Information recommendation method and device, computer device, and storage medium

Abstract

The application relates to an information recommendation method and device, computer equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: summing up the historical recommendation times of various types of information within N historical refresh times; adding a corresponding temporary rule to target type information within a current refresh time, the temporary rule being used to indicate a recommendation quantity constraint condition of the target type information within a first unit recommendation window of the current refresh time; the target type information being type information whose relationship between the sum of the historical recommendation times and a corresponding recommendation time threshold meets a set condition; generating an information recommendation sequence within the current refresh time based on the temporary rule and inherent rules of various types of information; and performing information recommendation on to-be-recommended information in the information recommendation sequence. Through the above method, the information recommendation sequence under the current refresh time can be different from the information recommendation sequence under the historical refresh time, so that the diversity of information recommendation between refresh times is increased.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of Internet, and particularly relate to an information recommendation method and device, a computer device and a storage medium. BACKGROUND

[0002] In the field of information stream recommendation, personalized recommendation faces the problem of balancing the relevance and diversity of the recommendation results, wherein the relevance refers to the matching degree between the recommended results and the user's interest preferences, and the diversity refers to the types of information and the number of information categories contained in the recommended results; if the diversity of the recommended results is not considered, the recommendation system will always be based on the known interests of the user for recommendation and utilization, resulting in serious homogenization of the recommended results, affecting the user experience, and by adding a diversity dispersion link, the user interest fatigue can be effectively avoided, and it is beneficial to the recommendation system to continuously explore and discover the user's potential unknown interests, and beneficial to improving the user experience and promoting the long-term healthy development of the recommendation system ecosystem. In the recommendation system, the rearrangement stage is generally the last module of the recommendation system and is closest to the user level, and the strategy of this stage has the most direct impact on the user. Therefore, the diversity dispersion link is also mostly set in the recommendation rearrangement stage.

[0003] In related technologies, a rule constraint-based dispersion method is usually used to disperse the recommendation order of the to-be-recommended information under the current refresh number, so as to ensure the diversity of the recommended results within the current refresh number.

[0004] However, the effect of the above dispersion method can only ensure the diversity of the recommended results under a single refresh number, and if the recommendation sequence is lengthened and extended to the recommendation sequence of previous refresh numbers, the recommended results of each refresh number in the short term will have a high repetition rate, thereby causing homogenization of the recommended content between different refresh numbers and affecting the recommendation effect. SUMMARY

[0005] Embodiments of the present application provide an information recommendation method, device, computer device and storage medium, which can avoid the same target user from receiving multiple information recommendations in a short period of time, avoid the situation that the target user is frequently disturbed, thereby improving the user experience and improving the effect of information recommendation. The technical solution is as follows:

[0006] In one aspect, an information recommendation method is provided, and the method comprises:

[0007] Sum up the sum of the historical recommendation numbers of each type of information within N historical refresh numbers; N is a positive integer;

[0008] add a corresponding temporary rule for the target type of information in the current refresh number, the temporary rule being used to indicate a recommended quantity constraint condition of the target type of information in a first unit recommended window of the current refresh number; the target type of information is a type of information whose relationship between a sum of historical recommended numbers and a corresponding recommended number threshold meets a set condition;

[0009] generate an information recommendation sequence in the current refresh number based on the temporary rule and a corresponding inherent rule of each type of information; the inherent rule is used to indicate a maximum recommendable quantity of the corresponding type of information in a second unit recommended window; wherein for the target type of information, a recommended frequency determined based on the corresponding inherent rule is different from a recommended frequency determined based on the corresponding temporary rule;

[0010] recommend each to-be-recommended information in the information recommendation sequence.

[0011] In another aspect, an information recommendation apparatus is provided, the apparatus comprising:

[0012] a number counting module configured to count a sum of historical recommended numbers of each type of information in N historical refresh numbers; N is a positive integer;

[0013] a rule adding module configured to add a corresponding temporary rule for a target type of information in the current refresh number, the temporary rule being used to indicate a recommended quantity constraint condition of the target type of information in a first unit recommended window of the current refresh number; the target type of information is a type of information whose relationship between a sum of historical recommended numbers and a corresponding recommended number threshold meets a set condition;

[0014] a sequence generating module configured to generate an information recommendation sequence in the current refresh number based on the temporary rule and a corresponding inherent rule of each type of information; the inherent rule is used to indicate a maximum recommendable quantity of the corresponding type of information in a second unit recommended window; wherein for the target type of information, a recommended frequency determined based on the corresponding inherent rule is different from a recommended frequency determined based on the corresponding temporary rule;

[0015] an information recommendation module configured to recommend each to-be-recommended information in the information recommendation sequence.

[0016] In an optional implementation, the rule adding module is configured to,

[0017] add a corresponding temporary rule for the target type of information in the current refresh number, the temporary rule being used to indicate a recommended quantity constraint condition of the target type of information in a first unit recommended window of the current refresh number; the target type of information is a type of information whose relationship between a sum of historical recommended numbers and a corresponding recommended number threshold meets a set condition;

[0018] In a case where the sum of the historical recommendation times of the target type of information is less than a second recommendation time threshold corresponding to the target type of information, a temporary rule corresponding to the target type of information is added for increasing the recommendation frequency within the current refresh time, where the application priority of the temporary rule is higher than the application priority of the inherent rule, and the first recommendation time threshold is greater than the second recommendation time threshold.

[0019] In an optional implementation, in a case where the sum of the historical recommendation times of the target type of information is greater than a first recommendation time threshold corresponding to the target type of information, the temporary rule is used to indicate the maximum recommendable quantity of the target type of information within a first unit recommendation window of the current refresh time.

[0020] The first recommendation proportion corresponding to the temporary rule of the target type of information is less than the second recommendation proportion corresponding to the inherent rule of the target type of information.

[0021] The recommendation proportion refers to the proportion between the maximum recommendable quantity indicated by the rule and the maximum quantity of information to be recommended that can be contained in the corresponding unit recommendation window.

[0022] In an optional implementation, in a case where the sum of the historical recommendation times of the target type of information is less than a second recommendation time threshold corresponding to the target type of information, the temporary rule is used to indicate the minimum recommendation quantity of the target type of information within a first unit recommendation window of the current refresh time.

[0023] In an optional implementation, the sequence generation module comprises:

[0024] A sequence acquisition submodule is configured to acquire an original recommendation sequence, where the original recommendation sequence is a recommendation sequence obtained after information sorting of each piece of information to be recommended.

[0025] A sequence generation submodule is configured to sort and adjust each piece of information to be recommended in the original recommendation sequence according to the recommendation rules of each type of information, to generate an information recommendation sequence within the current refresh time, where each type of information includes the target type of information and a non-target type of information, the recommendation rule corresponding to the target type of information includes the inherent rule and the temporary rule, and the recommendation rule corresponding to the non-target type of information includes the inherent rule.

