Information display frequency control method based on user behavior characteristics

Abstract

The application discloses a kind of information display frequency control method based on user behavior characteristics, it is related to information display control technical field, for solving the problem that display content appears excessive push, by initial user setting group division mechanism to obtain similar potential group, according to each group information display time length selects target information and extracts word frequency screening representative vocabulary, according to representative vocabulary screening target group, its information is displayed in user interface and sets display frequency, real-time monitoring user browsing time length and browsing volume and statistics, comprehensive result excludes target group, according to the number of screening generates update ratio and enters frequency adjustment mechanism, uses update ratio to adjust display frequency to obtain accurate frequency, real-time monitoring screening result judges whether to exit frequency adjustment mechanism, when exiting, collect user click search and active refresh times to determine whether to exit group division mechanism, avoid information redundancy to cause user fatigue, improve information utilization efficiency and resource allocation rationality.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information display control, more particularly, the present application relates to a method for controlling information display frequency based on user behavior characteristics. BACKGROUND

[0002] With the rapid development of Internet information services, there are more and more ways for users to obtain information through terminal devices, and information pushing and display have gradually become the key means for platforms to improve user stickiness and information dissemination efficiency. Most of the existing information display methods are based on user registration information, historical browsing records or platform recommendation algorithms.

[0003] The prior art has the following disadvantages:

[0004] At present, the prior art generally adopts fixed display frequency or rough frequency adjustment strategy, which cannot be dynamically adjusted in combination with real-time behavior characteristics of users, lacks depth analysis of potential group behavior characteristics, and thus leads to over-pushing of display content, insufficient information relevance and decline of user experience. Therefore, a method for controlling information display frequency based on user behavior characteristics is proposed.

[0005] The above information disclosed in the background section is only intended to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a method for controlling information display frequency based on user behavior characteristics, which uses a dynamic control strategy combining group division mechanism, word frequency analysis and similarity screening, activity calculation based on browsing behavior and frequency adjustment mechanism to solve the problems raised in the above background.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical scheme, a method for controlling information display frequency based on user behavior characteristics, comprising the following steps:

[0008] Step S1: setting a group division mechanism for an initial user to obtain a plurality of same-potential groups of the initial user, and selecting target information of the same-potential groups according to information display time lengths of the same-potential groups;

[0009] Step S2: extracting word frequencies of the target information of the same-potential groups and screening to obtain representative words of the same-potential groups, screening the same-potential groups according to the representative words to obtain a target group, and displaying information of the target group on an initial user interface;

[0010] Step S3: Set the display frequency, monitor and count the initial user's browsing time and browsing volume of each target group information in real time, filter the target group according to the browsing time and browsing volume, and generate an update rate according to the number of filtered target groups;

[0011] Step S4: After generating the update rate, enter the frequency adjustment mechanism, adjust the display frequency using the update rate and obtain the accurate display frequency, and monitor the filtering result of the target group in real time to determine whether to exit the frequency adjustment mechanism;

[0012] Step S5: When exiting the frequency adjustment mechanism, collect the number of user clicks and the number of active refreshes to determine whether to exit the group division mechanism.

[0013] In a preferred embodiment, in step S1, the initial user is substituted into the set group division mechanism, and the initial information filled in by the initial user during registration is collected;

[0014] The initial information is substituted into the preset group division mechanism to divide the initial user into groups, the initial user is divided into multiple preset groups, the preset group into which the initial user is divided is marked, and multiple types of potential groups similar to the initial user are obtained;

[0015] The information display time of each potential group similar to the initial user is counted by the built-in clock, and the information with the longest display time in each potential group similar to the initial user is selected as the target information of the potential group similar to the initial user.

[0016] In a preferred embodiment, in step S2, the frequency of each word in the target information of the potential group similar to the initial user is calculated by counting the number of occurrences of each word in the target information of the potential group similar to the initial user, and the frequency of each word in the target information of the potential group similar to the initial user is calculated by comparing the number of occurrences of each word with the total number of occurrences of each word.

[0017] The word frequencies of each word in the target information of the potential group similar to the initial user are sorted in descending order according to the numerical value, and the word with the largest word frequency is selected as the representative word of the potential group similar to the initial user.

[0018] The word frequency of the representative word of the potential group similar to the initial user is called, and the representative words are sorted in descending order according to the numerical value of the word frequency, and the preset number of representative words are selected as comparison representative words according to the arrangement order, and the remaining representative words are selected as candidate representative words.

