Method, device, equipment, storage medium and program product for sending display information

Abstract

The present application discloses a method, apparatus, device, storage medium and program product for sending display information, and belongs to the field of computer technology. The method comprises: screening multiple tags based on statistical data corresponding to multiple tags to obtain multiple recommendable tags; determining multiple reference tags corresponding to the target account based on the historical behavior data of the target account; determining the intersection between the multiple recommendable tags and the multiple reference tags as the target recommendation tag corresponding to the target account; obtaining the target display information corresponding to each target recommendation tag, and sending the target display information to the target terminal to which the target account belongs. By adopting the present application, it is possible to determine target recommendation tags that are of both quality and recommendability and that the user is interested in, thereby improving the recommendation effect.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, device, storage medium, and program product for sending presentation information. Background Art

[0002] Many current applications have recommendation functions, which improve user stickiness and enhance user experience by recommending information that may be of interest to users.

[0003] The usual recommendation method is to directly recommend display information with a high click-through rate or collection rate to users.

[0004] However, the above recommendation method is relatively crude and may recommend information that the user is not interested in, resulting in poor recommendation effect. Summary of the Invention

[0005] The embodiment of the present application provides a method for sending display information, which can solve the problem that the recommendation method in the prior art is relatively rough and thus leads to poor recommendation effect.

[0006] In a first aspect, a method for sending presentation information is provided, the method comprising:

[0007] Based on statistical data corresponding to the multiple tags, the multiple tags are screened to obtain multiple recommended tags, wherein the tags are information used to indicate content characteristics of the displayed information;

[0008] Determining a plurality of reference tags corresponding to the target account based on historical behavior data of the target account;

[0009] Determining an intersection of the multiple recommendable tags and the multiple reference tags as a target recommended tag corresponding to the target account;

[0010] Target display information corresponding to each target recommendation tag is obtained, and the target display information is sent to the target terminal to which the target account belongs.

[0011] In a possible implementation, the method further includes:

[0012] For each piece of display information, a tag corresponding to the display information is determined based on the display information and a keyword extraction model.

[0013] In a possible implementation, the plurality of tags are filtered based on statistical data corresponding to the plurality of tags to obtain a plurality of recommended tags, including:

[0014] Determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the multiple tags;

[0015] Based on the user interest intensity corresponding to each tag and a preset intensity threshold, the multiple tags are screened to obtain multiple recommendable tags.

[0016] In a possible implementation, the statistical data includes click statistical data and favorite statistical data.

[0017] In a possible implementation, the click statistics include a first click rate corresponding to all users within a first preset time period and a second click rate corresponding to historical click users within the first preset time period, wherein the historical click users are users who have clicked on any display information corresponding to the tag within a second preset time period before the first preset time period and have been exposed to the display information corresponding to the tag within the first preset time period;

[0018] The collection statistics include a first collection rate corresponding to all users in the first preset period and a second collection rate corresponding to historical collection users in the first preset period, wherein the historical collection users are users who have collected any display information corresponding to the tag in the second preset period and have been exposed to the display information corresponding to the tag in the first preset period;

[0019] The determining, based on the statistical data corresponding to the plurality of tags, the user interest intensity corresponding to each tag includes:

[0020] For each tag, determining the user click interest intensity corresponding to the tag based on the first click rate and the second click rate corresponding to the tag;

[0021] Determining the user's collection interest intensity corresponding to the tag based on the first collection rate and the second collection rate corresponding to the tag;

[0022] The user interest intensity corresponding to the tag is determined based on the user click interest intensity and the user collection interest intensity corresponding to the tag.

[0023] In one possible implementation, determining a plurality of reference tags corresponding to the target account based on historical behavior data of the target account includes:

[0024] Determining a short-term tag corresponding to the target account based on the historical behavior data of the target account within a third preset time period, wherein the third preset time period is a recent time period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information is less than a preset duration;

[0025] determining a long-term tag corresponding to the target account based on historical behavior data of the target account within a fourth preset time period;

[0026] Based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account is determined.

[0027] In one possible implementation, the historical behavior data includes historical click data and historical collection data, the historical click data includes one or more historical click display information and the click time point corresponding to each historical click display information, and the historical collection data includes one or more historical collection display information and the collection time point corresponding to each historical collection display information.

[0028] In a possible implementation, determining a short-term tag corresponding to the target account based on the historical behavior data of the target account within a third preset time period includes:

[0029] Obtaining short-term click tags corresponding to historical click data of the target account within a third preset time period and short-term collection tags corresponding to historical collection data of the target account within a third preset time period;

[0030] Determine the most recent click time point corresponding to each short-term click tag among the click time points corresponding to one or more historical click display information corresponding to each short-term click tag;

[0031] Selecting a first preset number of short-term click tags in descending order of the most recent click time points to obtain a target short-term click tag corresponding to the target account;

[0032] Determine the most recent collection time point corresponding to each short-term collection tag among the collection time points corresponding to one or more historical collection display information corresponding to each short-term collection tag;

[0033] Selecting a second preset number of short-term favorite tags in descending order of the most recent favorite time point to obtain a target short-term favorite tag corresponding to the target account;

[0034] Deduplication processing is performed on the target short-term click tag and the target short-term collection tag to obtain a short-term tag corresponding to the target account.

[0035] In a possible implementation, determining the long-term tag corresponding to the target account based on the historical behavior data of the target account within a fourth preset time period includes:

[0036] Obtaining a long-term click tag corresponding to the historical click data of the target account within a fourth preset time period and a long-term collection tag corresponding to the historical collection data of the target account within the fourth preset time period;

[0037] Deduplication is performed on the long-term click tag and the long-term favorite tag to obtain a reference long-term tag corresponding to the target account;

[0038] Determining, based on historical behavior data within a fourth preset time period corresponding to each reference long-term tag, a correlation between each reference long-term tag and the target account;

[0039] A third preset number of reference long-term tags are selected in descending order of the relevance to obtain a long-term tag corresponding to the target account.

