Content recommendation method, device, electronic device and storage medium

Abstract

The embodiments of the present invention provide a content recommendation method, device, electronic device and storage medium. The content recommendation method provided by the embodiments of the present invention obtains at least one target recommended content corresponding to the multiple target behavior data of the target user from a preset index library based on multiple target behavior data obtained within a preset time period of the target user. The preset index library stores multiple correspondences between multiple behavior data obtained based on the historical behavior data of each user and a recommended content, and then recommends the obtained target recommended content to the target user. In the embodiments of the present invention, the index library is preset based on the historical behavior data of each user, so that the preferences of each user for content are taken into account when recommending, thereby improving the accuracy of content recommendation. At the same time, the recommended content is obtained through the multiple behavior data of the target user, and the accuracy of content recommendation is higher than that of recommending content based only on the current single behavior data of the target user.

Description

Technical Field

[0001] The present invention relates to the field of intelligent recommendation technology, and in particular to a content recommendation method, device, electronic device and storage medium. Background Art

[0002] With the increasing development of intelligent technology, especially the richness of data collection methods, many business applications can make personalized recommendations for different users. For example, video applications recommend videos that users may like, shopping applications recommend products that users may like, etc. Personalized recommendations, that is, recommending objects of interest to users based on their interests, can help users quickly discover valuable information.

[0003] Among existing personalized recommendation technologies, a common method is to recommend content based on the user's current behavior, that is, to recommend content related to the user's current behavior. This method is prone to low recommendation quality and high noise. For example, in a shopping app, if a user is currently browsing a shirt of a certain brand, similar shirts of the brand will usually be recommended to the user, but the user may actually want to browse other styles of shirts from other brands. It can be seen that the content recommended to the user in this way may not be of interest to the user, that is, the accuracy of the recommendation is low. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide a content recommendation method, device, electronic device and storage medium to improve the accuracy of recommendation. The specific technical solution is as follows:

[0005] In a first aspect of the present invention, a content recommendation method is provided, the method comprising:

[0006] Acquire multiple target behavior data within a preset time period of the target user; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior;

[0007] Based on multiple target behavior data within a preset time period of the target user, at least one target recommended content corresponding to the behavior data of the target user is obtained from a preset index library; the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content;

[0008] The at least one target recommended content is recommended to a target user.

[0009] In one embodiment of the present invention, the preset index library stores a many-to-one correspondence between behavior data obtained based on historical behavior data of each user and recommended content, specifically: a first preset number of behavior data corresponds to one recommended content;

[0010] The preset index library is preset using the following process:

[0011] Obtain historical behavior data of each user;

[0012] For each user, the historical behavior data of the user is segmented according to a preset first segmentation quantity to obtain a plurality of historical behavior data combinations, wherein each of the historical behavior data combinations contains behavior data of a preset first segmentation quantity; the preset first segmentation quantity is greater than the first preset quantity;

[0013] For each historical behavior data combination, respectively use the operation content in each historical behavior data in the historical behavior data combination as the recommended content, obtain the corresponding relationship between the recommended content and a first preset number of other behavior data in the historical behavior data combination, and use the determined corresponding relationship as the first index information;

[0014] The obtained first index information is stored in the preset index library.

[0015] In one embodiment of the present invention, the step of storing the obtained first index information in the preset index library includes:

[0016] Among all the first index information, the first index information in which the number of repetitions of the correspondence between the first preset number of behavior data and one recommended content exceeds a preset number threshold is stored in the preset index library.

[0017] In one embodiment of the present invention, the step of acquiring at least one target recommended content corresponding to the target user's behavior data from a preset index library based on multiple target behavior data of the target user within a preset time period includes:

[0018] Dividing the plurality of target behavior data within the preset time period of the target user according to a preset second segmentation number to obtain a plurality of target segmentation behavior data combinations; wherein the preset second segmentation number is equal to the first preset number;

[0019] Searching for the target segmentation behavior data combination in the preset index library;

[0020] If the target segmentation behavior data combination is found in the preset index library, the recommended content corresponding to the target segmentation behavior data combination is used as the target recommended content.

[0021] In one embodiment of the present invention, the preset index library stores a plurality of many-to-one correspondences between behavior data obtained based on historical behavior data of each user and recommended content; wherein each many-to-one correspondence corresponds to a preset third number of cuts, and the preset third number of cuts is greater than the second preset number of behavior data in the correspondence;

[0022] The preset index library is preset using the following process:

[0023] Obtain historical behavior data of each user;

[0024] For each user, the historical behavior data of the user is divided according to each preset third division quantity, and a plurality of historical behavior data combinations corresponding to each many-to-one correspondence relationship are obtained respectively;

[0025] For each historical behavior data combination, based on the corresponding target many-to-one correspondence relationship, the operation content in each historical behavior data in the historical behavior data combination is respectively used as the recommended content, and the correspondence relationship between the recommended content and the target number of other behavior data in the historical behavior data combination is obtained, and the obtained correspondence relationship is used as the second index information; wherein the target number is: the second preset number of behavior data in the target many-to-one correspondence relationship;

[0026] The obtained second index information is stored in the preset index library.

[0027] In one embodiment of the present invention, each of the above-mentioned many-to-one correspondences corresponds to a preset fourth segmentation quantity, and the preset fourth segmentation quantity is equal to the second preset quantity of behavior data in the correspondence;

[0028] The step of acquiring at least one target recommended content corresponding to the target user's behavior data from a preset index library based on multiple target behavior data within a preset time period of the target user includes:

[0029] According to each of the preset fourth segmentation quantities, the multiple target behavior data within the preset time period of the target user are segmented respectively, and multiple target segmentation behavior data combinations corresponding to each many-to-one correspondence relationship are obtained respectively;

[0030] Searching for the target segmentation behavior data combination in the preset index library;

[0031] If the target segmentation behavior data combination is found in the preset index library, the recommended content corresponding to the target segmentation behavior data combination is used as the target recommended content.

[0032] In one embodiment of the present invention, the step of recommending the at least one target recommended content to the target user includes:

[0033] Sort the at least one target recommended content according to the time sequence of the target segmentation behavior data combination corresponding to the target recommended content in the multiple target behavior data of the target user;

[0034] The target recommended content is displayed to the target user according to the ranking.