[0019] The potential group similar to the initial user corresponding to the comparison representative word is marked as the target group by default.

[0020] In a preferred embodiment, in step S2, each candidate representative vocabulary is vectorized with each comparison representative vocabulary, and the similarity of each candidate representative vocabulary with each comparison representative vocabulary is calculated by substituting the cosine similarity calculation formula for all candidate representative vocabularies;

[0021] The similarity of each candidate representative vocabulary with each comparison representative vocabulary is grouped according to the number of candidate representative vocabularies to obtain each candidate representative vocabulary similarity group;

[0022] The target value of each candidate representative vocabulary is obtained by averaging each candidate representative vocabulary similarity group;

[0023] The target value of each candidate representative vocabulary is compared with a preset similarity threshold;

[0024] If the target value of the candidate representative vocabulary exceeds the similarity threshold, the same type of potential group corresponding to the current candidate representative vocabulary is excluded;

[0025] If the target value of the candidate representative vocabulary is lower than the similarity threshold, the same type of potential group corresponding to the current candidate representative vocabulary is marked as a target group;

[0026] The information corresponding to the target group obtained by screening is used for interface display of the initial user.

[0027] In a preferred embodiment, in step S3, the browsing time and the number of views of each target group information by the initial user are monitored in real time within a set display frequency;

[0028] During the browsing process of the initial user within the display frequency, the single stay time of each information belonging to the current target group is recorded, and the stay time of all information under the current target group within the display frequency is accumulated to obtain the browsing time of the target group information;

[0029] Each time the initial user clicks into any information of the current target group within the display frequency is counted as one view and is counted to obtain the number of views of the target group information.

[0030] In a preferred embodiment, in step S3, the browsing time and the number of views of each target group information after standardization are substituted into the geometric mean method to obtain the activity coefficient of each target group;

[0031] The activity coefficient of each target group is compared with an activity threshold;

[0032] If the activity coefficient of the target group exceeds the activity threshold, the current target group is retained;

[0033] If the activity coefficient of the target group is lower than the activity threshold, the current target group is excluded;

[0034] The number of the screened target groups is counted to obtain a total number of the screened target groups, and the total number of the screened target groups is multiplied by a preset adjustment coefficient to obtain an update ratio.

[0035] In a preferred embodiment, in step S4, after entering the frequency adjustment mechanism, the update ratio is multiplied by the display frequency to obtain a display frequency adjustment amount, and the display frequency is subtracted by the display frequency adjustment amount to obtain a precise display frequency.

[0036] After obtaining the precise display frequency, the initial user's browsing time and browsing volume of the information of each target group are monitored and counted again in the precise display frequency, and the target groups are screened based on the browsing time and the browsing volume.

[0037] In a preferred embodiment, in step S4, the number of the screened target groups is detected, and if the number of the screened target groups is zero, an exit signal is generated and executed to exit the frequency adjustment mechanism.

[0038] If the number of the screened target groups is not zero, an update ratio is generated, and the precise display frequency is adjusted again according to the update ratio to obtain an updated precise display frequency.

[0039] After obtaining the updated precise display frequency, the initial user's browsing time and browsing volume of the information of each target group are monitored and counted again in the precise display frequency, and the target groups are screened based on the browsing time and the browsing volume, until an exit signal is generated.

[0040] In a preferred embodiment, in step S5, when exiting the frequency adjustment mechanism, the user's search frequency and active refresh frequency are collected in the precise display frequency according to the precise display frequency.

[0041] The total number of times that the initial user clicks the search box to search in the precise display frequency is counted to obtain the user's search frequency.

[0042] The number of times that the initial user triggers information refresh by refreshing the page in the precise display frequency is counted as the active refresh frequency.

[0043] In a preferred embodiment, in step S5, after the user's search frequency and the active refresh frequency are standardized, they are input into a polynomial regression calculation to obtain an exit group coefficient.

[0044] The exit group coefficient is compared with a preset group threshold value.

[0045] If the exit group coefficient exceeds the group threshold value, the group division mechanism is exited, and an alarm signal is generated.

[0046] If the exit coefficient is lower than the group threshold, the group division mechanism is retained.