[0040] In a possible implementation, after determining the long-term tag corresponding to the target account, the method further includes:

[0041] Get the feature vector corresponding to each label;

[0042] For each long-term tag, respectively calculate the similarity between each tag and the long-term tag;

[0043] Determining an extended tag corresponding to each long-term tag based on a similarity between each tag and the long-term tag;

[0044] The determining, based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account includes:

[0045] A reference tag corresponding to the target account is determined based on the short-term tag corresponding to the target account, the long-term tag corresponding to the target account, and the extended tag corresponding to each long-term tag.

[0046] In a possible implementation, obtaining target display information corresponding to each target recommendation tag includes:

[0047] For each target recommendation tag, based on statistical data of each display information corresponding to the target recommendation tag, determine a score for each display information corresponding to the target recommendation tag;

[0048] A fourth preset number of display information is selected in descending order of scores to obtain target display information corresponding to the target recommendation tag.

[0049] In a second aspect, a device for sending presentation information is provided, the device comprising:

[0050] a screening module, configured to screen the multiple tags based on statistical data corresponding to the multiple tags to obtain multiple recommended tags, wherein the tags are information indicating content characteristics of the displayed information;

[0051] A first determining module, configured to determine a plurality of reference tags corresponding to the target account based on historical behavior data of the target account;

[0052] A second determining module is configured to determine an intersection of the plurality of recommendable tags and the plurality of reference tags as a target recommended tag corresponding to the target account;

[0053] The sending module is used to obtain target display information corresponding to each target recommendation tag and send the target display information to the target terminal to which the target account belongs.

[0054] In a possible implementation, the apparatus further includes a label determination module configured to:

[0055] For each piece of display information, a tag corresponding to the display information is determined based on the display information and a keyword extraction model.

[0056] In a possible implementation, the screening module is configured to:

[0057] Determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the multiple tags;

[0058] Based on the user interest intensity corresponding to each tag and a preset intensity threshold, the multiple tags are screened to obtain multiple recommendable tags.

[0059] In a possible implementation, the statistical data includes click statistical data and favorite statistical data.

[0060] In a possible implementation, the click statistics include a first click rate corresponding to all users within a first preset time period and a second click rate corresponding to historical click users within the first preset time period, wherein the historical click users are users who have clicked on any display information corresponding to the tag within a second preset time period before the first preset time period and have been exposed to the display information corresponding to the tag within the first preset time period;

[0061] The collection statistics include a first collection rate corresponding to all users in the first preset period and a second collection rate corresponding to historical collection users in the first preset period, wherein the historical collection users are users who have collected any display information corresponding to the tag in the second preset period and have been exposed to the display information corresponding to the tag in the first preset period;

[0062] The screening module is used to:

[0063] For each tag, determining the user click interest intensity corresponding to the tag based on the first click rate and the second click rate corresponding to the tag;

[0064] Determining the user's collection interest intensity corresponding to the tag based on the first collection rate and the second collection rate corresponding to the tag;

[0065] The user interest intensity corresponding to the tag is determined based on the user click interest intensity and the user collection interest intensity corresponding to the tag.

[0066] In a possible implementation, the first determining module is configured to:

[0067] Determining a short-term tag corresponding to the target account based on the historical behavior data of the target account within a third preset time period, wherein the third preset time period is a recent time period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information is less than a preset duration;

[0068] determining a long-term tag corresponding to the target account based on historical behavior data of the target account within a fourth preset time period;

[0069] Based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account is determined.

[0070] In one possible implementation, the historical behavior data includes historical click data and historical collection data, the historical click data includes one or more historical click display information and the click time point corresponding to each historical click display information, and the historical collection data includes one or more historical collection display information and the collection time point corresponding to each historical collection display information.

[0071] In a possible implementation, the first determining module is configured to:

[0072] Obtaining short-term click tags corresponding to historical click data of the target account within a third preset time period and short-term collection tags corresponding to historical collection data of the target account within a third preset time period;

[0073] Determine the most recent click time point corresponding to each short-term click tag among the click time points corresponding to one or more historical click display information corresponding to each short-term click tag;

[0074] Selecting a first preset number of short-term click tags in descending order of the most recent click time points to obtain a target short-term click tag corresponding to the target account;

[0075] Determine the most recent collection time point corresponding to each short-term collection tag among the collection time points corresponding to one or more historical collection display information corresponding to each short-term collection tag;

[0076] Selecting a second preset number of short-term favorite tags in descending order of the most recent favorite time point to obtain a target short-term favorite tag corresponding to the target account;

[0077] Deduplication processing is performed on the target short-term click tag and the target short-term collection tag to obtain a short-term tag corresponding to the target account.

[0078] In a possible implementation, the first determining module is configured to:

[0079] Obtaining a long-term click tag corresponding to the historical click data of the target account within a fourth preset time period and a long-term collection tag corresponding to the historical collection data of the target account within the fourth preset time period;

[0080] Deduplication is performed on the long-term click tag and the long-term favorite tag to obtain a reference long-term tag corresponding to the target account;

[0081] Determining, based on historical behavior data within a fourth preset time period corresponding to each reference long-term tag, a correlation between each reference long-term tag and the target account;

[0082] A third preset number of reference long-term tags are selected in descending order of the relevance to obtain a long-term tag corresponding to the target account.

[0083] In a possible implementation, after determining the long-term tag corresponding to the target account, the first determining module is further configured to:

[0084] Get the feature vector corresponding to each label;

[0085] For each long-term tag, respectively calculate the similarity between each tag and the long-term tag;

[0086] Determining an extended tag corresponding to each long-term tag based on a similarity between each tag and the long-term tag;

[0087] The determining, based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account includes:

[0088] A reference tag corresponding to the target account is determined based on the short-term tag corresponding to the target account, the long-term tag corresponding to the target account, and the extended tag corresponding to each long-term tag.