[0026] In an optional implementation, each rule has a corresponding application priority.

[0027] The sequence generation submodule is configured to generate the information recommendation sequence within the current refresh time based on the information sorting determined based on the first rule in a case where the information sorting determined based on the first rule conflicts with the information sorting determined based on the second rule.

[0028] The first rule and the second rule are respectively a recommendation rule of two types of information, or the first rule and the second rule are different rules of the same type of information; the application priority of the first rule is higher than the application priority of the second rule.

[0029] In an optional implementation, the apparatus further includes:

[0030] The storage module is configured to store, in a database, the historical recommendation times of the types of information within M historical refresh times; M > N, and M is a positive integer.

[0031] The deletion module is configured to delete, when the historical refresh times are greater than M, the historical recommendation times of the types of information stored in the database at the earliest historical refresh times.

[0032] In an optional implementation, the apparatus further includes:

[0033] The rule destruction module is configured to destroy the temporary rule corresponding to the target type of information after completing the information recommendation within the current refresh time.

[0034] In another aspect, a computer device is provided, which includes a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the information recommendation method described above.

[0035] In another aspect, a computer readable storage medium is provided, which stores at least one computer program, the computer program being loaded and executed by a processor to implement the information recommendation method described above.

[0036] In another aspect, a computer program product is provided, which includes at least one computer program, the computer program being loaded and executed by a processor to implement the information recommendation method provided in various optional implementations described above.

[0037] The technical solutions provided in the present application can include the following beneficial effects:

[0038] The information recommendation method provided in the embodiments of the present application is to count the sum of the historical recommendation times of each type of information in N historical refreshing times before performing information recommendation, and in the case that the relationship between the sum of the historical recommendation times of the target type of information and the corresponding recommendation time threshold satisfies a set condition, it is indicated that the recommendation strategy for the target type of information needs to be changed, and the corresponding temporary rule is added for the target type of information in the current refreshing time; then, the information recommendation sequence in the current refreshing time is generated based on the temporary rule of the target type of information and the inherent rule of each type of information, so as to perform information recommendation on each to-be-recommended information according to the recommendation order indicated by the information recommendation sequence, wherein the recommendation frequency of the target type of information determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule. Through the above method, when performing information recommendation in the current refreshing time, the computer device can adjust the recommendation rule in combination with the historical recommendation of each information category in the historical refreshing time, and adjust the recommendation frequency of the target type of information in each type of information through the temporary rule, so as to realize the push control of the target type of information, and also make the information recommendation sequence in the current refreshing time different from the information recommendation sequence in the historical refreshing time, thereby increasing the diversity of information recommendation between refreshing times.

[0039] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0040] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the present application.

[0041] Figure 1 A schematic diagram of sequence adjustment by the rule-constrained dispersing method provided in an exemplary embodiment of the present application is shown;

[0042] Figure 2 A flowchart of the information recommendation method provided in an exemplary embodiment of the present application is shown;

[0043] Figure 3 A flowchart of the information recommendation method provided in an exemplary embodiment of the present application is shown;

[0044] Figure 4 A process schematic diagram of generating an information recommendation sequence by the recommendation rules of each type of information provided in an exemplary embodiment of the present application is shown;

[0045] Figure 5 A block diagram of the information recommendation device provided in an exemplary embodiment of the present application is shown;

[0046] Figure 6A structural block diagram of a computer device according to an example embodiment of the present application is shown.

[0047] Figure 7 A structural block diagram of a computer device according to an example embodiment of the present application is shown. DETAILED DESCRIPTION

[0048] The example embodiments will be described in detail herein with reference to the attached drawings. The following description is made with reference to the accompanying drawings in which like reference numerals refer to like elements, unless otherwise indicated. The following description of example embodiments is not representative of all embodiments consistent with the present application. Rather, it is merely an example of apparatus and methods consistent with some aspects of the present application as detailed in the appended claims.

[0049] The dispersing method for rearrangement diversity in the related art includes a rule constraint-based dispersing method, a heuristic method-based dispersing method such as MMR (Maximal Marginal Relevance), DPP (Determinantal Point Process), and a deep model-based dispersing method; wherein the rule constraint-based dispersing method has weaker generalization ability than other methods, but is not prone to bad cases, is simple and controllable, and has strong interpretability, so the mainstream diversity dispersing technology is still the rule constraint-based dispersing method. The most commonly used rule constraint-based dispersing method is the sliding window dispersing method, the scheme idea of which is that the size of the sliding window is first set, and then in the sliding window, when the frequency of the item objects of the same category attribute exceeds the rule constraint, the item objects that meet the rule constraint in the subsequent sequence are adjusted to this position, and the original items in this position are sequentially moved backward until all the item objects in the window meet the rule constraint. If a special case occurs, i.e., there is no item object meeting the rule constraint, the window can continue to move backward until the sliding window adjusts the entire sequence. This method has a relatively low degree of damage to the original sequence order and can maximize the correlation preservation. An implementation method of a sliding window rule case is given below, Figure 1 A schematic diagram of sequence adjustment based on the rule constraint-based dispersing method according to an example embodiment of the present application is shown, as shown in Figure 1 The original sequence 110 contains 9 item objects arranged in order according to the recommended order, and it is assumed that the defined rule constraint is that no item object of the same information category appears in a recommended window with a size of 3. Then, with the movement of the sliding window, the following judgment process and item object position adjustment process occur.

[0050] The first window, item2 and item3 belong to the same information category, and the subsequent object item4 meets the rule constraint, and item4 is adjusted to the window, and thus all items in the second window meet the rule constraint.

[0051] The second window, item2 and item3 belong to the same information category, and the subsequent object item4 meets the rule constraint, and item4 is adjusted to the window, and thus all items in the second window meet the rule constraint.

[0052] The third window, all items in the third window meet the rule constraint, and no adjustment is needed, and the window continues to slide.

[0053] The fourth window, all items in the fourth window meet the rule constraint, and no adjustment is needed, and the window continues to slide.

[0054] The fifth window, item6 and item7 belong to the same information category, and the subsequent object item8 meets the rule constraint, and item8 is adjusted to the window, and thus all items in the fifth window meet the rule constraint.

[0055] The sixth window, item6 and item7 belong to the same information category, and the subsequent object item9 cannot meet the window rule constraint, and no processing is performed, and the window continues to slide.

[0056] The seventh window, the last window of the sequence, item8 and item9 belong to the same information category, and there is no adjustable item object, and no processing is performed, and thus the sliding window operation adjustment of the entire sequence ends.

[0057] However, the above diversity dispersion method is to disperse the recommended information in the current user refresh times, that is, only the dispersion of the current request recommendation result is considered, and the time sequence diversity is lacking. The time sequence diversity refers to the diversity of new recommendation results compared with the past recommendation results in a period of time. If only the dispersion in a refresh time is considered, although the results in the current refresh time ensure diversity, the recommendation sequence is extended to the recommendation sequence of the past refresh times, and the category repetition will appear in the short-term recommendation results of each refresh time. Because the recommendation sorting model generally cannot capture the change of user interest in real time, it will always recommend based on the past user interest for recommendation sorting, and thus the recommended content will be homogenized between different refresh times.