[0047] The technical effects and advantages of this invention are as follows:

[0048] This invention establishes a group segmentation mechanism for initial users to obtain multiple potential groups similar to the initial users. Based on the display duration of information for each potential group, target information for that group is selected, and word frequencies are extracted for filtering to obtain representative words for each potential group. Based on these representative words, the potential groups are further filtered to obtain target groups. This target group information is then displayed on the initial user interface, with a set display frequency. The browsing time and number of views for each target group are monitored and statistically analyzed in real time. Based on the combined browsing time and number of views, target groups are filtered out. An update ratio is generated based on the number of filtered target groups, and a frequency adjustment mechanism is implemented. This update ratio is used to adjust the display frequency and obtain a precise display frequency. The filtering results for target groups are monitored in real time to determine whether to exit the frequency adjustment mechanism. When exiting the frequency adjustment mechanism, the number of user clicks and searches, as well as the number of active refreshes, are collected to determine whether to exit the group segmentation mechanism. This achieves personalized and precise information display, avoids user fatigue caused by information redundancy, and improves the utilization efficiency of information display and the rationality of information resource allocation. Attached Figure Description

[0049] Fig. 1 This is a flowchart illustrating the implementation of an information display frequency control method based on user behavior characteristics according to the present invention.

[0050] Fig. 2 This is a schematic diagram illustrating the steps of an information display frequency control method based on user behavior characteristics according to the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1

[0053] Please see Figs. 1-2 A method for controlling the frequency of information display based on user behavior characteristics, the specific operation process is as follows:

[0054] Step S1: Set up a group segmentation mechanism for the initial users to obtain multiple potential groups of the same type as the initial users, and select the target information of the same potential groups based on the display duration of each piece of information of the same potential group.

[0055] Step S2: Extract the word frequency of the target information of the same potential group and filter it to obtain the representative words of the same potential group. Filter the same potential group according to the representative words to obtain the target group, and display the target group information on the initial user interface.

[0056] Step S3: Set the display frequency, monitor and count the browsing time and number of views of the initial users on each target group's information in real time, filter out the target groups based on the combined browsing time and number of views, and generate an update ratio based on the number of filtered target groups.

[0057] Step S4: After generating the update ratio, enter the frequency adjustment mechanism, use the update ratio to adjust the display frequency and obtain the accurate display frequency, and monitor the screening results of the target group in real time to determine whether to exit the frequency adjustment mechanism.

[0058] Step S5: When exiting the frequency adjustment mechanism, collect the number of user clicks on search and the number of active refreshes to determine whether to exit the group segmentation mechanism.

[0059] The specific implementation process is as follows:

[0060] In step S1, the initial users are substituted into the set group division mechanism, and the initial information filled in by the initial users during registration is collected;

[0061] Among them, an initial user refers to a user who completes the first registration or login in the system and whose behavior characteristics cannot be found in the user behavior characteristic database. In this case, the current user is judged to be an initial user. The specific conditions for identifying and judging initial users can be set by those skilled in the art according to actual application needs and user behavior characteristic collection constraints, and will not be elaborated here.

[0062] Furthermore, the user behavior feature database refers to a database used to store user behavior data generated in the system. The user behavior features recorded in this database include, but are not limited to, user browsing behavior features, user interaction behavior features, user preference behavior features, and user access environment features. The specific types and collection scope of the behavior features are not limited and will not be elaborated here.

[0063] The initial information includes the user's gender, age, interest tags, IP address, and registration channel information. The specific amount of initial information is determined by the researchers in the information input box set when the user enters the registration UI interface, and will not be elaborated here.

[0064] Furthermore, interest tags are set with core dimensions based on the type of application platform (product categories on e-commerce platforms or content areas and formats on information platforms, etc.), and are collected through multiple selections and semantic matching.

[0065] The initial information is substituted into a preset group division mechanism to divide the initial users into groups. The initial users are divided into multiple preset groups, and the preset groups into which the initial users are assigned are marked to obtain multiple potential groups of the same type as the initial users.

[0066] Specifically, group segmentation mechanisms include classification methods based on attribute features, methods based on geographic tags, and segmentation methods based on clustering algorithms. For example, classification methods based on attribute features divide users into corresponding groups according to static information such as gender and age range; methods based on geographic tags divide users into groups based on the geographical range of their IP address or their network access location; and segmentation methods based on clustering algorithms use unsupervised learning algorithms such as K-means, DBSCAN, and hierarchical clustering to divide initial users into multiple potential groups of the same type. The specific implementation of the group segmentation mechanism is not limited and will not be elaborated here.

[0067] The specific implementation process of the clustering method is common knowledge to those skilled in the art and will not be elaborated here.