[0089] In a possible implementation, the sending module is configured to:

[0090] For each target recommendation tag, based on statistical data of each display information corresponding to the target recommendation tag, determine a score for each display information corresponding to the target recommendation tag;

[0091] A fourth preset number of display information is selected in descending order of scores to obtain target display information corresponding to the target recommendation tag.

[0092] According to a third aspect, a computer device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the operation performed by the method for sending display information.

[0093] In a fourth aspect, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the instruction is loaded and executed by a processor to implement the operations performed by the method for sending display information.

[0094] In a fifth aspect, a computer program product is provided, wherein the computer program product includes at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the operations performed by the method for sending display information.

[0095] The beneficial effects of the technical solution provided by the embodiment of the present application are as follows: the solution mentioned in the embodiment of the present application can screen multiple tags based on the statistical data corresponding to the multiple tags, thereby obtaining multiple high-quality, recommendable and recommendable recommendable tags, and then based on the historical behavior data of the target account, determine the multiple reference tags that the target account corresponds to and that the users are more interested in, and then determine the intersection between the multiple recommendable tags and the multiple reference tags as the target recommendation tags corresponding to the target account, and finally obtain the target display information corresponding to each target recommendation tag, and send the target display information to the target terminal to which the target account belongs. By adopting this application, it is possible to determine target recommendation tags that are of high quality, recommendable, and of interest to users, thereby improving the recommendation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0097] Figure 1 This is a flow chart of a method for sending display information provided by an embodiment of the present application;

[0098] Figure 2 This is a flow chart of a method for determining a short-term tag provided by an embodiment of the present application;

[0099] Figure 3 This is a flow chart of a method for determining a long-term tag provided in an embodiment of the present application;

[0100] Figure 4 This is a schematic diagram of the structure of a device for sending display information provided in an embodiment of the present application;

[0101] Figure 5 This is a structural block diagram of a server provided in an embodiment of the present application. DETAILED DESCRIPTION

[0102] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0103] The embodiment of the present application provides a method for sending display information, which can be implemented by a server. The server can be a single server or a server cluster composed of multiple servers.

[0104] The server may include a processor, a memory, a communication component, etc., and the processor is connected to the memory and the communication component respectively.

[0105] The processor may be a CPU (Central Processing Unit). The processor may be used to read instructions and process data, for example, filtering multiple tags to obtain multiple recommended tags, determining multiple reference tags corresponding to a target account, determining target recommended tags corresponding to the target account, obtaining target display information corresponding to each target recommended tag, and so on.

[0106] The memory may include ROM (Read-Only Memory), RAM (Random Access Memory), CD-ROM (Compact Disc Read-Only Memory), magnetic disks, optical data storage devices, etc. The memory may be used for data storage, for example, for storing data on multiple tags, data on multiple recommendable tags, historical behavior data on target accounts, data on multiple reference tags corresponding to a determined target account, data on target recommended tags corresponding to a determined target account, data on target display information corresponding to each tag, and the like.

[0107] The communication component may be a wired network connector, a wireless fidelity module, a Bluetooth module, a cellular network communication module, etc. The communication component may be used to receive and send signals, for example, to send target display information to a target terminal to which a target account belongs, and so on.

[0108] Figure 1 This is a flow chart of a method for sending display information provided by an embodiment of the present application. Figure 1 , the embodiment includes:

[0109] 101. Based on the statistical data corresponding to the multiple tags, the multiple tags are filtered to obtain multiple recommended tags.

[0110] The tag is information used to indicate the content characteristics of the displayed information.

[0111] In practice, each display information corresponds to one or more tags. When recommending display information to a target account, you can first obtain the statistical data corresponding to each tag, and then filter out high-quality and recommendable tags based on the statistical data.

[0112] Optionally, the method of filtering multiple tags using statistical data may include determining the user interest strength corresponding to each tag based on the statistical data corresponding to each of the multiple tags. Based on the user interest strength corresponding to each tag and a preset strength threshold, the multiple tags are filtered to obtain multiple recommended tags.

[0113] During implementation, staff can pre-set a preset strength threshold. When filtering multiple tags, the user interest strength corresponding to each tag can be calculated based on the statistical data corresponding to each tag. This user interest strength is used to represent the user's level of interest in the tag. Tags with user interest strength greater than or equal to the preset strength threshold can then be determined as recommended tags.

[0114] 102. Determine multiple reference tags corresponding to the target account based on historical behavior data of the target account.

[0115] During implementation, when it is necessary to recommend display information to the target account, it is also necessary to determine multiple reference tags corresponding to the target account through the historical behavior data of the target account. The reference tags are tags that the user to whom the target account belongs is interested in.

[0116] Optionally, the method for determining a reference tag using the target account's historical behavior data may be as follows: determining a short-term tag corresponding to the target account based on the target account's historical behavior data within a third preset time period, wherein the third preset time period is the most recent time period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information is less than a preset duration. Determining a long-term tag corresponding to the target account based on the target account's historical behavior data within a fourth preset time period. Determining a reference tag corresponding to the target account based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account.

[0117] In implementation, short-term tags are tags that the user of the target account is interested in in the recent period, and long-term tags are tags that the user of the target account is interested in in the long term.

[0118] When determining the short-term tag corresponding to the target account, historical behavior data from a relatively short recent period can be obtained, i.e., the historical behavior data within the aforementioned third preset time period. The third preset time period is the most recent period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information within this period is less than a preset duration. Typically, this third preset time period can be called a session. The preset duration can be set to any reasonable duration, for example, half an hour, etc., and this embodiment of the application is not limited to this.

