Information recommendation method and device

Abstract

The present invention discloses an information recommendation method and device, and relates to the field of computer technology. The method comprises: in response to a preset trigger condition, determining whether the amount of historical behavior data of a user on a network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, determining the score values ​​of multiple candidate items; screening out items to be recommended from the multiple candidate items according to the score values, and pushing the information of the items to be recommended to the user. Through the above steps, the recommendation effect of the recommendation system can be improved, and the problem that the system recommendation effect is poor or even unable to make recommendations in the case of new users or less user behavior data is solved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an information recommendation method and device. Background Art

[0002] With the development of Internet technology, information recommendation on online platforms is becoming more and more common. For example, on e-commerce platforms, recommendation system functions have emerged to help users quickly find the items they want.

[0003] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art: Most existing recommendation systems make recommendations based on user historical behaviors. When encountering new users or when there is relatively little user behavior data, the recommendation effect of existing recommendation systems is often very poor. In addition, since old users have no behavior on many items, it is difficult for the recommendation system to make recommendations based on their preferences for these items and categories. Summary of the invention

[0004] In view of this, the present invention provides an information recommendation method and device, which can improve the recommendation effect of the recommendation system and solve the problem that the system recommendation effect is poor or even unable to make recommendations when there are new users or relatively little user behavior data.

[0005] To achieve the above objective, according to one aspect of the present invention, an information recommendation method is provided.

[0006] The information recommendation method of the present invention includes: in response to a preset trigger condition, determining whether the amount of historical behavior data of a user on a network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, determining the scoring values ​​of multiple candidate items; screening out items to be recommended from the multiple candidate items according to the scoring values, and pushing the information of the items to be recommended to the user.

[0007] Optionally, determining the scoring values ​​of multiple candidate items includes: querying a storage module according to the identification of the candidate items to obtain the scoring values ​​of the corresponding candidate items; wherein the scoring values ​​of the candidate items are obtained in advance according to the following processing method: calculating the original scores of the candidate items according to the values ​​of multiple attributes of the candidate items; comparing the original scores of the candidate items with a preset threshold, and selecting a corresponding normalization model according to the comparison result; and normalizing the original scores of the candidate items based on the corresponding normalization model to obtain the scoring values ​​of the candidate items.

[0008] Optionally, the threshold is determined according to the following method: determining the distribution of original scores of items in the same category; and determining the threshold corresponding to the category according to the distribution of original scores of items in the same category.

[0009] Optionally, comparing the original score of the candidate item with a preset threshold and selecting a corresponding normalization model based on the comparison result includes: comparing the candidate item with a threshold corresponding to the category to which it belongs; when the original score of the candidate item is greater than or equal to the threshold, selecting a first normalization model; when the original score of the candidate item is less than the threshold, selecting a second normalization model.

[0010] Optionally, calculating the original score of the candidate item based on the values ​​of multiple attributes of the candidate item includes: determining a first weight and a second weight for each of the multiple attributes; calculating a comprehensive weight of the attributes based on the first weight and the second weight; and calculating the original score of the candidate item based on the values ​​of the multiple attributes of the candidate item and the comprehensive weight of each attribute.

[0011] Optionally, the method further includes: when the amount of historical behavior data of the user on the network platform is sufficient, determining items preferred by the user based on the historical behavior data of the user, and pushing information of the items preferred by the user to the user.

[0012] To achieve the above objective, according to another aspect of the present invention, an information recommendation device is provided.

[0013] The information recommendation device of the present invention includes: a judgment module, which is used to judge whether the data volume of the user's historical behavior data on the network platform is insufficient in response to a preset trigger condition; a first recommendation module, which is used to determine the scoring values ​​of multiple candidate items when the data volume of the user's historical behavior data on the network platform is insufficient; and is also used to filter out items to be recommended from the multiple candidate items according to the scoring values, and push the information of the items to be recommended to the user.

[0014] Optionally, the first recommendation module determines the score values ​​of multiple candidate items, including: the first recommendation module queries the storage module according to the identification of the candidate item to obtain the score value of the corresponding candidate item; wherein the score value of the candidate item is obtained in advance according to the following processing method: calculating the original score of the candidate item according to the values ​​of multiple attributes of the candidate item; comparing the original score of the candidate item with a preset threshold, and selecting a corresponding normalization model according to the comparison result; and normalizing the original score of the candidate item based on the corresponding normalization model to obtain the score value of the candidate item.

[0015] To achieve the above objective, according to another aspect of the present invention, an electronic device is provided.

[0016] The electronic device of the present invention comprises: one or more processors; and a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the information recommendation method of the present invention.