[0068] Furthermore, the researchers can set the following conditions to be met when obtaining multiple potential groups of the same type as the initial user: for example, when matching static attributes, the overlap of the initial user and the preset group's "initial information such as gender, age range, and IP address" exceeds the preset overlap, etc.

[0069] It should be noted that the preset group is a set of user categories set in the group division mechanism, used to match the initial users with potential groups; the preset group is set through historical user behavior statistical analysis and expert experience, and the specific number of preset groups and the method of dividing the group boundaries are not limited, and will not be elaborated here.

[0070] The built-in clock is used to count the display time of each piece of information in each potential group, and the information with the longest display time in each potential group is selected as the target information for the potential group.

[0071] The built-in clock refers to a timing device or module in the system used to record the time a user spends on each information interface. It can collect the display duration of each piece of information on the user's homepage in real time with a preset time accuracy, and use the collected data for subsequent information display statistics within the group and target information identification.

[0072] In step S2, the frequency of each word appearing in the target information of the same potential group is statistically analyzed to obtain the frequency of each word and the total frequency of each word in the target information. The ratio of the frequency of each word to the total frequency of each word is calculated to obtain the word frequency of each word in the target information of the same potential group.

[0073] The word recognition in the target information is based on natural language processing technology and word segmentation algorithm. The specific word recognition method and its implementation are not limited here, but are selected by those skilled in the art based on the characteristics of the corpus and the requirements of the semantic application scenario, and will not be elaborated here.

[0074] The word frequencies of each word in the target information of the same potential group are sorted from largest to smallest according to their numerical values, and the word with the largest word frequency value is selected as the representative word of the same potential group.

[0075] Furthermore, the above method was used to select representative terms for each potential group in the same category, which will not be elaborated here.

[0076] The frequency of representative words of the same potential group is called up, and the representative words of the same potential group are sorted from largest to smallest according to the frequency value. A preset number of representative words are selected as comparison representative words according to the sorting order, and the remaining representative words are used as candidate representative words.

[0077] By default, potential groups corresponding to representative terms are marked as target groups.

[0078] It is understandable that for the potential groups corresponding to the comparison representative words, since the representative words best reflect the core information of the potential groups, they are marked as the target groups by default. The comparison representative words are used to analyze whether the potential groups corresponding to the candidate representative words can be marked as the target groups.

[0079] It should be noted that the word frequency calculation method for representative words of similar potential groups has been described above and will not be repeated here. The preset comparison quantity refers to the number of words used to construct the comparison representative word set, which is preset by the system or dynamically configured by the experimenters based on historical sample data after the representative words of similar potential groups are sorted, according to the application scenario or system requirements. Specifically, the preset comparison quantity can be a fixed value (e.g., the first 3 or the first 5 words) or a dynamic value, that is, selected proportionally according to the number of representative words, such as taking the top 10% or 20% of the representative words in the sort, or taking them proportionally to the total number of representative words (the first 3 for less than 10 words, the first 5 for 10 to 20 words, the first 8 for more than 20 words), etc. It can also be determined based on experimental statistics, that is, by analyzing the effect of historical user data to select the optimal comparison quantity, etc., which will not be repeated here.

[0080] Each candidate representative word and each comparison representative word are vectorized, and the cosine similarity calculation formula is used to calculate the similarity between each candidate representative word and each comparison representative word.

[0081] Specifically, the formula for calculating cosine similarity is as follows:

[0082] ;

[0083] In the formula, Let x be the similarity between the x-th candidate representative word and the y-th comparison representative word. Let x be the vectorized representation of the x-th candidate representative word. Let y be the vectorized representation of the word represented by the y-th alignment. For vector dot product operation, The Euclidean norm of a vector;

[0084] In this experiment, the researchers can use the Word2Vec model for vectorization. The parameters are set as follows: for example, the word vector dimension is 100, the window size is 5, the minimum word frequency is 2 (to filter low-frequency meaningless words), and the training corpus is the user interaction text (such as product descriptions, news titles) selected by the platform, etc., which will not be elaborated here.

[0085] For example, given the existing comparison representative words A, B, and C, and the candidate representative words D, E, F, and G, then for candidate representative word D, calculate the AD similarity, BD similarity, and CD similarity; for candidate representative word E, calculate the AE similarity, BE similarity, and CE similarity, and so on, to calculate the similarity between each candidate representative word and each comparison representative word.

[0086] The similarity between each candidate representative word and each comparison representative word is grouped according to the number of candidate representative words to obtain the similarity groups of each candidate representative word.