[0119] When determining the long-term tag corresponding to the target account, historical behavior data for a longer period of time can be obtained, namely the fourth preset period mentioned above. The fourth preset period can be set to a longer duration than the third preset period, for example, the fourth preset period can be set to half a year, etc., or other reasonable time periods, which are not limited in this embodiment of the present application.

[0120] After determining the short-term label and the corresponding long-term label corresponding to the target account, the short-term label and the long-term label can be directly determined as the reference label corresponding to the target account, or the determined short-term label and the long-term label can be filtered by other means, and the labels obtained after filtering can be determined as the reference label corresponding to the target account.

[0121] 103. Determine the intersection of the multiple recommendable tags and the multiple reference tags as the target recommendation tag corresponding to the target account.

[0122] In implementation, after determining multiple recommendable tags and multiple reference tags corresponding to the target account, the intersection of the two can be taken, that is, a target recommendation tag that has both recommendation quality and meets the interest requirements of the user to which the target account belongs can be obtained.

[0123] 104. Obtain target display information corresponding to each target recommendation tag, and send the target display information to the target terminal to which the target account belongs.

[0124] In practice, all display information corresponding to each target recommendation tag can be directly used as target display information and pushed to the target terminal. Alternatively, the display information corresponding to the target recommendation tag can be filtered to obtain the target display information and then pushed to the target terminal.

[0125] Optionally, a method for screening display information corresponding to a target recommendation tag may be as follows: for each target recommendation tag, based on statistical data of each display information corresponding to the target recommendation tag, a score for each display information corresponding to the target recommendation tag is determined. A fourth preset number of display information is selected in descending order of score to obtain the target display information corresponding to the target recommendation tag.

[0126] Among them, the statistical data of the display information may include the click rate and collection rate of the fifth preset time period, and the score of the display information may be determined by: the sum of the click rate and collection rate of the display information in the fifth preset time period. Of course, other calculation methods may also be used, and the embodiments of the present application are not limited to this.

[0127] Before step 101, a corresponding label may be determined for each display information. There are many methods for determining the label, and the following is one example:

[0128] For each piece of display information, a label corresponding to the display information is determined based on the display information and the keyword extraction model.

[0129] During implementation, information such as the attributes of the display information or the content of the display information can be input into a trained keyword extraction model to obtain one or more keywords corresponding to the display information. The keyword can be directly used as the label corresponding to the display information, or one or more keywords can be combined using a preset template to obtain one or more labels corresponding to the display information. For example, the display information keywords output by the keyword extraction model can be divided into the following categories: time, cuisine, taste, quality, venue, activity, place, etc. The preset templates may include: time + cuisine, taste + cuisine, category + cuisine, place + cuisine, venue + activity, time + venue, taste + venue, quality + venue, place + venue, etc.

[0130] In step 101, multiple tags may be filtered based on preset rules. The preset rules may be set as required. The following are two of the preset rules:

[0131] The first

[0132] The staff can pre-set a first threshold value for the number of displayed information. During screening, the number of displayed information corresponding to each tag can be obtained, and tags with a number of corresponding displayed information less than the first threshold value for the number of displayed information can be removed.

[0133] The second

[0134] Each display information corresponds to one or more categories. Categories are more coarse-grained than tags. Staff can pre-set the second display information number threshold.

[0135] During screening, for each tag, obtain one or more categories of display information corresponding to the tag, and determine the target category with the largest number of corresponding display information. If the number of display information corresponding to the target category is less than the second display information number threshold, the tag is removed.

[0136] The preset rules may also be any reasonable settings, which is not limited in the embodiments of the present application.

[0137] There are many possibilities for the statistical data in the above step 101. The following is a more detailed introduction using the statistical data including click statistical data and favorite statistical data as an example. Of course, other statistical data may also be included, and this embodiment of the application does not limit this.

[0138] More specifically, the click statistics may include a first click rate corresponding to all users within a first preset period and a second click rate corresponding to historical click users within the first preset period, wherein the historical click users are users who have clicked on any display information corresponding to the label within a second preset period before the first preset period and have exposed the display information corresponding to the label within the first preset period.

[0139] The first click-through rate for a tag is the ratio of the number of clicks on the display information corresponding to the tag by all users during the first preset time period to the number of impressions. The second click-through rate for a tag is the ratio of the number of clicks on the display information corresponding to the tag by historical click users during the first preset time period to the number of impressions.

[0140] The second preset period is adjacent to the first preset period and is located before the first preset period. For example, if the first preset period is set to one day and the second preset period is set to one week, when determining the second click-through rate corresponding to label A on March 10, users who clicked on any display information corresponding to label A between March 3 and March 9 and were exposed on March 10 can be obtained as historical click users.

[0141] The first click-through rate can be used to represent the interest level of all users in the tag, and the second click-through rate can be used to represent the user's interest level in the display information corresponding to the tag when it is recommended again to the user who clicked on the display information corresponding to the tag.

[0142] Similar to click statistics, collection statistics may include a first collection rate corresponding to all users within a first preset period and a second collection rate corresponding to historical collection users within the first preset period, wherein historical collection users are users who have performed a collection operation on any display information corresponding to the tag within the second preset period and have performed an exposure operation on the display information corresponding to the tag within the first preset period.

[0143] Based on the settings of the above click statistics and favorite statistics, the method for determining the user interest intensity corresponding to each tag can be as follows:

[0144] For each tag, the user's click interest strength for the tag is determined based on the tag's first click-through rate and second click-through rate. The user's favorite interest strength for the tag is determined based on the tag's first favorite rate and second favorite rate. The user's click interest strength and favorite interest strength for the tag are determined based on the tag's click interest strength and favorite interest strength.

[0145] In implementation, the user interest strength corresponding to a tag may be determined based on the following formulas 1-3.