[0035] In a second aspect of the present invention, a content recommendation device is provided, the device comprising:

[0036] The behavior data acquisition module is used to acquire multiple target behavior data within a preset time period of the target user; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior;

[0037] A recommended content acquisition module, configured to acquire at least one target recommended content corresponding to the behavior data of the target user from a preset index library based on multiple target behavior data within a preset time period of the target user; the preset index library stores at least one many-to-one correspondence between the behavior data acquired based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content;

[0038] The recommended content recommendation module is used to recommend the at least one target recommended content to a target user.

[0039] In another aspect of the present invention, there is provided an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0040] Memory, used to store computer programs;

[0041] The processor is used to implement any of the above-mentioned content recommendation method steps when executing the program stored in the memory.

[0042] In another aspect of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any of the above-mentioned content recommendation methods is implemented.

[0043] In yet another aspect of the present invention, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute any of the above-mentioned content recommendation methods.

[0044] The content recommendation method provided in the embodiment of the present invention first obtains multiple target behavior data of the target user within a preset time period, and based on the multiple target behavior data, obtains at least one target recommended content corresponding to the multiple target behavior data of the target user from a preset index library, and the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, specifically, multiple behavior data corresponds to one recommended content, and then the obtained target recommended content is recommended to the target user. In the embodiment of the present invention, the index library is preset based on the historical behavior data of each user, so that the recommended content is not only based on the similarity of the content itself, but also takes into account the user's preference for the content, thereby improving the accuracy of content recommendation. At the same time, the recommended content is obtained through the multiple behavior data of the target user, and the accuracy of content recommendation is higher than that of recommending content based only on the current single behavior data of the target user. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0046] Figure 1 A flow chart of a content recommendation method provided in an embodiment of the present invention;

[0047] Figure 2 A flow chart of the index library preset process provided in an embodiment of the present invention;

[0048] Figure 3 A second flow chart of the content recommendation method provided in an embodiment of the present invention;

[0049] Figure 4 A schematic diagram of a specific implementation flow of the content recommendation method provided in an embodiment of the present invention;

[0050] Figure 5 A second flow chart of the index library preset process provided in an embodiment of the present invention;

[0051] Figure 6 A third flow chart of the content recommendation method provided in an embodiment of the present invention;

[0052] Figure 7 A fourth flow chart of the content recommendation method provided in an embodiment of the present invention;

[0053] Figure 8 A schematic diagram of the structure of a content recommendation device provided in an embodiment of the present invention;

[0054] Fig. 9 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present invention will be described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0056] In order to improve the accuracy of content recommendation, embodiments of the present invention provide a content recommendation method, device, electronic device and storage medium.

[0057] The following first introduces a content recommendation method provided by the present invention.

[0058] like Figure 1 As shown, Figure 1 This is a flow chart of a content recommendation method provided in an embodiment of the present invention, and the specific steps may include:

[0059] Step 101, obtaining multiple target behavior data of a target user within a preset time period; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior;

[0060] Step 102, based on the target behavior data of the target user within a preset time period, obtaining at least one target recommended content corresponding to the behavior data of the target user from a preset index library; the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content;

[0061] Step 103: recommend the at least one target recommended content to a target user.

[0062] The content recommendation method provided in the embodiment of the present invention first obtains multiple target behavior data of the target user within a preset time period, and based on the multiple target behavior data, obtains at least one target recommended content corresponding to the multiple target behavior data of the target user from a preset index library, and the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, specifically, multiple behavior data corresponds to one recommended content, and then the obtained target recommended content is recommended to the target user. In the embodiment of the present invention, the index library is preset based on the historical behavior data of each user, so that the recommended content is not only based on the similarity of the content itself, but also takes into account the user's preference for the content, thereby improving the accuracy of content recommendation. At the same time, the recommended content is obtained through the multiple behavior data of the target user, and the accuracy of content recommendation is higher than that of recommending content based only on the current single behavior data of the target user.

[0063] In an embodiment of the present invention, the acquired user behavior data may include all operation behaviors generated by the user in the application and the operation content corresponding to the operation behavior. For example, in a video application, the acquired target user's behavior data may include search, viewing, collection and other behaviors, and correspondingly, the operation content corresponding to the operation behavior may be the search term, the video information viewed, the video information collected, etc. In a shopping application, the acquired target user's behavior data may include search, browsing, add to cart, purchase and other behaviors, and the operation content corresponding to the above operation behaviors may be the search term and the product information browsed, added to cart, and purchased, etc.

[0064] As a specific implementation of the embodiment of the present invention, when obtaining the behavior data of the target user, the behavior data of the target user within a certain time period can be obtained. The time period can be preset manually. At the same time, in order to obtain more behavior data of the target user, the time period can be set longer, such as two months, three months, four months, etc., that is, all the behavior data of the target user from the current time node to the previous few months can be obtained.

[0065] After obtaining the above-mentioned multiple behavior data of the target user, the target recommended content recommended to the target user can be obtained from the preset index library based on the multiple behavior data.

[0066] In the embodiment of the present invention, the preset index library may store only one many-to-one correspondence, for example, only storing the correspondence between 2 behavior data and one recommended content, or only storing the correspondence between 3 behavior data and one recommended content, and so on.

[0067] Of course, the preset index library can also store a variety of many-to-one correspondences, for example: not only the correspondence between 2 behavior data and one recommended content is stored, but also the correspondence between 3 behavior data and one recommended content is stored, and so on.

[0068] Each many-to-one correspondence stored in the above-mentioned preset index library contains multiple many-to-one relationships. For example: for this correspondence between 2 behavioral data and one recommended content, that is, a 2-to-1 correspondence, it can specifically include: a 2-to-1 correspondence between the behavioral data watching video A and watching video B and the recommended content video C; a 2-to-1 correspondence between watching video B and watching video C and the recommended content video D, and so on. Multiple 2-to-1 correspondences. Similarly, for this correspondence between 3 behavioral data and one recommended content, that is, a 3-to-1 correspondence, it can specifically include: a 3-to-1 correspondence between watching video A, watching video B, watching video C and video D; a 3-to-1 correspondence between purchasing product 1, purchasing product 2, and purchasing product 3 and recommended product 4, and so on. Multiple 3-to-1 correspondences.