[0017] To achieve the above objective, according to yet another aspect of the present invention, a computer readable medium is provided.

[0018] The computer-readable medium of the present invention stores a computer program, and when the program is executed by a processor, the information recommendation method of the present invention is implemented.

[0019] An embodiment of the above invention has the following advantages or beneficial effects: by responding to a preset trigger condition, determining whether the amount of historical behavior data of a user on the network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, determining the score values ​​of multiple candidate items, screening out items to be recommended from the multiple candidate items according to the score values, and pushing the information of the items to be recommended to the user. These steps can improve the recommendation effect of the recommendation system and solve the problem that the system recommendation effect is poor or even unable to make recommendations when there are new users or less user behavior data.

[0020] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0022] Figure 1 is a schematic diagram of the main flow of the information recommendation method according to the first embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of the main flow of the information recommendation method according to the second embodiment of the present invention;

[0024] Figure 3a is a schematic diagram of BI score distribution of items in a category according to an embodiment of the present invention;

[0025] Figure 3b is a schematic diagram of BI score distribution of items in another category according to an embodiment of the present invention;

[0026] Figure 4a This is one of the schematic diagrams of the effect after the BI scores of items in a certain category are normalized according to an embodiment of the present invention;

[0027] Figure 4bThis is the second schematic diagram of the effect of normalizing the BI scores of items in a certain category according to an embodiment of the present invention;

[0028] Figure 5 is a schematic diagram of main modules of an information recommendation device according to a third embodiment of the present invention;

[0029] Figure 6 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0030] Figure 7 It is a schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0032] It should be pointed out that the various embodiments of the present invention and the technical features in the embodiments may be combined with each other without affecting the implementation of the present invention.

[0033] Figure 1 FIG. 1 is a schematic diagram of the main flow of the information recommendation method according to the first embodiment of the present invention. Figure 1 As shown, the information recommendation method of the embodiment of the present invention includes:

[0034] Step S101: In response to a preset trigger condition, determine whether the amount of historical behavior data of the user on the network platform is insufficient.

[0035] Exemplarily, the preset trigger condition may be detecting a certain user behavior. For example, detecting that a user enters a certain page on a network platform, or detecting that a user scrolls the screen on a certain page, or detecting that a user stays on a certain page for a long time, etc. In addition, the preset trigger condition may also be that the current time meets the time requirements for scheduled recommendation or periodic recommendation, or other trigger conditions. After the preset trigger condition is met, the processing flow of the information recommendation method of the embodiment of the present invention may be executed.

[0036] In an optional implementation, the data volume information of the user's historical behavior data on the network platform may be obtained first; then the data volume of the user's historical behavior data on the network platform is compared with a preset threshold; if the data volume of the user's historical behavior data on the network platform is greater than or equal to the preset threshold, it is confirmed that the user's historical behavior data on the network platform is sufficient; if the data volume of the user's historical behavior data on the network platform is less than the preset threshold, it is confirmed that the data volume of the user's historical behavior data on the network platform is insufficient. The user's historical behavior data on the network platform may be the user's browsing, searching, clicking, and / or ordering behavior data on the network platform (such as a shopping platform). In addition, other methods may be used to determine whether the data volume of the user's historical behavior data on the network platform is insufficient without affecting the implementation of the present invention.

[0037] Step S102: when the amount of historical behavior data of the user on the network platform is insufficient, determining rating values ​​of a plurality of candidate items.

[0038] Among them, the score value can be used to measure the sales of the item and reflect the activity of the item. In this sense, it can also be called an "activity evaluation value". In an optional implementation of step S102, the corresponding score value can be calculated in real time based on the attribute data of each candidate item. Exemplarily, the attribute data used in calculating the score value of the candidate item may include one or more of the candidate item's click volume, number of comments, good review rate, bad review rate, 3-day average sales volume, 7-day average sales volume, 3-day average sales amount, 7-day average sales amount, PV (page view) conversion rate, UV (unique visitor number) conversion rate, return rate, offline discount and other indicators.

[0039] In another optional implementation, the scoring values ​​of the items on the network platform may be calculated in advance based on the attribute data of the candidate items and stored. In this way, in step S102, the scoring values ​​of multiple candidate items may be directly obtained by querying the storage module.

[0040] Step S103: Filter out items to be recommended from the multiple candidate items according to the score values, and push information of the items to be recommended to the user.

[0041] In an optional implementation, M (e.g., 5, 10 or other values) candidate items with the largest score values ​​may be selected as items to be recommended, and information about these items to be recommended may be pushed to the user. In another optional implementation, candidate items with score values ​​greater than a preset threshold may be selected as items to be recommended, and information about these items to be recommended may be pushed to the user.