[0146]

[0147]

[0148]

[0149] Among them, γ is the user interest intensity, is the user's click interest intensity, is the user's collection interest strength, is the first click rate, is the second click rate, A and B are weights, It is the first collection rate, It is the second collection rate.

[0150] Optionally, in the case where a tag has multiple corresponding display information, the user interest intensity corresponding to the tag can be determined based on the statistical data corresponding to each display information. For example, in an application with a consumption function, it is necessary to recommend review content to the target account, and in the application, each tag can correspond to at least one review content and at least one POI (Point Of Interest). When determining the user interest intensity corresponding to each tag, the user interest intensity corresponding to the review content can be calculated based on the statistical data of the review content corresponding to the tag, and the user interest intensity corresponding to the POI can be calculated based on the statistical data of the POI corresponding to the tag, and then the two can be added together to obtain the user interest intensity corresponding to the tag.

[0151] There are many possibilities for the historical behavior data in the above step 102. The following is a more detailed introduction using the historical behavior data including historical click data and historical collection data as an example. Of course, other statistical data may also be included, which is not limited in the embodiment of the present application.

[0152] More specifically, the historical click data may include one or more historical click display information and the click time point corresponding to each historical click display information, wherein the historical click display information is the display information clicked by the target account within the third preset time period or the fourth preset time period, and the click time point corresponding to the historical click display information is the time point when the user to which the target account belongs clicks the historical click display information.

[0153] Similar to the historical click data, the historical collection data may include one or more historical collection display information and the collection time point corresponding to each historical collection display information, wherein the historical collection display information is the display information that the target account has collected within the third preset time period or the fourth preset time period, and the collection time point corresponding to the historical collection display information is the time point when the user to which the target account belongs collects the historical collection display information.

[0154] Optionally, the above-mentioned historical click display information and historical collection display information may include multiple display information, for example, review content and POI, etc., which is not limited in this embodiment of the present application.

[0155] Based on the above settings of historical click data and historical collection data, the method of determining the short-term label corresponding to the target account can be found in Figure 2 , corresponding to the following:

[0156] 201. Obtain short-term click tags corresponding to historical click data of a target account within a third preset time period and short-term collection tags corresponding to historical collection data of the target account within a third preset time period.

[0157] In implementation, one or more tags corresponding to multiple pieces of historical click display information included in the historical click data within the third preset time period may be obtained, and then these tags may be deduplicated to obtain short-term click tags. Similarly, one or more tags corresponding to multiple pieces of historical collection display information included in the historical collection data within the third preset time period may be obtained, and then these tags may be deduplicated to obtain short-term collection tags.

[0158] 202. Determine the most recent click time point corresponding to each short-term click tag among the click time points corresponding to one or more historical click display information corresponding to each short-term click tag.

[0159] During implementation, for each short-term click tag, the click time points corresponding to one or more historical click display information corresponding to the short-term click tag are obtained, and then the click time point closest to the current time is determined as the most recent click time point corresponding to the short-term click tag.

[0160] 203. Select a first preset number of short-term click tags in descending order of the most recent click time points to obtain a target short-term click tag corresponding to the target account.

[0161] In implementation, multiple short-term click tags are arranged in descending order according to their most recent click time points, and the first first preset number of short-term click tags are determined as the target short-term click tags corresponding to the target account. The first preset number can be any reasonable number, for example, 50, etc., and is not limited in this embodiment of the application.

[0162] 204. Determine the most recent collection time point corresponding to each short-term collection tag among the collection time points corresponding to one or more historical collection display information corresponding to each short-term collection tag.

[0163] In implementation, for each short-term collection tag, the collection time points corresponding to one or more historical collection display information corresponding to the short-term collection tag are obtained, and then the collection time point closest to the current time is determined as the latest collection time point corresponding to the short-term collection tag.

[0164] 205. Select a second preset number of short-term favorite tags in descending order of the most recent favorite time point to obtain a target short-term favorite tag corresponding to the target account.

[0165] During implementation, the multiple short-term favorite tags are arranged in descending order of their most recent favorite time point, and the first second preset number of short-term favorite tags are determined as the target short-term favorite tags corresponding to the target account. The second preset number can be any reasonable number, such as 30, and can be the same as or different from the first preset number, and this embodiment of the application is not limited thereto.

[0166] It is understandable that there is no temporal order between steps 202 - 203 and steps 204 - 205 .

[0167] 206. De-duplicate the target short-term click tag and the target short-term favorite tag to obtain a short-term tag corresponding to the target account.

[0168] Based on the above settings of historical click data and historical collection data, the method for determining the long-term label corresponding to the target account can be found in Figure 3 , corresponding to the following:

[0169] 301. Obtain a long-term click tag corresponding to historical click data of a target account within a fourth preset time period and a long-term collection tag corresponding to historical collection data of the target account within a fourth preset time period.

[0170] In implementation, one or more tags corresponding to multiple pieces of historical click display information included in the historical click data within the fourth preset time period may be obtained, and then these tags may be deduplicated to obtain long-term click tags. Similarly, one or more tags corresponding to multiple pieces of historical collection display information included in the historical collection data within the fourth preset time period may be obtained, and then these tags may be deduplicated to obtain long-term collection tags.

[0171] 302. De-duplicate the long-term click tag and the long-term favorite tag to obtain a reference long-term tag corresponding to the target account.

[0172] 303. Determine the relevance between each reference long-term tag and the target account based on the historical behavior data within a fourth preset time period corresponding to each reference long-term tag.

[0173] In practice, there are many methods for calculating the relevance between the reference long-term tag and the target account. In this embodiment, the TF-IDF (Term Frequency–Inverse Document Frequency) method is used for calculation:

[0174] One or more historical click display information and click time points corresponding to the reference long-term tag can be obtained from the historical behavior data within the fourth preset time period. Then, a corresponding score can be assigned based on the distance between the click time point corresponding to each historical click display information and the end time point of the fourth preset time period. The closer the click time point is to the end time point of the fourth preset time period, the higher the score corresponding to the click time point. Specifically, the score corresponding to each click time point is:

[0175]

[0176] Wherein, D1 is the time length from the click time point to the start time point of the fourth preset time period, and D2 is the total time length of the fourth preset time period.