[0087] As shown in the example above, the AD similarity, BD similarity, and CD similarity are divided into similarity groups of candidate representative words D, and the AE similarity, BE similarity, and CE similarity are divided into similarity groups of candidate representative words E. And so on, to complete the grouping operation according to the number of candidate representative words and obtain the similarity groups of each candidate representative word.

[0088] The target value of each candidate representative word is obtained by averaging the similarity groups of each candidate representative word group.

[0089] As shown in the example above, the AD similarity, BD similarity, and CD similarity are summed and then compared with their numbers to calculate the ratio, i.e., (AD similarity + BD similarity + CD similarity) / 3, to obtain the target value of candidate representative words D. Similarly, the target values ​​of candidate representative words E, F, and G are obtained.

[0090] The target value of each candidate representative word is compared with the preset similarity threshold;

[0091] If the target value of a candidate representative word exceeds the similarity threshold, then the potential group of the same type corresponding to the current candidate representative word will be filtered out.

[0092] If the target value of the candidate representative word is lower than the similarity threshold, then the potential group of the same type corresponding to the current candidate representative word is marked as the target group;

[0093] It should be noted that the similarity threshold was set by the researchers based on the number of candidate representative words and the frequency distribution of the compared representative words, which will not be elaborated here.

[0094] Understandably, the higher the target value of the candidate representative word, the higher the similarity between the candidate representative word and the comparison representative word, which makes it easier to generate duplicate recommendations, resulting in information redundancy and increased user browsing burden. Therefore, it is more necessary to screen out the potential group corresponding to the candidate representative word. Conversely, the lower the target value of the candidate representative word, the lower the similarity between the candidate representative word and the comparison representative word, which means that the potential group corresponding to the word can provide new and differentiated information. Therefore, it is more necessary to mark the potential group corresponding to the candidate representative word as the target group for accurate information display to the initial users.

[0095] The information corresponding to the target group obtained through screening will be used for the initial user interface display.

[0096] Optionally, the information from each target group can be aggregated to generate a target group information set. Each information item in the target group information set can be sorted according to preset display rules, such as sorting by information importance or time priority. Then, the sorted target group information can be mapped and allocated according to the initial user interface layout requirements. For example, in an e-commerce platform scenario, it can be allocated to the homepage recommendation module, search results module, or personalized information stream. For each display position, the display capacity or the number of information items per page can be set to avoid the interface being too crowded. These design methods will not be elaborated here.

[0097] In step S3, the browsing time and number of views of the initial users on each target group's information are monitored in real time within the set display frequency.

[0098] It should be noted that the display frequency refers to the period during which information about the target group is displayed to users. In this experiment, the researchers used historical user behavior characteristic parameters and push capacity parameters to monitor the browsing time and number of views of information about each target group in real time within the display frequency. The specific monitoring method is not limited and will not be elaborated here.

[0099] The browsing time of target group information refers to the cumulative time that an initial user spends on all information content under the same target group within a set display frequency. The logic for obtaining this information is to record the single dwell time of each piece of information belonging to the current target group during the initial user's browsing process within the display frequency, and then add up the dwell time of all information under the current target group within the display frequency to obtain the browsing time of target group information.

[0100] It should be noted that the method for recording each piece of information for the current target group is to count the time spent on each piece of information on the page, which will not be elaborated here;

[0101] The number of views of target group information refers to the cumulative number of times an initial user accesses information content under the same target group within a set display frequency. The logic for obtaining this number is to count each time an initial user clicks into any information of the current target group within the display frequency as a view and then count the number of views of the target group information.

[0102] Furthermore, the logic for obtaining the browsing time and browsing volume of the target group information mentioned above can be applied to the collection methods for the browsing time and browsing volume of various target group information, and will not be elaborated here.

[0103] The browsing time and number of views for each target group's information are standardized so that the browsing time and number of views for each target group's information are kept under the same dimension and the numerical expression range is between 0 and 1.

[0104] It should be noted that the standardization methods include, but are not limited to, standard linear transformation based on interval scaling, statistical Z-Score standardization method, or normalization method based on nonlinear mapping function. The application methods of standardization will not be elaborated here.

[0105] The activity coefficient of each target group is calculated by substituting the browsing time and number of views of each target group after standardization into the geometric mean method.

[0106] Among them, the geometric mean method refers to the statistical method of multiplying two or more non-negative values ​​and taking the square root, which is common knowledge in the art. Specifically, the activity coefficient of the current target group is obtained by multiplying the standardized value of the browsing time and the standardized value of the browsing volume of the target group information and taking the square root.