[0042] In an embodiment of the present invention, by responding to a preset trigger condition, it is determined whether the amount of historical behavior data of a user on the network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, the score values ​​of multiple candidate items are determined, and items to be recommended are screened out from the multiple candidate items according to the score values, and the information of the items to be recommended is pushed to the user. These steps can improve the recommendation effect of the recommendation system and solve the problem that the system recommendation effect is poor or even unable to make recommendations when there are new users or relatively little user behavior data.

[0043] Figure 2 FIG. 2 is a schematic diagram of the main flow of the information recommendation method according to the second embodiment of the present invention. Figure 2 As shown, the information recommendation method of the embodiment of the present invention includes:

[0044] Step S201: In response to a preset trigger condition, the user's historical behavior data on the network platform is obtained.

[0045] For example, the preset trigger condition may be detecting that a user enters a certain page on a network platform, or detecting that a user scrolls the screen on a certain page, or detecting that a user stays on a certain page for a long time, etc. In addition, the preset trigger condition may also be that the current time meets the time requirements of scheduled recommendation or periodic recommendation, or other trigger conditions. After the preset trigger condition is met, the processing flow of the information recommendation method of the embodiment of the present invention may be executed.

[0046] In this step, the historical behavior data of the user on the network platform can be queried based on the user identifier (such as the user's login account on the network platform, or the user's mobile phone number, etc.). The historical behavior data of the user on the network platform can be the user's browsing, searching, clicking, adding to cart (i.e., adding to shopping cart), and / or ordering behavior data on the network platform in the recent period of time (such as the past year or the past six months, etc.).

[0047] Step S202: Determine whether the amount of historical behavior data is insufficient.

[0048] In an optional implementation, the amount of the user's historical behavior data on the network platform can be compared with a preset data volume threshold; if the amount of the user's historical behavior data on the network platform is greater than or equal to the preset data volume threshold, it is confirmed that the user's historical behavior data on the network platform is sufficient; if the amount of the user's historical behavior data on the network platform is less than the preset data volume threshold, it is confirmed that the amount of the user's historical behavior data on the network platform is insufficient. In addition, other methods can also be used to determine whether the amount of the user's historical behavior data on the network platform is insufficient without affecting the implementation of the present invention.

[0049] Further, when the amount of the historical behavior data of the user on the network platform is insufficient, step S203 is executed; when the amount of the historical behavior data of the user on the network platform is sufficient, step S205 is executed.

[0050] Step S203: query the storage module according to the identifier of the candidate item to obtain the score value of the corresponding candidate item.

[0051] The score value can be used to measure the sales of the item and reflect the activity of the item. In this sense, it can also be called "activity evaluation value" or "heat evaluation value". In the embodiment of the present invention, the score value of the item on the network platform can be calculated and stored in advance based on the attribute data of the candidate item. In this way, in step S203, the score values ​​of multiple candidate items can be directly obtained by querying the storage module.

[0052] Furthermore, the score of the candidate item can be obtained according to the preferred processing method shown in step A1 to step A3:

[0053] Step A1: Calculate the original score of the candidate item according to the values ​​of multiple attributes of the candidate item.

[0054] The original score of the candidate item can be used to measure the sales of the item and reflect the activity of the item, which can also be called an activity score or a BI (Business Intelligence) score. In an optional implementation, step A1 specifically includes determining a first weight and a second weight for each of the multiple attributes; calculating a comprehensive weight of the attribute based on the first weight and the second weight; and calculating the original score of the candidate item based on the values ​​of the multiple attributes of the candidate item and the comprehensive weight of each attribute.

[0055] Among them, the first weight and the second weight are two weight values ​​set from different angles. Among them, the first weight is the weight assignment for the attributes of the item. In specific implementation, the first weight of each attribute can be set according to the business recommendation experience and the online test effect. For example, the attributes of the item can be divided into multiple categories first, such as browsing, evaluation, sales, amount, conversion rate, return rate, and discount. Then, the corresponding first weight is set for the attributes under each category. For example, the first weight corresponding to the attributes under the "browsing" category can be set to 25, the first weight corresponding to the attributes under the "sales" category can be set to 30, and the first weights under the categories of "evaluation", "amount", "conversion rate", "return rate", and "discount" can be set to 15, 15, 10, -5, and 10 respectively.