[0058] In order to solve the problems in the diversity dispersion manner in the related art, the information recommendation method provided by the present application can ensure the diversity of the recommendation results between the refresh times, Figure 2 A flowchart of an information recommendation method provided by an example embodiment of the present application is shown, which can be executed by a computer device, which can be implemented as a server or a terminal, as shown in Figure 2 The information recommendation method can include the following steps:

[0059] In step 210, the sum of the historical recommendation times of each type of information in N historical refresh times is counted; N is a positive integer.

[0060] In order to ensure the timeliness of the historical recommendation times obtained by counting, the historical refresh times can be the N historical refresh times close to the current refresh time, for example, the sum of the historical recommendation times of each type of information in the last 3 refresh times; the value of N can be set by relevant personnel, which is not limited by the present application. The value of N is used to indicate the dimension of dispersion across refresh times, that is, to indicate dispersion between N refresh times.

[0061] The historical recommendation times of each type of information refer to the times when the information belonging to each information category (i.e., each type of information) is recommended to the user in the historical refresh times.

[0062] In an optional implementation, the information categories can be divided based on the information field to which the information belongs, for example, information categories divided based on the information fields of entertainment, technology, life, etc.; or, in another optional implementation, the information categories can be divided based on the publishing account of the information, for example, information categories divided according to the information published by account A, the information published by account B, etc.; the division method of the information categories can be set by relevant personnel, which is not limited by the present application.

[0063] In step 220, a corresponding temporary rule is added for the target type of information in the current refresh time, which is used to indicate the recommendation quantity constraint condition of the target type of information in the first unit recommendation window of the current refresh time; the target type of information is the type of information whose sum of historical recommendation times and corresponding recommendation time threshold satisfy a set condition.

[0064] The target type of information is the information belonging to the target information category; the target information category refers to the information category among the information categories contained in the current refresh time, whose corresponding sum of historical recommendation times and corresponding recommendation time threshold satisfy a set condition. The set condition can be a condition corresponding to the numerical relationship between the sum of historical recommendation times and the corresponding recommendation time threshold.

[0065] In the embodiments of the present application, the computer device corresponding to different information categories can be provided with different recommendation times thresholds; the recommendation times threshold can be used to limit the maximum number of recommendations within N refresh times, or can also be used to limit the minimum number of recommendations within N refresh times; when the sum of the historical recommendation times of the target category information within N historical refresh times and the recommendation times threshold corresponding thereto meet the set condition, it indicates that the recommendation frequency of the target category information needs to be adjusted, and the corresponding temporary rule is added to the target category information within the current refresh time to recommend the target category information, so as to achieve the purpose of dispersing the target category information within the current refresh time.

[0066] The recommendation times threshold corresponding to each information category can be set by relevant personnel based on actual needs, and the present application does not limit this.

[0067] The recommendation number constraint condition can be a constraint condition for limiting the maximum number of recommendations of the target category information within the first unit recommendation window, or can be a constraint condition for limiting the minimum number of recommendations of the target category information within the first unit recommendation window. The recommendation number constraint condition can be set based on actual conditions.

[0068] Step 230, generating an information recommendation sequence within the current refresh time based on the temporary rule and the inherent rule corresponding to each category of information; the inherent rule is used to indicate the maximum recommendable number of the corresponding category of information within the second unit recommendation window; wherein for the target category information, the recommendation frequency determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule.

[0069] The recommendation frequency is used to represent the number of recommendations within a unit time; the temporary rule can be used to control the number of recommendations of the target category information.

[0070] In the embodiments of the present application, the inherent rule and the temporary rule of the target category information act on the ordering of the target category information in the information recommendation sequence, so when generating the information recommendation sequence within the current refresh time, the ordering position of the target category information in the information recommendation sequence needs to satisfy the corresponding inherent rule and temporary rule as much as possible, so as to achieve the purpose of reordering the target category information; when it is impossible to satisfy the corresponding inherent rule and temporary rule at the same time, the rule with higher priority can be satisfied first.

[0071] In the generation of the information recommendation sequence, the computer device performs rule increase judgment on the information categories to which the plurality of to-be-recommended information contained in the current refresh number belongs, to determine the target category information from the information categories, wherein the rule increase judgment refers to a judgment process of determining whether to increase the corresponding temporary rule for the information category, that is, in the case that the sum of the historical recommendations corresponding to the information category for which the rule increase judgment is performed and the recommendation number threshold corresponding thereto satisfies the set condition, the information category is determined to be the target information category, and the corresponding temporary rule is added to the to-be-recommended information under the information category, so as to sort the to-be-recommended information under the information category based on the temporary rule corresponding to the information category and the inherent rule; and in the case that the sum of the historical recommendations corresponding to the information category for which the rule increase judgment is performed and the recommendation number threshold corresponding thereto does not satisfy the set condition, the information category is determined to be a non-target information category, and the to-be-recommended information under the information category is sorted based on the inherent rule corresponding thereto. Wherein the temporary rules corresponding to each information category can be the same or different, and the inherent rules corresponding to each information category can be the same or different, which are not limited by the present application.

[0072] After completing the rule increase judgment on each information category within the current recommendation number, the information categories are sorted and adjusted according to the rules corresponding to each information category respectively, to obtain an information recommendation sequence, and the information sorting in the information recommendation sequence conforms to the rules of each information category as much as possible.

[0073] Step 240, information recommendation is performed on each to-be-recommended information in the information recommendation sequence.

[0074] According to the recommendation order indicated by the information recommendation sequence, each to-be-recommended information is recommended in turn.

[0075] To sum up, the information recommendation method provided in the embodiments of the present application, before information recommendation, sums up the sum of the historical recommendation times of each type of information in N historical refresh times, and in the case that the relationship between the sum of the historical recommendation times of the target type of information and the recommendation time threshold corresponding to the target type of information satisfies the set condition, it is indicated that the recommendation strategy for the target type of information needs to be changed, and the corresponding temporary rule is added for the target type of information in the current refresh time; then, based on the temporary rule of the target type of information and the inherent rule of each type of information, the information recommendation sequence in the current refresh time is generated, so as to recommend each to-be-recommended information according to the recommendation order indicated by the information recommendation sequence, wherein for the target type of information, the recommendation frequency determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule. Through the above method, when the information recommendation in the current refresh time is performed, the computer device can adjust the recommendation rule in combination with the historical recommendation of each information category in the historical refresh time, and adjust the recommendation frequency of the target type of information in each type of information through the temporary rule, so as to realize the push control of the target type of information, and also make the information recommendation sequence in the current refresh time different from the information recommendation sequence in the historical refresh time, thereby increasing the diversity of information recommendation between refresh times.