[0069] In the case where only one many-to-one correspondence relationship can be stored in the preset index library, the number of behavior data corresponding to the above-mentioned one recommended content can be manually preset, and for the convenience of description, the number can be recorded as the first preset number. That is, the above-mentioned preset index library can store the relationship between the first preset number of behavior data and one recommended content.

[0070] As described above, as a specific implementation of an embodiment of the present invention, the preset index library may store a plurality of many-to-one correspondences of a many-to-one correspondence, where the many-to-one relationship may be that a first preset number of behavior data corresponds to one recommended content. Accordingly, as a specific implementation of the present invention, Figure 2 As shown, Figure 2 A flowchart of the index library preset process provided in an embodiment of the present invention may specifically include the following steps:

[0071] Step 201, obtaining historical behavior data of each user;

[0072] As a specific implementation, when obtaining the historical behavior data of each user, it is also possible to obtain the behavior data of each user within a period of time. The time period can also be set in advance by the developer. For example, the time period can be set to one week, that is, all the historical behavior data of each user within one week is obtained. Similar to the behavior data of the target user obtained, the historical behavior data of each user also includes the operation behavior and the operation content corresponding to the operation behavior. For an application (APP) that recommends content, the above-mentioned users can be all users of the APP.

[0073] Step 202, for each user, the historical behavior data of the user is segmented according to a preset first segmentation number to obtain a plurality of historical behavior data combinations, wherein each of the above historical behavior data combinations contains a preset first segmentation number of behavior data; the above preset first segmentation number is greater than the above first preset number;

[0074] As described above, the many-to-one relationship stored in the preset index library is that the first preset number of behavior data corresponds to one recommended content. Therefore, after obtaining the historical behavior data of each user, the number of splits can be set according to the number of multiple behavior data in the corresponding relationship to be constructed, and the historical behavior data of each user can be split based on the number of splits, and each combination of historical behavior data obtained will contain the above-mentioned number of behavior data. The splitting can be performed in the order of the user's historical behavior data, or it can be performed in a non-ordered manner, which is not specifically limited here.

[0075] For example, in a specific embodiment of a video application, the index library to be constructed stores the correspondence between 2 behavior data and one recommended content, that is, in this example, the first preset number is 2. Correspondingly, since 2 behavior data correspond to 1 recommended content, the number of divisions can be set to 3, that is, for each user's historical behavior data, the historical behavior data of the user is divided into groups of 3 behavior data, and after the division, multiple groups of historical behavior data combinations can be obtained, each of which contains 3 behavior data.

[0076] Based on the above example, the historical behavior data of a certain user obtained is watching video A, searching B (B is the search term), watching video C, and collecting video D. According to the order of the historical behavior data of the user, and dividing the historical behavior of the user into groups of 3 behavior data, the following multiple groups of historical behavior data combinations can be obtained: watching video A, searching B, watching video C as a group; searching B, watching video C, collecting video D as a group; watching video A, watching video C, collecting video D as a group; watching video A, searching B, collecting video D as a group.

[0077] Step 203, for each historical behavior data combination, respectively use the operation content in each historical behavior data in the historical behavior data combination as recommended content, obtain the corresponding relationship between the recommended content and a first preset number of other behavior data in the historical behavior data combination, and use the determined corresponding relationship as first index information;

[0078] After the above-mentioned multiple groups of historical behavior data combinations are obtained, the above-mentioned many-to-one correspondence relationship can be established based on the behavior data in each historical behavior data combination.

[0079] For the example in step 202 above, the index library to be constructed stores the correspondence between two behavior data and one recommended content. In each combination of historical behavior data obtained, each combination of historical behavior data contains three behavior data. Therefore, in this example, for each combination of historical behavior data, the operation content of one of the behavior data can be selected as the recommended content, and the correspondence between it and the other two behavior data can be established.

[0080] For example, for the above historical behavior data combination of watching video A, searching B, and watching video C, video C can be selected as the recommended content, so that a correspondence relationship between video C and the behavior data of watching video A and searching B can be established. Search word B can also be selected as the recommended content, so that a correspondence relationship between search word B and the behavior data of watching video A and watching video C can be established. In the embodiment of the present invention, for each historical behavior data combination, the operation content in each behavior data therein can be used as the recommended content, and a correspondence relationship between the recommended content and other behavior data in the historical behavior data combination can be established.

[0081] Each many-to-one relationship obtained for each historical behavior combination may be recorded as first index information.

[0082] See the following table, which shows part of the above first index information:

[0083]

[0084] As shown in the above table, in the embodiment of the present invention, the same first preset number of behavior data can correspond to different recommended contents, such as the behavior data of watching video A and watching video B can correspond to the recommended content video C; the behavior data of watching video A and watching video B can also correspond to the recommended content video D. In addition, in the embodiment of the present invention, different first preset numbers of behavior data can correspond to the same recommended content, such as the behavior data of watching video A and watching video B shown in the above table can correspond to the recommended content video C; the behavior data of watching video E and watching video F can also correspond to the recommended content video C.

[0085] Step 204: store the obtained first index information into the preset index library.

[0086] As a specific implementation method in the embodiment of the present invention, a number threshold can be set, and the above-mentioned first index information whose number of repetitions of the correspondence between a first preset number of behavior data and a recommended content exceeds the number threshold is stored in a preset index library. The number threshold can be set by the developer. For example, the number threshold can be 5 times, 10 times, etc. Of course, the threshold can also be other numbers, which is not specifically limited here.

[0087] For example, the number threshold is set to 5. For the following first index information: watch video A, search B corresponding to video C; watch video A, watch video C corresponding to video D; search E, search F corresponding to video D, the first index information "watch video A, search B corresponding to video C" is repeated 8 times, the first index information "watch video A, watch video C corresponding to video D" is repeated 6 times, and the first index information "search E, search F corresponding to video" is repeated only once. Then, the first index information storage value with a number of repetitions exceeding the number threshold of 5 can be preset in the index library. For the first index information "search E, search F corresponding to video D" with a number of repetitions below the number threshold, it can be determined to be a specific search of a certain user. For example, the above-mentioned search terms E and F may be a separate search of a certain user containing a lot of noise, that is, only this user has the first index information. Since it is not universal, the first index information can be not stored in the preset index library.