[0107] It should be noted that the longer the browsing time and the more views the target group information has, the higher the initial users' attention and interest in the current target group. This means that the current target group information is more representative and has more value for retention, and therefore, it is more necessary to retain the current target group.

[0108] Compare the activity coefficients of each target group with the activity thresholds;

[0109] If the activity level of the target group exceeds the activity threshold, the current target group will be retained.

[0110] If the activity coefficient of the target group is lower than the activity threshold, the current target group will be filtered out.

[0111] It should be noted that the activity threshold was set by the researchers based on historical user behavior data and the size of the target group, and will not be elaborated here.

[0112] The total number of target groups to be screened is obtained by counting the number of target groups to be screened, and the update ratio is generated by multiplying the total number of target groups to be screened by the preset adjustment coefficient.

[0113] The preset adjustment coefficients were obtained by the researchers based on historical display data and user behavior feedback. For example, the preset adjustment coefficients were set according to platform type and user activity level, and the update ratio was kept between 0 and 1. This will not be elaborated here.

[0114] It should be noted that the larger the total number of target groups that are filtered out, the higher the update ratio will be, which will allow for a greater adjustment to the display frequency of the remaining target groups, in order to reduce the display of low-activity or redundant information.

[0115] In step S4, after entering the frequency adjustment mechanism, the update ratio and the display frequency are multiplied to obtain the display frequency adjustment amount. The display frequency is then subtracted from the display frequency adjustment amount to obtain the precise display frequency.

[0116] Furthermore, if the display frequency adjustment amount is consistent with the display frequency, then the precise display frequency is zero, and there is no need to display information to the remaining target group, thus ending the method operation;

[0117] After obtaining the precise display frequency, the browsing time and number of views of the initial users on each target group's information are monitored and statistically analyzed again within the precise display frequency. The target groups are then filtered out based on the combined browsing time and number of views.

[0118] The number of target groups to be screened is detected. If the number of target groups to be screened is zero, an exit signal is generated and executed to exit the frequency adjustment mechanism.

[0119] If the number of target groups removed is not zero, an update ratio is generated, and the precision display frequency is adjusted again based on the update ratio to obtain the updated precision display frequency.

[0120] Furthermore, after obtaining the updated accurate display frequency, the system repeatedly monitors and statistically analyzes the browsing time and number of views of the initial users on each target group's information in real time, and filters out the target groups based on the combined browsing time and number of views until an exit signal is generated.

[0121] It is understandable that the loop operation in the frequency adjustment mechanism is actually a repetition of steps S3 to S4. Its purpose is to update and control the display frequency so that it can complete the capture of the initial user interest content in a short time. Optionally, the experimenters can set a bottom value in the frequency adjustment mechanism. When the accurate display frequency is less than the bottom value, the frequency adjustment mechanism is forcibly exited to avoid the target group information not being effectively displayed due to the low frequency.

[0122] It should be noted that the minimum effective display frequency is set as the bottoming-out value in the frequency adjustment mechanism. Specifically, the researchers set it as follows according to the platform type: for example, the bottoming-out value for e-commerce platforms is to update and display target group information once every 3 minutes to ensure that users have enough time to browse; the bottoming-out value for news platforms is to update and display target group information once every 2 minutes. Due to the high update frequency of news, a higher display frequency is required.

[0123] Furthermore, the forced exit logic exits the adjustment mechanism when the frequency of precise display is less than the bottom value, to avoid having no information to display.

[0124] In step S5, when exiting the frequency adjustment mechanism, the number of times the user clicks to search and the number of times the user actively refreshes are collected within the precise display frequency, based on the precise display frequency.

[0125] It should be noted that in practical applications, the frequency adjustment mechanism may cycle through multiple rounds, resulting in multiple updated accurate display frequencies. In this embodiment, only the accurate display frequency obtained for the first time after entering the frequency adjustment mechanism is used as an example for explanation. In fact, the collection and processing methods of the accurate display frequencies after multiple updates are the same as in this embodiment, and will not be repeated here.

[0126] The logic for obtaining the number of user search clicks is to count the total number of times the initial user clicked the search box to perform a search behavior within the precise display frequency.

[0127] The search box is an input interaction component set up by the researchers on the information display platform. Specific search behaviors include, but are not limited to, text input search, voice input search, and history click search. The display position and interface style of the search box are not limited and will not be described in detail here.