[0076] In an optional application scenario, the corresponding temporary rule is added for the target type of information in order to limit the recommendation times of the target type of information in the current refresh time, so as to avoid the homogenization of refresh results between different refresh times. In another optional application scenario, the corresponding temporary rule is added for the target type of information in order to increase the recommendation times of the target type of information in the current refresh time, so as to increase the recommendation amount of the target type of information. Figure 3 A flowchart of an information recommendation method provided by an example embodiment of the present application is shown, which can be executed by a computer device, which can be implemented as a server or a terminal, as shown in Figure 3 The information recommendation method can include the following steps.

[0077] In step 310, the sum of the historical recommendation times of each type of information in N historical refresh times is counted; N is a positive integer.

[0078] In the embodiments of the present application, the computer device can obtain the historical recommendation times of each type of information in N historical refresh times from the database to calculate the sum of the historical recommendation times of each type of information.

[0079] In order to reduce the storage pressure of the database, in the embodiments of the present application, the method further includes:

[0080] The historical recommendation times of each type of information in M historical refresh times are stored in the database; M≥N, and M is a positive integer;

[0081] When the historical refresh number is greater than M, the historical recommendation number of each type of information stored in the database under the earliest refresh number is deleted.

[0082] That is, only the category information of the information exposed to the user within M refresh numbers is recorded in the database, if the latest refresh number of the user exceeds M refresh numbers, the record of the historical recommendation number of each type of information stored in the database under the earliest refresh number is deleted, only the record of the historical recommendation number of each type of information under the latest M refresh numbers is kept, thereby reducing the data storage pressure of the database.

[0083] In the embodiment of the present application, the database can be a graph storage database.

[0084] The computer device can obtain the historical recommendation number of each type of information under the latest N refresh numbers from the database; when the historical refresh number recorded in the database is less than N, the historical recommendation number of each type of information under the historical refresh number recorded in the database is counted, at this time, the value of N is equal to the value of the historical refresh number recorded in the database; illustratively, if the computer device needs to obtain the historical recommendation number of each type of information under the latest 5 refresh numbers from the database, that is, N=5, but at this time, 5 refreshes have not been performed, only the historical recommendation number of each type of information under 3 refresh numbers is recorded in the database, then the historical recommendation number of each type of information under the 3 refresh numbers is extracted, the number is counted and judged, at this time, N=3.

[0085] Step 320, adding a corresponding temporary rule for the target category information within the current refresh number, the temporary rule is used to indicate the recommendation quantity constraint condition of the target category information within the first unit recommendation window of the current refresh number; the target category information is the category information whose relationship between the sum of the historical recommendation number and the corresponding recommendation number threshold satisfies the set condition.

[0086] In the same judgment process, multiple types of rule increase judgment can be performed simultaneously. Illustratively, if the computer device sets the maximum recommendation number threshold of information under the same publishing account in 3 historical refresh times as 5, the maximum recommendation number threshold of information in the same information field in 3 historical refresh times as 12, and the like, at this time, the computer device needs to count the sum of the historical recommendation numbers of each publishing account in 3 historical refresh times, and count the sum of the historical recommendation numbers of each information field in 3 historical refresh times, and compare the sum of the six historical recommendation numbers with the corresponding recommendation number threshold to determine whether a temporary rule needs to be added for the corresponding information category; for example, if the sum of the historical recommendation numbers of publishing account A in 3 historical refresh times is 4, which is less than 5, no temporary rule is added, while the sum of the historical recommendation numbers of publishing account B in 3 historical refresh times is 6, which is greater than 5, and the corresponding temporary rule is added. It should be noted that the setting of the above-mentioned recommendation number threshold is only illustrative, and the recommendation number threshold corresponding to different information categories can be the same or different, and the temporary rules corresponding to each information category can be the same or different.

[0087] In the embodiment of the present application, each information category corresponds to an inherent rule, which is used to indicate the maximum number of recommendations of the corresponding information category in a second unit recommendation window. The inherent rules corresponding to each information category can be the same, for example, the inherent rules corresponding to each information category are “in a unit recommendation window with a size of 3, the same information category information does not appear”. Alternatively, the inherent rules corresponding to each information category can also be different, for example, the inherent rule corresponding to information category 1 is “in a unit recommendation window with a size of 3, a maximum of 2 times”, the inherent rule corresponding to information category 2 is “in a unit recommendation window with a size of 2, a maximum of 1 time”, and the like, and the inherent rules of each information category can be set based on actual needs, which are not limited by the present application.

[0088] In an optional case, in the case where the sum of the historical recommendation numbers of the target information category is greater than the first recommendation number threshold corresponding to the target information category, a temporary rule corresponding to the target information category is added in the current refresh time to reduce the recommendation frequency.

[0089] The first recommendation number threshold is used to limit the maximum number of recommendations of the target information category in N historical refresh times. In this case, the temporary rule is used to indicate the maximum number of recommendations of the target information category in the first unit recommendation window of the current refresh time.

[0090] In order to achieve the effect of controlling the recommended quantity of target class information, in the embodiments of the present application, when the temporary rule is used to limit the recommended quantity of target class information, the first recommended proportion corresponding to the temporary rule of the target class information is less than the second recommended proportion corresponding to the inherent rule of the target class information.

[0091] The recommended proportion refers to the proportion between the numerical value of the maximum recommended quantity indicated by the rule and the maximum quantity of the to-be-recommended information that can be contained in the corresponding unit recommendation window.

[0092] In order to achieve the above requirement, in an optional case, the specification of the first unit recommendation window in the temporary rule is the same as the specification of the second unit recommendation window in the inherent rule, but the maximum recommended quantity indicated in the temporary rule is less than the maximum recommended quantity indicated in the inherent rule, wherein the specification of the unit recommendation window is used to indicate the quantity of to-be-recommended information that can be contained in one unit recommendation window; or, in another optional case, the maximum recommended quantity indicated in the temporary rule is the same as the maximum recommended quantity indicated in the inherent rule, but the specification of the first unit recommendation window is greater than the specification of the second unit recommendation window; or, in another optional case, the specification of the first unit recommendation window is different from the specification of the second unit recommendation window, and the maximum recommended quantity corresponding to the inherent rule is also different from the maximum recommended quantity corresponding to the temporary rule, but the condition that the first recommended proportion is less than the second recommended proportion is met, for example, the temporary rule can be "appears at most 2 times in one unit recommendation window with a size of 5", the inherent rule is "appears at most 3 times in one unit recommendation window with a size of 4", and the like.

[0093] Optionally, the type of the unit recommendation window includes at least one of a sliding window and a head window. In the embodiments of the present application, the information recommendation method provided by the present application is described by taking the type of the unit recommendation window as the sliding window as an example.