[0088] By setting the number threshold, the first index information that is not universal can be filtered, thereby further improving the accuracy of content recommendation and saving a lot of storage space.

[0089] Accordingly, based on Figure 1 ,like Figure 3 As shown, Figure 3 A second flow chart of the content recommendation method provided by an embodiment of the present invention, Figure 1 Step 102 in the above method, based on the target behavior data of the target user within a preset time period, obtains at least one target recommendation content corresponding to the behavior data of the target user from a preset index library, which may specifically include the following steps:

[0090] Step 302, dividing the plurality of target behavior data of the target user within the preset time period according to a preset second division number to obtain a plurality of target division behavior data combinations, wherein the preset second division number is equal to the first preset number;

[0091] As described above, the preset index library stores the correspondence between a plurality of first preset number of behavior data and a recommended content, so after obtaining the plurality of behavior data of the target user, the plurality of behavior data of the target user can be segmented according to the preset segmentation length to obtain the target segmented behavior data combination. The segmentation length can be equal to the first preset number.

[0092] For example, for the above example, the preset index library stores a plurality of correspondences between two behavior data and one recommended content, that is, the first preset number is 2. When segmenting the behavior data of the target user, the behavior data of the target user can be segmented according to the segmentation number of 2, that is, each group of the obtained target segmented behavior data combination contains 2 behavior data.

[0093] Step 303, searching the above-mentioned target segmentation behavior data combination in the above-mentioned preset index library;

[0094] Based on the example in step 302 above, the preset index library stores a plurality of correspondences between two behavior data and one recommended content based on the historical behavior data of each user, and after obtaining the plurality of behavior data of the target user, the behavior data of the target user is segmented according to the segmentation length of two behavior data in each target segmented behavior data combination, and a plurality of target segmented behavior data combinations are obtained. Therefore, when searching for the target segmented behavior data combination in the preset index library, the target segmented behavior data combination can be matched with a plurality of first preset number of behavior data in the preset index library.

[0095] Step 304, determining whether the target segmentation behavior data combination is found in the preset index library, if found, executing step 305, if not found, executing step 306;

[0096] Step 305, taking the recommended content corresponding to the target segmentation behavior data combination as the target recommended content;

[0097] Step 306, combining recommended content without targeting the above target segmentation behavior data.

[0098] Based on the example of step 303, for a target segmentation behavior data combination, if one of the two behavior data groups is matched in the preset index library, the recommended content corresponding to the two behavior data groups can be used as the recommended content corresponding to the target segmentation behavior data combination. If no one of the two behavior data groups in the preset index library matches the target segmentation behavior data combination, that is, the target segmentation behavior data combination is not found in the preset index library, then no content is recommended for the target segmentation behavior data combination, that is, in this case, the recommended content is empty.

[0099] For example, the preset index library stores three correspondences between two behavior data and one recommended content, namely, watching video A, searching B and video C correspond; watching video A, watching video C correspond to search term B; searching B, watching video C correspond to video A. A target segmentation behavior data combination is watching video A and watching video C. Then, a behavior data group matching the target segmentation behavior data combination can be found in the above preset index library, so the recommended content search term B corresponding to the behavior data group can be recommended to the user. Specifically, the search term can be displayed in the search box or in the "Recommended Search" below the search box, which is not specifically limited here. As for the target segmentation behavior data combination watching video D and watching video C, a behavior data group matching the target segmentation behavior data combination cannot be found in the above preset index library, so it is impossible to recommend content for the target segmentation behavior data combination.

[0100] like Figure 4 As shown, Figure 4 This is a schematic diagram of a specific implementation process of a content recommendation method in an embodiment of the present invention:

[0101] First, collect the behavior sequence of each user offline, that is, historical behavior data. Figure 4 In the example, the behavior sequence of each user collected offline is item1, item2, item3, item1, item2, item3. The above items are the historical behavior data of the user. For video applications, the above items can be watching video A, collecting video B, searching for C, etc.; for shopping applications, the above items can be buying product X, searching for Y, adding product Z, etc. Figure 4 As shown in FIG. 1 , after obtaining the behavior sequences of the above-mentioned users, the above-mentioned historical behavior data can be segmented according to the segmentation number of 3 to obtain historical behavior data combinations, and a corresponding relationship between two behavior data and one recommended content is constructed for each historical behavior data combination (that is, n is 2 in the constructed ngram index). Figure 4For the user behavior sequence in, the 2-to-1 correspondence relationship can be item1, item2 corresponds to item3; item2, item3 corresponds to item1; item3, item1 corresponds to item2. Then, count the number of corresponding ngrams (i.e., many-to-one correspondences). For example, count that the above correspondence between item1, item2 and item3 is repeated 2 times, and the above correspondence between item2, item3 and item1 is repeated 1 time, etc. Then, store the correspondences (i.e., high-frequency ngrams) with a number of repetitions exceeding a preset number threshold into a preset index library to build an index library (i.e., ngram index) to control the size of the index. For example, the number threshold can be 1, and accordingly, the correspondence between item1, item2 and item3 can be stored in the above index library. Then, obtain the user's behavior sequence online, that is, obtain multiple behavior data of the target user. Figure 4 The user behavior sequence obtained online is item1, item2, item4, item5. The items obtained on the online side have the same meaning as the items in the user behavior sequence obtained on the offline side. Then, the user behavior sequence obtained online is segmented according to the segmentation number of 2 to obtain multiple target segmentation behavior data combinations, such as the target segmentation behavior data combination item1, item2 and the target segmentation behavior data combination item2, item4, etc. Then, the above target segmentation behavior data combination is queried in the preset index library (that is, the index is queried to achieve recall), and the corresponding recommended content is recalled as the target recommended content and recommended to the target user. For this example, the target segmentation behavior data combination item1 and item2 can be queried in the preset index library, and the corresponding recommended content is item3. Therefore, the operation content corresponding to item3 can be recalled as the recommended content and recommended to the target user.

[0102] As a specific implementation of the present invention, as mentioned above, the preset index library may also store multiple many-to-one relationships, that is, it may simultaneously include the corresponding relationship between different numbers of behavior data and one recommended content. For example, the multiple many-to-one relationships may include the corresponding relationship between three behavior data and one recommended content, and the corresponding relationship between two behavior data and one recommended content.