[0177] Through the above method, the score corresponding to the click time point of one or more historical click display information corresponding to each reference long-term tag can be determined, and the TF corresponding to the reference long-term tag 点击 It is the sum of the scores corresponding to one or more click time points corresponding to the reference long-term label.

[0178] Then, the IDF corresponding to each reference long-term label can be calculated 点击 , the corresponding calculation formula is:

[0179]

[0180] Wherein, ω1 is the total number of users of the application program, and ω2 is the number of users whose historical click display information corresponding to the long-term click tag is included in the historical click data within the fourth preset time period.

[0181] Calculate the TF corresponding to each reference long-term label using the above method 点击 and IDF 点击 After that, you can multiply the two together to get the click correlation between each reference long-term tag and the target account. Alternatively, you can multiply the TF corresponding to the reference long-term tag by 点击 and IDF 点击 After multiplication, add Multiply, or you can multiplied by the first preset power of , thereby obtaining the click correlation between the reference long-term tag and the target account.

[0182] Similarly, the TFIDF method mentioned above can also be used to calculate the collection relevance between each reference long-term tag and the target account.

[0183] One or more historical collection display information and the collection time points corresponding to the historical collection display information corresponding to the reference long-term tag can be obtained from the historical behavior data within the fourth preset time period. Then, a corresponding score can be assigned based on the distance between the collection time point corresponding to each historical collection display information and the end time point of the fourth preset time period. The closer the collection time point is to the end time point of the fourth preset time period, the higher the score corresponding to the collection time point. Specifically, the score corresponding to each collection time point is:

[0184]

[0185] Among them, D3 is the time length from the collection time point to the start time point of the fourth preset time period, and D4 is the total time length of the fourth preset time period.

[0186] Through the above method, the score corresponding to the collection time point of one or more historical collection display information corresponding to each reference long-term tag can be determined, and the TF corresponding to the reference long-term tag 收藏 It is the sum of the scores corresponding to one or more collection time points corresponding to the reference long-term tag.

[0187] Then, the IDF corresponding to each reference long-term label can be calculated 收藏 , the corresponding calculation formula is:

[0188]

[0189] Wherein, ω3 is the total number of users of the application program, and ω4 is the number of users whose historical collection data within the fourth preset time period includes historical collection display information corresponding to the reference long-term tag.

[0190] Calculate the TF corresponding to each reference long-term label using the above method 收藏 and IDF 收藏 After that, you can multiply the two together to get the collection relevance between each reference long-term tag and the target account. Alternatively, you can multiply the TF corresponding to the reference long-term tag by 收藏 and IDF 收藏 After multiplication, add Multiply, or you can Multiply by the second preset power of to obtain the collection correlation between the reference long-term tag and the target account.

[0191] Then, for each reference long-term tag, the click relevance and favorite relevance corresponding to the reference long-term tag can be added together to determine the relevance between the reference long-term tag and the target account. Alternatively, weights can be assigned to the click relevance and favorite relevance based on their respective priorities, and the resulting value, multiplied by the corresponding weights, is used to determine the relevance between the reference long-term tag and the target account.

[0192] 304. Select a third preset number of reference long-term tags in descending order of relevance to obtain a long-term tag corresponding to the target account.

[0193] In implementation, multiple reference long-term tags are arranged in descending order according to their corresponding relevance, and the first third preset number of reference long-term tags are determined as long-term tags corresponding to the target account.

[0194] The third preset number may be any reasonable number, and is not limited in this embodiment of the present application.

[0195] Optionally, in addition to obtaining the short-term and long-term tags corresponding to the target account, you can also obtain tags that are similar to the long-term tags. The corresponding processing is as follows:

[0196] For each long-term tag, calculate the similarity between each tag and the long-term tag. Based on the similarity between each tag and the long-term tag, determine the extended tag corresponding to each long-term tag.

[0197] In implementation, the following processing can be performed for each long-term tag: the similarity between the long-term tag and each tag is calculated. The tags can then be sorted in descending order of similarity, and the first five predetermined numbers of tags can be determined as the extended tags corresponding to the long-term tag. Alternatively, tags with similarities greater than a predetermined similarity threshold can be determined as the extended tags corresponding to the long-term tag.

[0198] The method for determining the similarity between the long-term label and each label can be: use a feature extraction model to extract features from each label to obtain a feature vector corresponding to each label, and calculate the similarity through the feature vector corresponding to the long-term label and the feature vector corresponding to each other label.

[0199] Among them, the feature extraction model can be a word2vec (word to vector, word embedding) model, for example, it can be a CBOW (Continuous Bag-Of-Words, continuous bag of words model) model, etc., which is not limited in the embodiment of the present application.

[0200] After determining the short-term, long-term, and extended tags corresponding to the target account, a reference tag corresponding to the target account can be determined based on these three types of tags. In practice, duplicates can be removed from the short-term, long-term, and extended tags corresponding to the target account to obtain the reference tag corresponding to the target account.

[0201] All of the above optional technical solutions can be combined in any way to form optional embodiments of the present application, and will not be described in detail here.

[0202] The solution mentioned in the embodiment of the present application can screen multiple tags based on the statistical data corresponding to the multiple tags, thereby obtaining multiple high-quality, recommendable recommendable tags, and then based on the historical behavior data of the target account, determine the multiple reference tags that the target account corresponds to that the user is more interested in. Then, the intersection between the multiple recommendable tags and the multiple reference tags is determined as the target recommendation tag corresponding to the target account, and finally the target display information corresponding to each target recommendation tag is obtained, and the target display information is sent to the target terminal to which the target account belongs. By using this application, it is possible to determine target recommendation tags that are both high-quality and recommendable and that users are interested in, thereby improving the recommendation effect.