[0094] Taking the division of information categories according to information fields as an example, the following illustratively shows the adding process of a temporary rule:

[0095] Illustratively, the definition form of the inherent rule of each information field is as follows:

[0096] "id=1;class=Slide Window;tag=1001;priority=1;args=5,1;"

[0097] Wherein, id is the number of rules, one rule corresponds to one id; class is the window rule type, SlideWindow is the sliding window rule, Top Window is the head fixed window rule, and the above inherent rules have the sliding window rule; tag is the mark of the information meeting the conditions, indicating the category of a rule, such as the tag of the inherent rule of each information field is defined as 1001, and each tag also carries the information field identifier of this tag, which is used to distinguish specific information fields, such as each information field contains entertainment stars, society, military, and humor, etc., and the inherent rule will only have information with the same information field identifier to be scattered, such as the entertainment star field, the information under the field is marked with 1001, and an information field identifier indicating "entertainment star" is recorded, and the society field is scattered, the information under the field is also marked with 1001, and an information field identifier indicating "society" is recorded, they belong to the same rule under tag=1001, and share the same rule constraint, but the rule constraint is only effective for information with the same information field identifier; priority represents the priority, and when the window rules conflict, the one with high priority takes effect; the first parameter in args represents the window size, and the rule in the example indicates that a unit recommendation window with a size of 5 is defined, and the second parameter is the maximum number of times the corresponding information field in the unit recommendation window can appear, and the rule in the example represents that the maximum number of times is 1.

[0098] If the number of the same category information field appearing in the three historical refresh times exceeds 12 times, a temporary rule is added to the information field with exposure times exceeding 12 times in the three historical refresh times:

[0099] "id=2;class=Slide Window;tag=1002;priority=1;args=6,1;"

[0100] For example, the exposure of "entertainment stars" and "society" to users in the three historical refresh times exceeds 12 times, so the tag of the two information fields in the current refresh time is added, and the tag number is recorded as 1002, and the temporary rule is that the maximum number of times in the unit recommendation window with a size of 6 can be 1, at this time, for the "entertainment star" and "society" information fields, each recommended information has two tags, i.e. 1001 and 1002, and the recommended information under other information fields has one tag, i.e. 1001.

[0101] In another optional case, in a case where the sum of the historical recommendation times of the target category information is less than the second recommendation time threshold corresponding to the target category information, a temporary rule corresponding to the target category information for increasing the recommendation frequency is added within the current refresh time, wherein the application priority of the temporary rule is higher than the application priority of the inherent rule, and the first recommendation time threshold is greater than the second recommendation time threshold.

[0102] The second recommendation time threshold is used to limit the minimum recommendation quantity of the target category information within N historical refresh times.

[0103] In a case where the same category information simultaneously has the highest recommendation time limit and the lowest recommendation time limit within N historical refresh times, the second recommendation time threshold corresponding to the same category information is less than the first recommendation time threshold.

[0104] Since the temporary rule of the target category information and the inherent rule simultaneously act on the sorting of the target category information, and in a case where the temporary rule is used to increase the recommendation frequency of the target category information, there may be a conflict with the information sorting determined by the inherent rule, therefore, the application priority of the temporary rule needs to be set to be higher than the application priority of the inherent rule, so as to ensure the purpose of increasing the recommendation frequency of the target category information.

[0105] Optionally, the temporary rule can be used to indicate the maximum recommendable quantity of the target category information within the first unit recommendation window of the current refresh time, at this time, the first recommendation proportion corresponding to the temporary rule is greater than the second recommendation proportion corresponding to the inherent rule of the target category information; for example, the inherent rule is “within a unit recommendation window with a size of 4, a maximum of 2 times”, and the temporary rule can be “within a unit recommendation window with a size of 4, a maximum of 3 times”.

[0106] However, increasing the recommendation frequency of the target category information by changing the maximum recommendable quantity within the unit recommendation window has instability and uncertainty; therefore, in another optional implementation manner, in a case where the sum of the historical recommendation times of the target category information is less than the second recommendation time threshold corresponding to the target category information, the temporary rule is used to indicate the minimum recommendation quantity of the target category information within the first unit recommendation window of the current refresh time.

[0107] That is, the inherent rule of the target category information indicates the maximum number of recommendations within the second unit recommendation window, but the temporary rule indicates the minimum number of recommendations within the first unit recommendation window, and the temporary rule is preferentially met to ensure that the target category information appears the minimum number of times corresponding to the minimum number of recommendations within the first unit recommendation window, thereby effectively improving the recommendation frequency of the target category information. For example, the inherent rule is "appear at most 2 times within a unit recommendation window with a size of 4", and the temporary rule is set to "appear at least 2 times within a unit recommendation window with a size of 4". At this time, the inherent rule and the temporary rule can be met at the same time when the target category information is sorted, and the two do not conflict, but there will be a case where the temporary rule is met but the inherent rule is not met when actually sorting the information. However, if the temporary rule is "appear at most 3 times within a unit recommendation window with a size of 4", the temporary rule and the inherent rule conflict, and at this time, the temporary rule is preferentially met based on the priority setting.

[0108] In step 330, an original recommendation sequence is obtained, which is a recommendation sequence obtained after information sorting of each to-be-recommended information.

[0109] The information sorting is a sorting process based on the sorting scores of each to-be-recommended information.

[0110] In step 340, each to-be-recommended information in the original recommendation sequence is sorted and adjusted according to the recommendation rules of each category information to generate an information recommendation sequence within the current refresh number; the target category information and the non-target category information are included in the category information, the inherent rule and the temporary rule are included in the recommendation rule corresponding to the target category information, and the inherent rule is included in the recommendation rule corresponding to the non-target category information.

[0111] In the embodiment of the present application, the computer device can create an empty sequence, which can include K to-be-recommended information, K being a positive integer; according to the recommendation rules of each category information, information meeting the respective recommendation rules of each category information is selected according to the arrangement order of the original recommendation sequence and is sequentially filled into the empty sequence, and in view of the conflict between the rules, in an optional implementation manner, each rule has a corresponding application priority.

[0112] In the case where the information sorting determined based on the first rule and the information sorting determined based on the second rule conflict, the information sorting determined based on the first rule generates an information recommendation sequence within the current refresh number;

[0113] The first rule and the second rule are the respective recommendation rules of the two categories of information, or the first rule and the second rule are different rules of the same category of information; the application priority of the first rule is higher than the application priority of the second rule.

[0114] That is, when filling the recommended information in the empty sequence according to the respective rules, if there is a conflict between two rules, the recommended information is sorted according to the rule with higher priority.

[0115] The priority comparison between the above rules can exist between the respective rules of different types of information, such as the priority comparison between the respective inherent rules of different types of information; or exist between different rules of the same type of information, such as the priority comparison between the inherent rule and the temporary rule of the same type of information, and the like. Through the setting of the priority, the rule setting can be more flexible and adjustable, so that the information recommendation method can adapt to various scenarios.