[0103] Correspondingly, such as Figure 5 As shown, Figure 5 The second flow chart of the index library preset process provided by the embodiment of the present invention may include the following specific steps:

[0104] Step 501, obtaining historical behavior data of each user;

[0105] Step 502: for each user, divide the historical behavior data of the user according to each preset third division quantity, and obtain multiple historical behavior data combinations corresponding to each many-to-one correspondence relationship;

[0106] As described above, the preset index library stores a variety of many-to-one relationships, and each many-to-one relationship corresponds to a different number of behavior data and a recommended content. Therefore, after obtaining the historical behavior data of each user, the historical behavior data can be segmented according to different segmentation numbers. For different types of many-to-one relationships, the corresponding segmentation numbers are different.

[0107] For example, the preset index library stores a plurality of correspondences between three behavior data and one recommended content, and a correspondence between two behavior data and one recommended content. For the convenience of description, the correspondence between three behavior data and one recommended content is recorded as the first correspondence, and the correspondence between two behavior data and one recommended content is recorded as the second correspondence. The first correspondence and the second correspondence correspond to different numbers of segmentations. In the embodiment of the present invention, for each many-to-one correspondence, the corresponding number of segmentations is greater than the number of behavior data corresponding to one recommended content in the correspondence. Specifically, the number of segmentations corresponding to the correspondence may be 1 more than the number of behavior data corresponding to the recommended content in the correspondence. Therefore, the number of segmentations corresponding to the first correspondence is 4, and the number of segmentations corresponding to the second correspondence is 3.

[0108] For a corresponding relationship, the process of segmenting the acquired historical data of each user according to the corresponding segmentation length is the same as step 202, which will not be repeated here.

[0109] In an embodiment of the present invention, the order of segmenting the above-mentioned historical behavior data of each user according to the segmentation lengths corresponding to different corresponding relationships is not specifically limited. For example, for the above-mentioned preset index library storing the first corresponding relationship and the second relationship, the above-mentioned historical behavior data of each user may be first segmented according to a segmentation length of 4, and then segmented according to a segmentation length of 3; or the above-mentioned historical behavior data may be first segmented according to a segmentation length of 3, and then segmented according to a segmentation length of 4.

[0110] After segmenting the above-mentioned historical behavior data of each user according to the segmentation lengths corresponding to different corresponding relationships, multiple historical behavior data combinations corresponding to each many-to-one corresponding relationship can be obtained. The historical behavior data combination corresponding to each many-to-one relationship includes the preset third segmentation number of behavior data corresponding to the many-to-one relationship.

[0111] Step 503: for each historical behavior data combination, based on the corresponding target many-to-one correspondence relationship, the operation content in each historical behavior data in the historical behavior data combination is respectively used as the recommended content, and the correspondence relationship between the recommended content and the target number of other behavior data in the historical behavior data combination is obtained, and the obtained correspondence relationship is used as the second index information; wherein the target number is: the second preset number of behavior data in the target many-to-one correspondence relationship;

[0112] Based on the example in step 502, for the combination of historical behavior data obtained by segmenting the historical behavior data of each user according to the number of segments being 4, it corresponds to the first corresponding relationship. Therefore, for each combination of historical behavior data in this combination of historical behavior data, the operation content of any one of the behavior data can be selected as the recommended content, and a corresponding relationship between it and the other three behavior data in the combination of historical behavior data can be established. For the combination of historical behavior data obtained by segmenting the historical behavior data of each user according to the number of segments being 3, it corresponds to the second corresponding relationship. Therefore, for each combination of historical behavior data in this combination of historical behavior data, the operation content contained in any one of the behavior data can be selected as the recommended content, and a corresponding relationship between it and the other behavior data in the combination of historical behavior data can be established.

[0113] The process of establishing a corresponding number of behavior data and recommended content for a historical behavior data combination is the same as the above step 203, and will not be repeated here.

[0114] Step 504: store the obtained second index information into the preset index library.

[0115] The obtained multiple many-to-one correspondence relationships of different types may be recorded as second index information, and the second index information may be stored in the preset index library.

[0116] As a specific implementation, a number threshold can be set, and the second index information in which the number of repetitions of the correspondence between each second preset number of behavior data and a recommended content exceeds the above number threshold is stored in the preset index library. Step 504 is the same as the above step 204 and will not be repeated here.

[0117] As described above, the preset index library stores different types of many-to-one correspondences. Different many-to-one relationships contain different numbers of correspondences between behavior data and one recommended content. Figure 1 ,like Figure 6As shown, different types of many-to-one relationships correspond to different segmentation lengths of target user behavior data. The segmentation lengths of different target user behavior data can be recorded as the fourth segmentation quantity. For each different type of many-to-one relationship, the fourth segmentation quantity is equal to the number of multiple behavior data corresponding to the recommended content in the many-to-one relationship. Figure 6 This is a third flow chart of the content recommendation method provided by an embodiment of the present invention. Figure 1 Step 102 in the above method, based on the target behavior data of the target user within a preset time period, obtains at least one target recommendation content corresponding to the behavior data of the target user from a preset index library, which may specifically include the following steps:

[0118] Step 602, dividing the plurality of target behavior data within the preset time period of the target user respectively according to each of the preset fourth division quantities, and obtaining a plurality of target segmented behavior data combinations corresponding to each many-to-one correspondence relationship respectively;

[0119] As described above, different types of many-to-one relationships correspond to different numbers of segments for segmenting the target user's behavior data.

[0120] Based on the example in step 502, the first correspondence is a correspondence between three behavior data and one recommended content, and the corresponding fourth segmentation number is 3. Therefore, for the first correspondence, the behavior data of the target user can be segmented according to the segmentation number of 3, thereby obtaining multiple target segmentation behavior data combinations, each of which contains three behavior data. Secondly, the second correspondence is a correspondence between two behavior data and one recommended content, and the corresponding fourth segmentation number is 2. The multiple behavior data of the target user can be segmented again according to the segmentation number of 2 to obtain multiple target segmentation behavior data combinations, each of which contains two behavior data.