[0128] The logic for obtaining the number of active refreshes is to count the number of information refreshes triggered by the initial user refreshing the page within the precise display frequency as the number of active refreshes;

[0129] The number of user clicks for search and the number of active refreshes are standardized to keep the number of user clicks for search and the number of active refreshes under the same dimension and the value range is between 0 and 1.

[0130] The standardization process has been described above and will not be repeated here.

[0131] Substituting the standardized user search clicks and active refreshes into a multinomial regression calculation, we obtain the exit group coefficient, expressed by the following formula:

[0132] ;

[0133] In the formula, The coefficient for leaving the group. The number of user clicks for search after standardization. This refers to the number of active refreshes after standardization. To adjust the parameters, as well as The weighting coefficients corresponding to the standardized user click search count and the number of active refreshes;

[0134] It should be noted that the order and form of the polynomial regression calculation formula in this invention can be selected according to the actual application requirements. The specific order and function form are not limited here. Those skilled in the art can determine them based on the actual initial user behavior characteristics, which will not be elaborated here.

[0135] The weight coefficients mentioned in the multinomial regression calculation formula were obtained by our researchers based on the principle of minimizing the loss function and corresponding optimization algorithms (such as ordinary least squares, gradient descent, etc.). This is common knowledge to those skilled in the art and will not be elaborated here.

[0136] Specifically, the greater the number of times a user clicks to search and actively refreshes the page, the less satisfied the initial user's interest in the currently displayed content is. This results in a higher exit coefficient, indicating that the system judges the current group segmentation mechanism to be ineffective in capturing user information, and thus the greater the need for the exit mechanism.

[0137] The exit coefficient is compared with the preset group threshold.

[0138] If the exit coefficient exceeds the group threshold, the group will exit the group partitioning mechanism and an alarm signal will be generated.

[0139] If the exit coefficient is lower than the group threshold, the group division mechanism is retained;

[0140] It should be noted that the group threshold was set by the researchers based on historical user behavior data analysis and system information capture strategies, and will not be elaborated here.

[0141] Furthermore, when generating alarm signals, the system also generates messages such as "User interests not covered," "Group segmentation needs optimization," and "Abnormal attention frequency" and uploads them to the backend visualization port to notify system administrators, trigger automatic optimization processes, and record system operation logs.

[0142] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0143] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0144] In this document, the singular forms “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that terms such as “comprising / including” or “having” specify the presence of the stated features, integrals, steps, operations, components, parts, or combinations thereof, but do not preclude the possibility of the presence or addition of one or more other features, integrals, steps, operations, components, parts, or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0145] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0146] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for controlling the frequency of information display based on user behavior characteristics, characterized in that: Includes the following steps: Step S1: Set up a group segmentation mechanism for the initial users to obtain multiple potential groups of the same type as the initial users, and select the target information of the same potential groups based on the display duration of each piece of information of the same potential groups; Step S2: Extract the word frequency of the target information of the same potential group and filter it to obtain the representative words of the same potential group. Filter the same potential group according to the representative words to obtain the target group, and display the target group information on the initial user interface. Step S3: Set the display frequency, monitor and count the browsing time and number of views of the initial users on each target group's information in real time, filter out the target groups based on the combined browsing time and number of views, and generate an update ratio based on the number of filtered target groups. Step S4: After generating the update ratio, enter the frequency adjustment mechanism, use the update ratio to adjust the display frequency and obtain the accurate display frequency, and monitor the screening results of the target group in real time to determine whether to exit the frequency adjustment mechanism. Step S5: When exiting the frequency adjustment mechanism, collect the number of user clicks on search and the number of active refreshes to determine whether to exit the group segmentation mechanism.

2. The information display frequency control method based on user behavior characteristics according to claim 1, characterized in that: In step S1, the initial users are substituted into the set group division mechanism, and the initial information filled in by the initial users during registration is collected; The initial information is substituted into a preset group division mechanism to divide the initial users into groups. The initial users are divided into multiple preset groups, and the preset groups into which the initial users are assigned are marked to obtain multiple potential groups of the same type as the initial users. The system uses a built-in clock to track the display duration of information within each potential group and selects the information with the longest display duration as the target information for that potential group.