[0116] If there is no recommended information satisfying the rule to fill in a certain position, the recommended information with the highest ranking in the original sequence that has not been selected is filled in, until the empty sequence is filled, and the information recommendation sequence is generated.

[0117] Alternatively, in another optional case, the computer device can also adjust the position of each recommended information in the original recommendation sequence based on the recommendation rules of each type of information, sequentially move the recommended information that does not meet the recommendation rules, until the sliding window operation of the entire sequence is completed, and the information recommendation sequence is generated.

[0118] For example, the information recommendation sequence is generated in the manner of filling the empty sequence, Figure 4 The process of generating the information recommendation sequence according to the recommendation rules of each type of information provided by an exemplary embodiment of the present application is shown in the process diagram, Figure 4 As shown in the original sequence 410, nine item objects (i.e. recommended information) are sorted according to the recommendation ranking, assuming that a inherent rule is defined as "in a second unit recommendation window with a size of 3, no recommended information of the same type of information appears", and the judgment condition of the sum of the historical recommendation times is "the maximum number of appearance threshold parameter of the same author in 3 brushes is 5", and the temporary rule is "in a first unit recommendation window with a size of 5, no recommended information of the same type of information appears". Take the example of limiting the recommendation frequency of information, and apply it to the original recommendation sequence as shown in Figure 1 The information categories of item2 and item3 meet the above judgment condition of the sum of the historical recommendation times, that is, the positions of item2 and item3 in the sequence need to meet the temporary rule; an empty sequence 420 is created, and as the sliding window moves, each item object is filled in the empty sequence, and the following judgment process and item object position adjustment process will occur:

[0119] The first window, item2 and item3 belong to the same information category, and the subsequent object item4 meets the inherent rule, item4 is adjusted to the window, and thus all items in the second window meet the inherent rule.

[0120] The second window, item2 and item3 belong to the same information category, and the subsequent object item4 meets the inherent rule, item4 is adjusted to the window, and thus all items in the second window meet the inherent rule.

[0121] The third window, all items in the third window meet the inherent rule, but item2 and item3 do not meet the temporary rule, the subsequent objects item6 and item7 are in the same information field as item5, which does not meet the inherent rule, item8 meets the inherent rule, item8 is adjusted to the window, and the original item3 is sequentially moved, thus all items in the third window meet the inherent rule, and item2 and item3 meet the temporary rule.

[0122] The fourth window, all items in the fourth window meet the rule constraint, and no adjustment is needed, and the window continues to slide.

[0123] The fifth window, after item6 is added to the empty sequence, all items in the fifth window meet the rule constraint, and no adjustment is needed, and the window continues to slide.

[0124] The sixth window, item7 and item6 belong to the same information category, and the subsequent object item9 meets the fixed condition, item9 is added to the empty sequence, and all items in the sixth window meet the inherent rule.

[0125] The seventh window, the last window of the sequence, item7 and item6 belong to the same information category, and there is no adjustable object in the subsequent, and no processing is done, and thus the sliding window operation adjustment of the entire sequence is completed.

[0126] It should be noted that, Figure 4 The inherent rule, the temporary rule, and the determination condition of the sum of the number of historical recommendations shown are illustrative, and various rules can be set and combined based on actual needs in the application process, and the present application does not limit this.

[0127] Step 350, information recommendation is performed on each information to be recommended in the information recommendation sequence.

[0128] In the embodiment of the present application, after completing the information recommendation within the current refresh number, the temporary rule corresponding to the target category information is destroyed.

[0129] That is, the temporary rule for the target category information is only valid within the current refresh number, and at the end of the current refresh number, the temporary rule applied within the current refresh number is destroyed; in the next refresh number, the judgment process and the setting process of whether to increase the temporary rule are performed again.

[0130] After completing the information recommendation within the current refresh number, the recommendation result within the current refresh number is written into the database for subsequent refresh numbers to perform historical refresh number statistics and judgment.

[0131] To sum up, the information recommendation method provided in the embodiment of the present application, before performing information recommendation, sums the historical recommendation number of each category information within N historical refresh numbers, and in the case that the relationship between the historical recommendation number of the target category information and the corresponding recommendation number threshold satisfies the set condition, it is explained that the recommendation strategy for the target category information needs to be changed, and a corresponding temporary rule is added for the target category information within the current refresh number; then, based on the temporary rule of the target category information and the inherent rule of each category information, an information recommendation sequence within the current refresh number is generated to perform information recommendation on each to-be-recommended information according to the recommendation order indicated by the information recommendation sequence, wherein for the target category information, the recommendation frequency determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule. Through the above method, when performing information recommendation in the current refresh number, the computer device can adjust the recommendation rule in combination with the historical recommendation of each information category in the historical refresh number, and adjust the recommendation frequency of the target category information in each category information through the temporary rule, so as to realize the push control of the target category information, and also make the information recommendation sequence in the current refresh number different from the information recommendation sequence in the historical refresh number, thereby increasing the diversity of information recommendation between refresh numbers.

[0132] At the same time, when the recommendation frequency of the target category information is too high, the push of the target category information can be limited through the increased temporary rule; when the recommendation frequency of the target category information is too low, the push of the target category information can be improved through the increased temporary rule, so that the adjustment of information push is more flexible and meets the actual demand.

[0133] Figure 5 A block diagram of an information recommendation device provided by an example embodiment of the present application is shown, which can be used to execute all or part of the steps of the embodiments shown in Figure 2 or Figure 3 , such as shown in Figure 5 , the information recommendation device can include:

[0134] The number statistics module 510 is configured to count a sum of historical recommendation numbers of each type of information in N historical refresh numbers; N is a positive integer.

[0135] The rule adding module 520 is configured to add a corresponding temporary rule for the target type of information in the current refresh number, where the temporary rule is used to indicate a recommendation quantity constraint condition of the target type of information in a first unit recommendation window of the current refresh number; the target type of information is a type of information whose relationship between the sum of historical recommendation numbers and a corresponding recommendation number threshold meets a set condition.

[0136] The sequence generating module 530 is configured to generate an information recommendation sequence in the current refresh number based on the temporary rule and a corresponding inherent rule of each type of information; the inherent rule is used to indicate a maximum recommendable quantity of the corresponding type of information in a second unit recommendation window; and for the target type of information, a recommendation frequency determined based on the corresponding inherent rule is different from a recommendation frequency determined based on the corresponding temporary rule.

[0137] The information recommendation module 540 is configured to perform information recommendation on each to-be-recommended information in the information recommendation sequence.