[0121] In the embodiment of the present invention, there is no specific limitation on the order of segmenting the multiple behavior data of the target user according to different quantities. For the above example, the behavior data of the target user may be segmented according to the segmentation quantity of 3 first, and then segmented according to the segmentation quantity of 2; or the behavior data of the target user may be segmented according to the segmentation quantity of 2 first, and then segmented according to the segmentation quantity of 3.

[0122] Step 603, searching the preset index library for the target segmentation behavior data combination;

[0123] Based on the example in step 602, after obtaining two or more target segmentation behavior data combinations, the above-mentioned multiple target segmentation behavior data combinations can be searched in the above-mentioned preset index library. When searching, you can first search for a target segmentation behavior data combination containing three behavior data, and then search for a target segmentation behavior data combination containing two behavior data. The present invention does not specifically limit the order of searching for different types of behavior data combinations.

[0124] Step 604, determining whether the target segmentation behavior data combination is found in the preset index library, if found, executing step 605, if not found, executing step 606;

[0125] Step 605, taking the recommended content corresponding to the target segmentation behavior data combination as the target recommended content;

[0126] Step 606: Recommending content without targeting the target segmentation behavior data combination.

[0127] The process of the above steps 603 to 606 is the same as that of the above steps 303 to 306, and will not be repeated here.

[0128] By storing multiple many-to-one relationships and recommending content to target users based on the multiple many-to-one relationships and multiple behavioral data of the target users, the accuracy of content recommendation can be improved while the coverage of content recommendation can be improved.

[0129] As a specific implementation of the embodiment of the present invention, when displaying the above-mentioned recommended contents, the obtained recommended contents can be sorted based on Figure 1 ,like Figure 7 As shown above Figure 1 Step 103 in may include:

[0130] Step 703, sorting the at least one target recommended content according to the time sequence of the target segmentation behavior data combination corresponding to the target recommended content in the multiple target behavior data of the target user;

[0131] The time sequence of the target user's multiple target behavior data may be to arrange the acquired target user's multiple target behavior data in order from near to far from the current time point. For example, if the target user watched video A 15 minutes ago, watched video B 30 minutes ago, and watched video C 45 minutes ago, the acquired behavior data of the target user may be arranged in time sequence as watching video A, watching video B, and watching video C. Of course, in the embodiment of the present invention, the target user's behavior data may also be arranged in order from far to near from the current time point. Accordingly, the behavior data in the above example may be arranged in time sequence as watching video C, watching video B, and watching video A.

[0132] For example, the behavior data of the target user obtained is arranged in chronological order as watching video A, searching B, watching video C, and watching video D, and the above behavior data is segmented according to the segmentation number of 2, and multiple target segmentation behavior data combinations are obtained: watching video A, searching B; searching B, watching video C; watching video C, watching video D; watching video D, watching video A. Based on the chronological order of the behavior data of the target user, the order of the above target segmentation behavior data combinations is that the target segmentation behavior data combination watching video A and searching B ranks first, the target segmentation behavior data combination searching B and watching video C ranks second, the target segmentation behavior data combination watching video C and watching video D ranks third, and the target segmentation behavior data combination watching video D and watching video A ranks last. After searching the above target segmentation behavior data combinations in the preset index library, the target recommended contents corresponding to the above behavior data combinations are search term E, video X, video Y, and video Z, respectively. Since the target segmentation behavior data combination "watch video A, search B" corresponding to search term E is at the top of the target user's behavior data, search term E is ranked first in the recommended content, and correspondingly, video X is ranked second, video Y is ranked third, and video Z is ranked fourth.

[0133] Of course, when the target recommended contents are sorted, they can also be sorted based on the popularity of each target recommended content, specifically, the target recommended contents can be sorted in descending order of popularity. For example, for the target recommended contents video X, video Y, and video Z, if video X has the highest popularity, video Z has the lowest popularity, and video Y has a lower popularity than video X but higher than video Z, then the target recommended contents are sorted such that video X ranks first, video Y ranks second, and video Z ranks third.

[0134] When ranking the target recommended content, it is also possible to comprehensively consider factors such as the time sequence of the corresponding target segmentation behavior data combination in the target user's behavior data and the popularity of the target recommended content. For example, different weight coefficients can be set for different influencing factors (such as the popularity of the target recommended content, the time sequence of the target segmentation behavior data combination corresponding to the target recommended content in the target user's behavior data, etc.), and the comprehensive value of each target recommended content is calculated according to the ranking results of the target recommended content according to the above-mentioned influencing factors and the weight coefficients of each influencing factor, and the target recommended content is ranked according to the comprehensive value.

[0135] Step 704: Display the target recommended content to the target user according to the ranking.

[0136] For the example in step 703, the order of the target recommended contents is that search word E is ranked first, video X is ranked second, video Y is ranked third, and video Z is ranked fourth. Since the display position of the search word in the video application is different from the display position of the video, when displaying the recommended contents, the recommended contents in different display positions can be sorted according to the above sorting results. For this example, the display position of the search word E can be in the search box or in the "Recommended Search" below the search box, and video X, video Y, and video Z can be displayed in order in the "Guess You Like" section of the homepage of the video application.

[0137] As another implementation, the names of the above-mentioned video X, video Y, and video Z may also be displayed, and the names together with the search term E may be displayed in the search box or "recommended search" according to the above-mentioned sorting results, which is not specifically limited here.

[0138] Based on the same technical concept as the above content recommendation method, the embodiment of the present invention also provides a content recommendation device, such as Figure 8 As shown, the device may include:

[0139] The behavior data acquisition module 801 is used to acquire multiple target behavior data of the target user within a preset time period; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior;

[0140] The recommended content acquisition module 802 is used to acquire at least one target recommended content corresponding to the behavior data of the target user from a preset index library based on multiple target behavior data within a preset time period of the target user; the preset index library stores at least one many-to-one correspondence between the behavior data acquired based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content;

[0141] The recommended content recommendation module 803 is configured to recommend the at least one target recommended content to a target user.