3. The information display frequency control method based on user behavior characteristics according to claim 1, characterized in that: In step S2, the frequency of each word appearing in the target information of the same potential group is statistically analyzed to obtain the frequency of each word and the total frequency of each word in the target information. The ratio of the frequency of each word to the total frequency of each word is calculated to obtain the word frequency of each word in the target information of the same potential group. The word frequencies of each word in the target information of the same potential group are sorted from largest to smallest according to their numerical values, and the word with the largest word frequency value is selected as the representative word of the same potential group. The frequency of representative words of the same potential group is called up, and the representative words of the same potential group are sorted from largest to smallest according to the frequency value. A preset number of representative words are selected as comparison representative words according to the sorting order, and the remaining representative words are used as candidate representative words. By default, potential groups of the same type corresponding to the representative terms are marked as the target group.

4. The information display frequency control method based on user behavior characteristics according to claim 2, characterized in that: In step S2, each candidate representative word and each comparison representative word are vectorized, and the cosine similarity calculation formula is used to calculate the similarity between each candidate representative word and each comparison representative word. The similarity between each candidate representative word and each comparison representative word is grouped according to the number of candidate representative words to obtain the candidate representative word similarity groups; The target value of each candidate representative word is obtained by averaging the similarity groups of each candidate representative word. The target value of each candidate representative word is compared with the preset similarity threshold; If the target value of a candidate representative word exceeds the similarity threshold, then the potential group of the same type corresponding to the current candidate representative word will be filtered out. If the target value of the candidate representative word is lower than the similarity threshold, then the potential group of the same type corresponding to the current candidate representative word is marked as the target group; The information corresponding to the selected target group will be used for the initial user interface display.

5. The information display frequency control method based on user behavior characteristics according to claim 1, characterized in that: In step S3, the browsing time and number of views of the initial users on each target group's information are monitored in real time within the set display frequency. During the initial user browsing process within the display frequency, the single dwell time of each piece of information belonging to the current target group is recorded, and the dwell time of all information under the current target group within the display frequency is accumulated to obtain the browsing time of the target group information; Each time an initial user clicks on any information for the current target group within the display frequency is counted as one view, and the number of views for the target group's information is calculated.

6. The information display frequency control method based on user behavior characteristics according to claim 5, characterized in that: In step S3, the browsing time and number of views of each target group after standardization are substituted into the geometric mean method to calculate the activity coefficient of each target group. Compare the activity coefficients of each target group with the activity thresholds; If the activity level of the target group exceeds the activity threshold, the current target group will be retained. If the activity coefficient of the target group is lower than the activity threshold, the current target group will be filtered out. The total number of target groups to be screened is obtained by counting the number of target groups to be screened. The update ratio is generated by multiplying the total number of target groups to be screened by the preset adjustment coefficient.

7. The information display frequency control method based on user behavior characteristics according to claim 1, characterized in that: In step S4, after entering the frequency adjustment mechanism, the update ratio and the display frequency are multiplied to obtain the display frequency adjustment amount. The display frequency is then subtracted from the display frequency adjustment amount to obtain the precise display frequency. After obtaining the precise display frequency, the browsing time and number of views of the initial users on each target group's information are monitored and statistically analyzed again within the precise display frequency. The target group is then filtered out based on the combined browsing time and number of views.

8. The information display frequency control method based on user behavior characteristics according to claim 7, characterized in that: In step S4, the number of target groups to be screened is detected. If the number of target groups to be screened is zero, an exit signal is generated and executed to exit the frequency adjustment mechanism. If the number of target groups removed is not zero, an update ratio is generated, and the precision display frequency is adjusted again based on the update ratio to obtain the updated precision display frequency. After obtaining the updated accurate display frequency, the initial users' browsing time and number of views for each target group's information are repeatedly monitored and statistically analyzed. The target groups are then filtered out based on the combined browsing time and number of views until an exit signal is generated.

9. The information display frequency control method based on user behavior characteristics according to claim 1, characterized in that: In step S5, when exiting the frequency adjustment mechanism, the number of times the user clicks to search and the number of times the user actively refreshes are collected within the precise display frequency, based on the precise display frequency. The total number of times a user clicks the search box to perform a search action is counted within the precise display frequency range to obtain the user search click count; Within the precise display frequency, the number of times the initial user refreshes the information triggered by refreshing the page is counted as the number of active refreshes.

10. The information display frequency control method based on user behavior characteristics according to claim 9, characterized in that: In step S5, the number of user clicks on the search and the number of active refreshes are standardized and then entered into a multinomial regression calculation to obtain the exit coefficient. The exit coefficient is compared with the preset group threshold. If the exit coefficient exceeds the group threshold, the group will exit the group partitioning mechanism and an alarm signal will be generated. If the exit coefficient is lower than the group threshold, the group division mechanism is retained.

Images