[0138] In an optional implementation, the rule adding module 520 is configured to,

[0139] add, in the current refresh number, a corresponding temporary rule for reducing the recommendation frequency for the target type of information, in a case where the sum of historical recommendation numbers of the target type of information is greater than a first recommendation number threshold corresponding to the target type of information;

[0140] add, in the current refresh number, a corresponding temporary rule for increasing the recommendation frequency for the target type of information, in a case where the sum of historical recommendation numbers of the target type of information is less than a second recommendation number threshold corresponding to the target type of information, where an application priority of the temporary rule is higher than an application priority of the inherent rule, and the first recommendation number threshold is greater than the second recommendation number threshold.

[0141] In an optional implementation, in a case where the sum of historical recommendation numbers of the target type of information is greater than a first recommendation number threshold corresponding to the target type of information, the temporary rule is used to indicate a maximum recommendable quantity of the target type of information in a first unit recommendation window of the current refresh number.

[0142] A first recommendation proportion corresponding to the temporary rule of the target type of information is less than a second recommendation proportion corresponding to the inherent rule of the target type of information.

[0143] The recommendation proportion refers to a proportion between the maximum recommendable quantity indicated by the rule and a maximum number of to-be-recommended information that can be contained in the corresponding unit recommendation window.

[0144] In an optional implementation, the temporary rule is used to indicate a minimum recommendation quantity of the target category information in a first unit recommendation window of the current refresh quantity, in a case where a sum of historical recommendation quantities of the target category information is less than a second recommendation quantity threshold corresponding to the target category information.

[0145] In an optional implementation, the sequence generation module 530 comprises:

[0146] a sequence acquisition sub-module, configured to acquire an original recommendation sequence, the original recommendation sequence being a recommendation sequence obtained after information fine arrangement is performed on each to-be-recommended information;

[0147] a sequence generation sub-module, configured to perform sorting adjustment on each to-be-recommended information in the original recommendation sequence according to a recommendation rule of each category information, to generate an information recommendation sequence in a current refresh quantity; the each category information comprises the target category information and non-target category information, the recommendation rule corresponding to the target category information comprises an inherent rule and a temporary rule, and the recommendation rule corresponding to the non-target category information comprises an inherent rule.

[0148] In an optional implementation, each rule has a corresponding application priority.

[0149] The sequence generation sub-module is configured to generate the information recommendation sequence in the current refresh quantity based on information sorting determined based on a first rule, in a case where the information sorting determined based on the first rule conflicts with information sorting determined based on a second rule.

[0150] The first rule and the second rule are respectively recommendation rules of two categories of information, or the first rule and the second rule are different rules of a same category of information; and the application priority of the first rule is higher than the application priority of the second rule.

[0151] In an optional implementation, the apparatus further comprises:

[0152] a storage module, configured to store, in a database, historical recommendation quantities of each category of information in M historical refresh quantities; M≥N, M being a positive integer.

[0153] a deletion module, configured to delete, in a case where the historical refresh quantity is greater than M, the historical recommendation quantity of each category of information stored in the database in the earliest historical refresh quantity.

[0154] In an optional implementation, the apparatus further comprises:

[0155] A rule destroying module is configured to destroy the temporary rule corresponding to the target category information after completing the information recommendation in the current refresh number.

[0156] To sum up, the information recommendation device provided by the embodiments of the present application is used to count the sum of the historical recommendation numbers of each category of information in N historical refresh numbers before information recommendation, and in the case that the relationship between the sum of the historical recommendation numbers of the target category information and the corresponding recommendation number threshold satisfies the set condition, it is indicated that the recommendation strategy for the target category information needs to be changed, and the corresponding temporary rule is added for the target category information in the current refresh number. Then, the information recommendation sequence in the current refresh number is generated based on the temporary rule of the target category information and the inherent rule of each category of information, so as to perform information recommendation on each to-be-recommended information according to the recommendation order indicated by the information recommendation sequence, wherein the recommendation frequency of the target category information determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule. Through the above method, when the information recommendation in the current refresh number is performed, the computer device can adjust the recommendation rule in combination with the historical recommendation of each information category in the historical refresh number, and adjust the recommendation frequency of the target category information in each category of information through the temporary rule, so as to realize the push control of the target category information, and also make the information recommendation sequence in the current refresh number different from the information recommendation sequence in the historical refresh number, thereby increasing the diversity of information recommendation between refresh numbers.

[0157] Figure 6 The structural block diagram of the computer device 600 shown in an example embodiment of the present application is shown. The computer device can be implemented as a server in the above-mentioned scheme of the present application. The computer device 600 includes a central processing unit (CPU) 601, a system memory 604 including a random access memory (RAM) 602 and a read-only memory (ROM) 603, and a system bus 605 connecting the system memory 604 and the central processing unit 601. The computer device 600 also includes a mass storage device 606 for storing an operating system 609, application programs 610 and other program modules 611.

[0158] Without loss of generality, the computer readable medium can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, Erasable Programmable Read Only Memory (EPROM), Electrically-Erasable Programmable Read-Only memory (EEPROM), flash memory or other solid state memory technology, CD-ROM, Digital Versatile Disc (DVD), or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. It should be understood by those skilled in the art that computer storage media does not limit to the above-mentioned several kinds. The system memory 604 and the mass storage device 606 mentioned above can be collectively referred to as memory.

[0159] According to various embodiments of the present application, the computer device 600 can also operate in connection with a remote computer through a network such as the Internet. That is, the computer device 600 can connect to the network 608 through the network interface unit 607 connected to the system bus 605, or can be connected to other types of networks or remote computer systems (not shown) using the network interface unit 607.

[0160] The memory further includes at least one instruction, at least one program, a code set or an instruction set, which are stored in the memory, and the central processing unit 601 implements all or part of the steps of the information recommendation method shown in the various embodiments by executing the at least one instruction, at least one program, code set or instruction set.

[0161] Figure 7 A structural block diagram of a computer device 700 according to an exemplary embodiment of the present application is shown. The computer device 700 can be implemented as the terminal described above, such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart watch, a television and the like. The computer device 700 can also be referred to as a user equipment, a portable terminal, a laptop terminal, a desktop terminal and other names.

[0162] Generally, the computer device 700 includes a processor 701 and a memory 702.

[0163] In some embodiments, the computer device 700 can further optionally include a peripheral device interface 703 and at least one peripheral device. The processor 701, the memory 702 and the peripheral device interface 703 can be connected through a bus or a signal line. Each peripheral device can be connected to the peripheral device interface 703 through a bus, a signal line or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 704, a display screen 705, a camera component 706, an audio circuit 707 and a power supply 708.

[0164] In some embodiments, the computer device 700 further includes one or more sensors 709. The one or more sensors 709 include, but are not limited to, an acceleration sensor 710, a gyroscope sensor 711, a pressure sensor 712, an optical sensor 713 and a proximity sensor 714.

[0165] Those skilled in the art can understand that the structure shown in the above embodiments does not constitute a limitation on the computer device 700, and the computer device 700 can include more or fewer components than those shown in the figure, or combine certain components, or adopt a different arrangement of components. Figure 7

[0166] In an exemplary embodiment, a computer readable storage medium is also provided, in which at least one computer program is stored, the computer program being loaded and executed by a processor to implement all or part of the steps of the information recommendation method described above. For example, the computer readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk and an optical data storage device, etc.