[0203] The embodiment of the present application provides a device for sending display information, which may be the computer device in the above embodiment, such as Figure 4 As shown, the device includes:

[0204] A screening module 410 is configured to screen the multiple tags based on statistical data corresponding to the multiple tags to obtain multiple recommended tags, wherein the tags are information indicating content characteristics of the displayed information;

[0205] A first determining module 420 is configured to determine a plurality of reference tags corresponding to a target account based on historical behavior data of the target account;

[0206] A second determining module 430 is configured to determine an intersection of the plurality of recommendable tags and the plurality of reference tags as a target recommended tag corresponding to the target account;

[0207] The sending module 440 is configured to obtain target display information corresponding to each target recommendation tag, and send the target display information to the target terminal to which the target account belongs.

[0208] In a possible implementation, the apparatus further includes a label determination module configured to:

[0209] For each piece of display information, a tag corresponding to the display information is determined based on the display information and a keyword extraction model.

[0210] In a possible implementation, the screening module 410 is configured to:

[0211] Determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the multiple tags;

[0212] Based on the user interest intensity corresponding to each tag and a preset intensity threshold, the multiple tags are screened to obtain multiple recommendable tags.

[0213] In a possible implementation, the statistical data includes click statistical data and favorite statistical data.

[0214] In a possible implementation, the click statistics include a first click rate corresponding to all users within a first preset time period and a second click rate corresponding to historical click users within the first preset time period, wherein the historical click users are users who have clicked on any display information corresponding to the tag within a second preset time period before the first preset time period and have been exposed to the display information corresponding to the tag within the first preset time period;

[0215] The collection statistics include a first collection rate corresponding to all users in the first preset period and a second collection rate corresponding to historical collection users in the first preset period, wherein the historical collection users are users who have collected any display information corresponding to the tag in the second preset period and have been exposed to the display information corresponding to the tag in the first preset period;

[0216] The screening module 410 is used to:

[0217] For each tag, determining the user click interest intensity corresponding to the tag based on the first click rate and the second click rate corresponding to the tag;

[0218] Determining the user's collection interest intensity corresponding to the tag based on the first collection rate and the second collection rate corresponding to the tag;

[0219] The user interest intensity corresponding to the tag is determined based on the user click interest intensity and the user collection interest intensity corresponding to the tag.

[0220] In a possible implementation, the first determining module 420 is configured to:

[0221] Determining a short-term tag corresponding to the target account based on the historical behavior data of the target account within a third preset time period, wherein the third preset time period is a recent time period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information is less than a preset duration;

[0222] determining a long-term tag corresponding to the target account based on historical behavior data of the target account within a fourth preset time period;

[0223] Based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account is determined.

[0224] In one possible implementation, the historical behavior data includes historical click data and historical collection data, the historical click data includes one or more historical click display information and the click time point corresponding to each historical click display information, and the historical collection data includes one or more historical collection display information and the collection time point corresponding to each historical collection display information.

[0225] In a possible implementation, the first determining module 420 is configured to:

[0226] Obtaining short-term click tags corresponding to historical click data of the target account within a third preset time period and short-term collection tags corresponding to historical collection data of the target account within a third preset time period;

[0227] Determine the most recent click time point corresponding to each short-term click tag among the click time points corresponding to one or more historical click display information corresponding to each short-term click tag;

[0228] Selecting a first preset number of short-term click tags in descending order of the most recent click time points to obtain a target short-term click tag corresponding to the target account;

[0229] Determine the most recent collection time point corresponding to each short-term collection tag among the collection time points corresponding to one or more historical collection display information corresponding to each short-term collection tag;

[0230] Selecting a second preset number of short-term favorite tags in descending order of the most recent favorite time point to obtain a target short-term favorite tag corresponding to the target account;

[0231] Deduplication processing is performed on the target short-term click tag and the target short-term collection tag to obtain a short-term tag corresponding to the target account.

[0232] In a possible implementation, the first determining module 420 is configured to:

[0233] Obtaining a long-term click tag corresponding to the historical click data of the target account within a fourth preset time period and a long-term collection tag corresponding to the historical collection data of the target account within the fourth preset time period;

[0234] Deduplication is performed on the long-term click tag and the long-term favorite tag to obtain a reference long-term tag corresponding to the target account;

[0235] Determining, based on historical behavior data within a fourth preset time period corresponding to each reference long-term tag, a correlation between each reference long-term tag and the target account;

[0236] A third preset number of reference long-term tags are selected in descending order of the relevance to obtain a long-term tag corresponding to the target account.

[0237] In a possible implementation, after determining the long-term tag corresponding to the target account, the first determining module 420 is further configured to:

[0238] Get the feature vector corresponding to each label;

[0239] For each long-term tag, respectively calculate the similarity between each tag and the long-term tag;

[0240] Determining an extended tag corresponding to each long-term tag based on a similarity between each tag and the long-term tag;

[0241] The determining, based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account includes:

[0242] A reference tag corresponding to the target account is determined based on the short-term tag corresponding to the target account, the long-term tag corresponding to the target account, and the extended tag corresponding to each long-term tag.

[0243] In a possible implementation, the sending module 440 is configured to:

[0244] For each target recommendation tag, based on statistical data of each display information corresponding to the target recommendation tag, determine a score for each display information corresponding to the target recommendation tag;

[0245] A fourth preset number of display information is selected in descending order of scores to obtain target display information corresponding to the target recommendation tag.