[0142] The content recommendation method provided in the embodiment of the present invention first obtains multiple target behavior data of the target user within a preset time period, and based on the multiple target behavior data, obtains at least one target recommended content corresponding to the multiple target behavior data of the target user from a preset index library, and the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, specifically, multiple behavior data corresponds to one recommended content, and then the obtained target recommended content is recommended to the user. In the embodiment of the present invention, the index library is preset based on the historical behavior data of each user, so that the recommended content is not only based on the similarity of the content itself, but also takes into account the user's preference for the content, thereby improving the accuracy of content recommendation. At the same time, the recommended content is obtained through the multiple behavior data of the target user, and the accuracy of content recommendation is higher than that of recommending content based only on the current single behavior data of the target user.

[0143] As a specific implementation of the embodiment of the present invention, the preset index library stores a many-to-one correspondence between behavior data obtained based on historical behavior data of each user and recommended content, specifically: a first preset number of behavior data corresponds to one recommended content;

[0144] The preset index library is preset using the following process:

[0145] Obtain historical behavior data of each user;

[0146] For each user, the historical behavior data of the user is segmented according to a preset first segmentation quantity to obtain a plurality of historical behavior data combinations, wherein each of the historical behavior data combinations contains behavior data of a preset first segmentation quantity; the preset first segmentation quantity is greater than the first preset quantity;

[0147] For each historical behavior data combination, respectively use the operation content in each historical behavior data in the historical behavior data combination as the recommended content, obtain the corresponding relationship between the recommended content and a first preset number of other behavior data in the historical behavior data combination, and use the determined corresponding relationship as the first index information;

[0148] The obtained first index information is stored in the preset index library.

[0149] As a specific implementation of the embodiment of the present invention, the step of storing the obtained first index information into the preset index library includes:

[0150] Among all the first index information, the first index information in which the number of repetitions of the correspondence between the first preset number of behavior data and one recommended content exceeds a preset number threshold is stored in the preset index library.

[0151] As a specific implementation of the embodiment of the present invention, the recommended content acquisition module 802 includes:

[0152] A target segmentation behavior data combination acquisition submodule is used to segment the multiple target behavior data of the target user within the preset time period according to a preset second segmentation number to obtain multiple target segmentation behavior data combinations; wherein the preset second segmentation number is equal to the first preset number;

[0153] A target segmentation behavior data combination search submodule, used to search for the target segmentation behavior data combination in the preset index library;

[0154] The first recommended content acquisition submodule is used to use the recommended content corresponding to the target segmentation behavior data combination as the target recommended content if the target segmentation behavior data combination is found in the preset index library.

[0155] As a specific implementation of an embodiment of the present invention, the preset index library stores a plurality of many-to-one correspondences between behavior data obtained based on historical behavior data of each user and recommended content; wherein each many-to-one correspondence corresponds to a preset third number of cuts, and the preset third number of cuts is greater than the second preset number of behavior data in the correspondence;

[0156] The preset index library is preset using the following process:

[0157] Obtain historical behavior data of each user;

[0158] For each user, the historical behavior data of the user is divided according to each preset third division quantity, and a plurality of historical behavior data combinations corresponding to each many-to-one correspondence relationship are obtained respectively;

[0159] For each historical behavior data combination, based on the corresponding target many-to-one correspondence relationship, the operation content in each historical behavior data in the historical behavior data combination is respectively used as the recommended content, and the correspondence relationship between the recommended content and the target number of other behavior data in the historical behavior data combination is obtained, and the obtained correspondence relationship is used as the second index information; wherein the target number is: the second preset number of behavior data in the target many-to-one correspondence relationship;

[0160] The obtained second index information is stored in the preset index library.

[0161] As a specific implementation of the embodiment of the present invention, each of the many-to-one correspondence relationships corresponds to a preset fourth segmentation quantity, and the preset fourth segmentation quantity is equal to the second preset quantity of behavior data in the correspondence relationship;

[0162] The recommended content acquisition module 802 includes:

[0163] A behavior data segmentation submodule, for segmenting the plurality of target behavior data within the preset time period of the target user according to each preset fourth segmentation quantity, and obtaining a plurality of target segmentation behavior data combinations corresponding to each many-to-one correspondence relationship;

[0164] A behavior data combination search submodule, used to search for the target segmentation behavior data combination in the preset index library;

[0165] The second recommended content acquisition submodule is used to use the recommended content corresponding to the target segmentation behavior data combination as the target recommended content if the target segmentation behavior data combination is found in the preset index library.

[0166] As a specific implementation of the embodiment of the present invention, the recommended content recommendation module 803 includes:

[0167] A content sorting submodule, configured to sort the at least one target recommended content according to the time sequence of the target segmentation behavior data combination corresponding to the target recommended content in the multiple target behavior data of the target user;

[0168] The content recommendation submodule is used to display the target recommended content to the target user according to the ranking.

[0169] The embodiment of the present invention further provides an electronic device, such as Fig. 9 As shown, it includes a processor 901, a communication interface 902, a memory 903 and a communication bus 904, wherein the processor 901, the communication interface 902, and the memory 903 communicate with each other through the communication bus 904.

[0170] Memory 903, used for storing computer programs;

[0171] The processor 901 is used to execute the program stored in the memory 903 to implement the following steps:

[0172] Acquire multiple target behavior data within a preset time period of the target user; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior;

[0173] Based on multiple target behavior data within a preset time period of the target user, at least one target recommended content corresponding to the behavior data of the target user is obtained from a preset index library; the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content;

[0174] The at least one target recommended content is recommended to a target user.

[0175] The content recommendation method provided in the embodiment of the present invention first obtains multiple target behavior data of the target user within a preset time period, and based on the multiple target behavior data, obtains at least one target recommended content corresponding to the multiple target behavior data of the target user from a preset index library, and the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, specifically, multiple behavior data corresponds to one recommended content, and then the obtained target recommended content is recommended to the user. In the embodiment of the present invention, the index library is preset based on the historical behavior data of each user, so that the recommended content is not only based on the similarity of the content itself, but also takes into account the user's preference for the content, thereby improving the accuracy of content recommendation. At the same time, the recommended content is obtained through the multiple behavior data of the target user, and the accuracy of content recommendation is higher than that of recommending content based only on the current single behavior data of the target user.

[0176] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0177] The communication interface is used for communication between the above terminal and other devices.

[0178] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0179] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0180] In another embodiment of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the content recommendation method described in any one of the above embodiments is implemented.

