained as follows: Count the number of propagation links in the propagation path graph, and count the number of nodes in each propagation link as the length of each propagation link; Compare the nodes existing in the propagation path graph and count the number of propagation links involved in the same node; The length of each propagation link is combined with the number of propagation links involved in each node and the propagation time of each propagation link into the formula , and obtain the corresponding communication influence of the communication path diagram , where Indicates the propagation path diagram. The length of the propagation link, Indicates the transmission link number, , Indicates the propagation path diagram. The number of propagation links involved in each node, Indicates the node number, , Indicates the propagation path diagram. The propagation time of the propagation link; The average propagation influence is obtained by calculating the mean of the propagation influence corresponding to each propagation path diagram.

5. The method for automatic network information evaluation based on artificial intelligence according to claim 1, characterized in that: The invention also includes S6, constructing a communication effect curve and a quality curve according to the communication effect value and quality value of the target network news in each communication period, and then correlating and comparing the communication effect curve and the quality curve to obtain the synchronization degree and the advantageous communication period, wherein the synchronization degree is obtained as follows: Mark the extreme points on the propagation effect curve and the quality curve respectively, and number the marked extreme points in the order from the front to the back on the curve; Based on the same numbering principle, the extreme value points marked on the propagation effect curve and the quality curve are formed into an associated group; Obtain the propagation time periods of each associated group on the propagation effect curve and the quality curve respectively, and subtract the propagation time periods to obtain the time differences corresponding to each associated group; The synchronization degree is calculated based on the time difference corresponding to each associated group.

6. The method for automatic network information evaluation based on artificial intelligence as claimed in claim 5, characterized in that: The advantageous propagation period is obtained as follows: Based on the extreme value points marked in the propagation effect curve and the quality curve, the propagation period corresponding to the maximum value point is selected as the effect advantage period and the quality advantage period; Compare the effect advantage period and the quality advantage period. If they are consistent, the corresponding communication period will be taken as the advantageous communication period. If they are inconsistent, superimpose the communication effect value and quality value of each communication period to obtain the comprehensive advantage value of each communication period, and compare them. Select the communication period corresponding to the maximum comprehensive advantage value as the advantageous communication period.

Images