[0246] It should be noted that the apparatus for transmitting display information provided in the above embodiments uses the division of the aforementioned functional modules as an example only. In actual applications, the aforementioned functions can be assigned to different functional modules as needed, that is, the internal structure of the apparatus can be divided into different functional modules to perform all or part of the functions described above. Furthermore, the apparatus for transmitting display information provided in the above embodiments and the method for transmitting display information are conceptually identical. The specific implementation process is detailed in the method embodiments and will not be further elaborated here.

[0247] Figure 5 : This is a schematic diagram of the structure of a server provided in an embodiment of the present application. The server 500 may have relatively large differences due to different configurations or performances, and may include one or more processors (Central Processing Units, CPU) 501 and one or more memories 502, wherein the memory 502 stores at least one instruction, and the at least one instruction is loaded and executed by the processor 501 to implement the methods provided in the above-mentioned various method embodiments. Of course, the server may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output. The server may also include other components for implementing device functions, which will not be described in detail here.

[0248] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions. The instructions are executable by a processor in a terminal to implement the method for sending presentation information in the above-described embodiment. The computer-readable storage medium may be non-transitory. For example, the computer-readable storage medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0249] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0250] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals (including but not limited to signals transmitted between user terminals and other devices, etc.) involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions. For example, the "statistical data" involved in this application were obtained with full authorization.

[0251] The above description is merely an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for sending display information, characterized in that: The method comprises: Based on statistical data corresponding to the multiple tags, the multiple tags are screened to obtain multiple recommended tags, wherein the tags are information used to indicate content characteristics of the displayed information; Determining a plurality of reference tags corresponding to the target account based on historical behavior data of the target account; Determining an intersection of the multiple recommendable tags and the multiple reference tags as a target recommended tag corresponding to the target account; Obtaining target display information corresponding to each target recommendation tag, and sending the target display information to the target terminal to which the target account belongs; filtering the multiple tags based on the statistical data corresponding to the multiple tags to obtain multiple recommendable tags, including: Determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the multiple tags; Based on the user interest intensity corresponding to each tag and a preset intensity threshold, the multiple tags are screened to obtain multiple recommendable tags; the statistical data includes click statistics and favorite statistics; The click statistics include a first click rate corresponding to all users in a first preset period and a second click rate corresponding to historical click users in the first preset period, wherein the historical click users are users who have clicked on any display information corresponding to the tag in a second preset period before the first preset period and have been exposed to the display information corresponding to the tag in the first preset period; The collection statistics include a first collection rate corresponding to all users in the first preset period and a second collection rate corresponding to historical collection users in the first preset period, wherein the historical collection users are users who have collected any display information corresponding to the tag in the second preset period and have been exposed to the display information corresponding to the tag in the first preset period; The determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the plurality of tags includes: for each tag, determining the user click interest intensity corresponding to the tag based on a first click rate and a second click rate corresponding to the tag; The user collection interest strength corresponding to the tag is determined based on the first collection rate and the second collection rate corresponding to the tag; and the user interest strength corresponding to the tag is determined based on the user click interest strength and the user collection interest strength corresponding to the tag.

2. The method according to claim 1, characterized in that The determining of a plurality of reference tags corresponding to the target account based on the historical behavior data of the target account includes: Determining a short-term tag corresponding to the target account based on the historical behavior data of the target account within a third preset time period, wherein the third preset time period is a recent time period in which multiple user behavior information exists, and the time difference between each two adjacent user behavior information is less than a preset duration; Based on the historical behavior data of the target account within a fourth preset time period, a long-term tag corresponding to the target account is determined; based on the short-term tag corresponding to the target account and the long-term tag corresponding to the target account, a reference tag corresponding to the target account is determined.

3. The method according to claim 2, characterized in that The historical behavior data includes historical click data and historical collection data. The historical click data includes one or more historical click display information and the click time point corresponding to each historical click display information. The historical collection data includes one or more historical collection display information and the collection time point corresponding to each historical collection display information.

4. A device for sending display information, characterized in that: The device comprises: A screening module is configured to screen the multiple tags based on statistical data corresponding to the multiple tags to obtain multiple recommended tags, wherein the screening module includes: Determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the multiple tags; Based on the user interest intensity corresponding to each tag and a preset intensity threshold, the multiple tags are screened to obtain multiple recommendable tags, and the statistical data include click statistics and collection statistics, wherein the tag is information used to indicate the content characteristics of the display information, and the click statistics include a first click rate corresponding to all users in a first preset time period and a second click rate corresponding to historical click users in the first preset time period, wherein the historical click users are users who have clicked on any display information corresponding to the tag in a second preset time period before the first preset time period and have been exposed to the display information corresponding to the tag in the first preset time period; the collection statistics include a first collection rate corresponding to all users in the first preset time period and a second collection rate corresponding to historical collection users in the first preset time period, wherein the historical collection users are users who have collected any display information corresponding to the tag in the second preset time period and have been exposed to the display information corresponding to the tag in the first preset time period; The determining the user interest intensity corresponding to each tag based on the statistical data corresponding to the plurality of tags includes: for each tag, determining the user click interest intensity corresponding to the tag based on a first click rate and a second click rate corresponding to the tag; Determine the user's collection interest strength corresponding to the tag based on the first collection rate and the second collection rate corresponding to the tag; determine the user's interest strength corresponding to the tag based on the user's click interest strength and the user's collection interest strength corresponding to the tag; A first determining module, configured to determine a plurality of reference tags corresponding to the target account based on historical behavior data of the target account; A second determining module is configured to determine an intersection of the plurality of recommendable tags and the plurality of reference tags as a target recommended tag corresponding to the target account; The sending module is used to obtain target display information corresponding to each target recommendation tag and send the target display information to the target terminal to which the target account belongs.

5. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the operation performed by the method for sending presentation information according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that The storage medium stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the operation performed by the method for sending display information according to any one of claims 1 to 3.

7. A computer program product, characterized in that The computer program product includes at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the operations performed by the method for sending display information according to any one of claims 1 to 3.

Images