[0181] In another embodiment of the present invention, a computer program product including instructions is provided. When the computer program product is executed on a computer, the computer executes the content recommendation method described in any one of the above embodiments.

[0182] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0183] It should be noted that, in this article, relational terms such as first and second, etc. are only used 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. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0184] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, electronic device, and storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0185] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A content recommendation method, characterized in that: The method comprises: Acquire multiple target behavior data within a preset time period of the target user; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior; Based on multiple target behavior data within a preset time period of the target user, at least one target recommended content corresponding to the behavior data of the target user is obtained from a preset index library; the preset index library stores at least one many-to-one correspondence between the behavior data obtained based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content; recommending the at least one target recommended content to a target user; The preset index library stores a many-to-one correspondence between behavior data obtained based on historical behavior data of each user and recommended content, specifically: a first preset number of behavior data corresponds to one recommended content; The step of acquiring at least one target recommended content corresponding to the target user's behavior data from a preset index library based on multiple target behavior data within a preset time period of the target user includes: Dividing the plurality of target behavior data within the preset time period of the target user according to a preset second segmentation number to obtain a plurality of target segmentation behavior data combinations; wherein the preset second segmentation number is equal to the first preset number; Searching for the target segmentation behavior data combination in the preset index library; If the target segmentation behavior data combination is found in the preset index library, the recommended content corresponding to the target segmentation behavior data combination is used as the target recommended content; The preset index library is preset using the following process: Obtain historical behavior data of each user; For each user, the historical behavior data of the user is segmented according to a preset first segmentation quantity to obtain a plurality of historical behavior data combinations, wherein each of the historical behavior data combinations contains behavior data of a preset first segmentation quantity; the preset first segmentation quantity is greater than the first preset quantity; For each historical behavior data combination, respectively use the operation content in each historical behavior data in the historical behavior data combination as the recommended content, obtain the corresponding relationship between the recommended content and a first preset number of other behavior data in the historical behavior data combination, and use the determined corresponding relationship as the first index information; The obtained first index information is stored in the preset index library.

2. The method according to claim 1, characterized in that: The step of storing the obtained first index information into the preset index library includes: Among all the first index information, the first index information in which the number of repetitions of the correspondence between the first preset number of behavior data and one recommended content exceeds a preset number threshold is stored in the preset index library.

3. The method according to claim 1, characterized in that The preset index library stores a plurality of many-to-one correspondences between behavior data obtained based on historical behavior data of each user and recommended content; wherein each many-to-one correspondence corresponds to a preset third number of segments, and the preset third number of segments is greater than the second preset number of behavior data in the correspondence; The preset index library is preset using the following process: Obtain historical behavior data of each user; For each user, the historical behavior data of the user is divided according to each preset third division quantity, and a plurality of historical behavior data combinations corresponding to each many-to-one correspondence relationship are obtained respectively; For each historical behavior data combination, based on the corresponding target many-to-one correspondence relationship, the operation content in each historical behavior data in the historical behavior data combination is respectively used as the recommended content, and the correspondence relationship between the recommended content and the target number of other behavior data in the historical behavior data combination is obtained, and the obtained correspondence relationship is used as the second index information; wherein the target number is: the second preset number of behavior data in the target many-to-one correspondence relationship; The obtained second index information is stored in the preset index library.

4. The method according to claim 3, characterized in that Each of the many-to-one correspondences corresponds to a preset fourth segmentation quantity, and the preset fourth segmentation quantity is equal to the second preset quantity of behavior data in the correspondence; The step of acquiring at least one target recommended content corresponding to the target user's behavior data from a preset index library based on multiple target behavior data within a preset time period of the target user includes: According to each of the preset fourth segmentation quantities, the multiple target behavior data within the preset time period of the target user are segmented respectively, and multiple target segmentation behavior data combinations corresponding to each many-to-one correspondence relationship are obtained respectively; Searching for the target segmentation behavior data combination in the preset index library; If the target segmentation behavior data combination is found in the preset index library, the recommended content corresponding to the target segmentation behavior data combination is used as the target recommended content.

5. The method according to claim 1, characterized in that The step of recommending the at least one target recommended content to a target user comprises: Sort the at least one target recommended content according to the time sequence of the target segmentation behavior data combination corresponding to the target recommended content in the multiple target behavior data of the target user; The target recommended content is displayed to the target user according to the ranking.

6. A content recommendation device, characterized in that: The device comprises: The behavior data acquisition module is used to acquire multiple target behavior data within a preset time period of the target user; the preset time period is: a time period of a preset length from the current moment forward; wherein each behavior data includes: an operation behavior and an operation content corresponding to the operation behavior; A recommended content acquisition module, configured to acquire at least one target recommended content corresponding to the behavior data of the target user from a preset index library based on multiple target behavior data within a preset time period of the target user; the preset index library stores at least one many-to-one correspondence between the behavior data acquired based on the historical behavior data of each user and the recommended content, wherein the correspondence is that multiple behavior data correspond to one recommended content; A recommended content recommendation module, used to recommend the at least one target recommended content to a target user; The preset index library stores a many-to-one correspondence between behavior data obtained based on historical behavior data of each user and recommended content, specifically: a first preset number of behavior data corresponds to one recommended content; The recommended content acquisition module is specifically used to: divide the multiple target behavior data of the target user within the preset time period according to the preset second division number to obtain multiple target division behavior data combinations; wherein the preset second division number is equal to the first preset number; Searching for the target segmentation behavior data combination in the preset index library; If the target segmentation behavior data combination is found in the preset index library, the recommended content corresponding to the target segmentation behavior data combination is used as the target recommended content; The preset index library is preset using the following process: Obtain historical behavior data of each user; For each user, the historical behavior data of the user is segmented according to a preset first segmentation quantity to obtain a plurality of historical behavior data combinations, wherein each of the historical behavior data combinations contains behavior data of a preset first segmentation quantity; the preset first segmentation quantity is greater than the first preset quantity; For each historical behavior data combination, respectively use the operation content in each historical behavior data in the historical behavior data combination as the recommended content, obtain the corresponding relationship between the recommended content and a first preset number of other behavior data in the historical behavior data combination, and use the determined corresponding relationship as the first index information; The obtained first index information is stored in the preset index library.

7. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 5 when executing a program stored in a memory.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 5 are implemented.

Images