[0167] In an exemplary embodiment, a computer readable storage medium is also provided, in which at least one computer program is stored, the computer program being loaded and executed by a processor to implement all or part of the steps of the information recommendation method described above. For example, the computer readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk and an optical data storage device, etc.

[0168] In an exemplary embodiment, a computer program product is also provided, which includes at least one computer program, the computer program being loaded and executed by a processor to implement all or part of the steps of the information recommendation method described above. Figure 2 or​Figure 3 all or part of the steps of the information recommendation method shown in any embodiment.

[0169] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0170] It should be understood that the application is not limited to the precise construction that has been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application. The scope of the application is indicated only by the appended claims.

Claims

1. An information recommendation method characterized by comprising: The method comprises: Statistics of each type of information in the sum of the respective historical recommended number of times within N historical refresh times; N is a positive integer; In the current refresh number of times, the corresponding temporary rule is added to the target type of information, and the temporary rule is used to indicate the recommended number of constraints of the target type of information in the first unit recommended window of the current refresh number of times; The target type of information is the type of information whose relationship between the sum of the historical recommended number of times and the corresponding recommended number of times threshold meets the set condition; Based on the temporary rule and the inherent rule corresponding to each type of information, an information recommendation sequence within the current refresh number of times is generated; The inherent rule is used to indicate the maximum recommended number of the corresponding type of information in the second unit recommended window; Wherein, for the target type of information, the recommendation frequency determined based on the corresponding inherent rule is different from the recommendation frequency determined based on the corresponding temporary rule; Information recommendation is performed on each to-be-recommended information in the information recommendation sequence; The temporary rule corresponding to the target type of information is added in the current refresh number of times, comprising: In the case that the sum of the historical recommended number of times of the target type of information is greater than the first recommended number of times threshold corresponding to the target type of information, the temporary rule corresponding to the target type of information is added in the current refresh number of times; Wherein, the application priority of the temporary rule is higher than the application priority of the inherent rule.

2. The method of claim 1, wherein, The temporary rule corresponding to the target type of information is added in the current refresh number of times, comprising: In the case that the sum of the historical recommended number of times of the target type of information is less than the second recommended number of times threshold corresponding to the target type of information, the temporary rule corresponding to the target type of information is added in the current refresh number of times; Wherein, the first recommended number of times threshold is greater than the second recommended number of times threshold.

3. The method according to claim 1 or 2, characterized in that, In the case that the sum of the historical recommended number of times of the target type of information is greater than the first recommended number of times threshold corresponding to the target type of information, the temporary rule is used to indicate the maximum recommended number of the target type of information in the first unit recommended window of the current refresh number of times; The first recommended proportion corresponding to the temporary rule of the target type of information is less than the second recommended proportion corresponding to the inherent rule of the target type of information; Wherein, the recommended proportion refers to the proportion between the maximum recommended number indicated by the rule and the maximum number of to-be-recommended information that can be contained in the corresponding unit recommended window.

4. The method according to claim 1 or 2, characterized in that, In the case that the sum of the historical recommended number of times of the target type of information is less than the second recommended number of times threshold corresponding to the target type of information, the temporary rule is used to indicate the minimum recommended number of the target type of information in the first unit recommended window of the current refresh number of times.

5. The method of claim 1, wherein, The temporary rule and the inherent rule corresponding to each type of information are used to generate an information recommendation sequence within the current refresh number of times, comprising: An original recommendation sequence is obtained, and the original recommendation sequence is a recommendation sequence obtained after information sorting of each to-be-recommended information. The method comprises the following steps: adjusting the order of each information to be recommended in the original recommendation sequence according to the recommendation rules of each type of information, and generating an information recommendation sequence within the current refresh number; the information recommendation sequence comprises target information and non-target information; the recommendation rules of the target information comprise inherent rules and temporary rules; and the recommendation rules of the non-target information comprise inherent rules.

6. The method of claim 5, wherein, Each rule has a corresponding application priority. The method comprises the following steps: adjusting the order of each information to be recommended in the original recommendation sequence according to the recommendation rules of each type of information, and generating an information recommendation sequence within the current refresh number; the information recommendation sequence comprises target information and non-target information; the recommendation rules of the target information comprise inherent rules and temporary rules; and the recommendation rules of the non-target information comprise inherent rules. In the case that the information order determined based on the first rule conflicts with the information order determined based on the second rule, the information order determined based on the first rule is used to generate an information recommendation sequence within the current refresh number. The first rule and the second rule are respectively the recommendation rules of two types of information, or the first rule and the second rule are different rules of the same type of information; and the application priority of the first rule is higher than the application priority of the second rule.

7. The method of claim 1, wherein, The method further comprises the following steps: Storing the historical recommendation numbers of each type of information within M historical refresh numbers in a database; M≥N, and M is a positive integer. When the historical refresh number is greater than M, deleting the historical recommendation number of each type of information stored in the database at the earliest historical refresh number.

8. The method of claim 1, wherein, After the information recommendation sequence is generated, the method further comprises the following steps: After the information recommendation within the current refresh number is completed, destroying the temporary rules corresponding to the target information.

9. An information recommendation device characterized by comprising: The device comprises: A number counting module configured to count the sum of the historical recommendation numbers of each type of information within N historical refresh numbers; N is a positive integer. A rule adding module configured to add a temporary rule corresponding to the target information within the current refresh number, wherein the temporary rule is used to indicate a recommendation number constraint condition of the target information within a first unit recommendation window of the current refresh number; and the target information is a type of information whose relationship between the sum of the historical recommendation numbers and a corresponding recommendation number threshold satisfies a set condition. A sequence generating module configured to generate an information recommendation sequence within the current refresh number based on the temporary rule and the inherent rules corresponding to each type of information; the inherent rules are used to indicate the maximum recommendable number of the corresponding type of information within a second unit recommendation window; and for the target information, a first recommendation frequency determined based on the corresponding inherent rule is different from a second recommendation frequency determined based on the corresponding temporary rule. An information recommendation module configured to perform information recommendation on each information to be recommended in the information recommendation sequence. The rule adding module is configured to add a temporary rule corresponding to the target information within the current refresh number, wherein the temporary rule is used to reduce the recommendation frequency of the target information; and the application priority of the temporary rule is higher than the application priority of the inherent rule.

10. A computer device, comprising: The computer device comprises a processor and a memory, and the memory stores at least one computer program, which is loaded and executed by the processor to implement the information recommendation method according to any one of claims 1 to 8.

11. A computer readable storage medium, characterized in that, The computer readable storage medium stores at least one computer program, which is loaded and executed by the processor to implement the information recommendation method according to any one of claims 1 to 8.

Images