[0056] The second weight reflects the correlation between the item attribute and the order quantity. In an optional example, the second weight can be expressed by the correlation coefficient between the item attribute and the order quantity. Further, the correlation coefficient between the item attribute and the order quantity can be calculated according to the following formula:

[0057]

[0058] In the formula, r(X,Y) represents the correlation coefficient between the attribute X of the item and the order quantity Y, Cov(X,Y) represents the covariance between X and Y, Var(X) represents the variance of X, and Var(Y) represents the variance of Y. Taking the browsing behavior attribute as an example, X is the number of user browsing times, and the number of browsing times of all users constitutes a random variable. Y is the user's purchase situation, which also constitutes a random variable. By calculating the covariance and variance of the browsing behavior attribute and the user's purchase situation, the correlation coefficient between the two can be calculated.

[0059] Furthermore, after obtaining the first weight and the second weight of the item attribute, the comprehensive weight of the item attribute can be calculated based on the first weight and the second weight. For example, the first weight and the second weight of the same item attribute can be multiplied together to obtain the comprehensive weight of the item attribute. Next, the original score of the candidate item can be calculated based on the values ​​of multiple attributes of the candidate item and the comprehensive weight of each attribute. For example, the comprehensive weight of the attribute can be used as a weighting coefficient to perform weighted summation on the values ​​of multiple attributes of the candidate item to obtain the original score of the candidate item.

[0060] Step A2: compare the original score of the candidate item with a preset threshold (the threshold may be called a "heat threshold", "activity threshold", or "BI score threshold"), and select a corresponding normalization model based on the comparison result.

[0061] The inventor of the present invention has found through analysis that: the BI scores of the head items (i.e., the items with the highest sales) and the tail items (i.e., the items with the lowest sales) are very different. If the items are recommended directly based on the BI scores, it may affect the performance and accuracy of the recommendation system. At the same time, the difference in BI scores between different categories is also relatively large, which results in that when items of different categories are mixed and sorted according to the BI scores, items of some categories are always ranked in front, while items of other categories are always ranked in the back, and thus the items of the categories ranked in the back are always not recommended, affecting the performance of the recommendation system. In view of this, in the embodiment of the present invention, the BI scores are normalized through steps A2 and A3.

[0062] Exemplarily, step A2 may specifically include: comparing the candidate item with a threshold value corresponding to the category to which it belongs; when the original score of the candidate item is greater than or equal to the threshold value, selecting a first normalization model; when the original score of the candidate item is less than the threshold value, selecting a second normalization model.

[0063] The first normalization model and the second normalization model together constitute a segmented normalization model. Further, the segmented normalization model can be specifically set as:

[0064]

[0065] Among them, y represents the normalized BI score (i.e., the original score), represents the first normalized model, represents the second normalized model, bi represents the BI score, and α represents the preset threshold.

[0066] Preferably, the threshold value can be determined in the following manner: determine the distribution of the original scores of items under the same category; determine the threshold value corresponding to the category according to the distribution of the original scores of the items under the same category. Taking the third-level category of thermal underwear as an example, the BI score of each product under the category can be calculated first, and then the distribution of the BI score of each product under the category can be statistically analyzed. For example, a BI score distribution graph can be drawn with the BI score of each product under the category as the horizontal axis and the number of items as the vertical axis, and then the threshold value corresponding to the category can be determined according to the BI score distribution graph. For example, for the BI score distribution graph of the third-level category of thermal underwear, the BI score of the 90% quantile (i.e., 90% of the items under the category) differs by an order of magnitude from the BI score of the 99% quantile, and the BI scores of the 90% quantile are very concentrated, so the BI score corresponding to the 90% quantile can be used as the threshold value of the category.

[0067] Step A3: normalize the original scores of the candidate items based on the corresponding normalization model to obtain the scoring values ​​of the candidate items.

[0068] In this step, if the selected normalization model is the first normalization model, the BI scores of the candidate items are normalized according to the first normalization model; if the selected normalization model is the second normalization model, the BI scores of the candidate items are normalized according to the second normalization model.

[0069] In the embodiment of the present invention, by selecting a segmented normalization model to normalize the BI score, it is possible to ensure the discrimination while avoiding the situation where the BI score of long-tail products is too low, thereby ensuring the recommendation effect of long-tail products. In addition, compared with the normalization method of "directly dividing the BI score by the maximum BI score under the category", the problem that the normalized BI score of long-tail products is still low and the discrimination of the BI score of long-tail products is still poor in the normalization method is solved, and the normalization effect is improved, which in turn helps to improve the effect of subsequent recommendation of item information based on the normalized BI score.

[0070] Step S204: Filter out items to be recommended from the multiple candidate items according to the score values, and push information of the items to be recommended to the user.

[0071] In an optional implementation, M (e.g., 5, 10 or other values) candidate items with the largest score values ​​may be selected as items to be recommended, and information about these items to be recommended may be pushed to the user. In another optional implementation, candidate items with score values ​​greater than a preset threshold may be selected as items to be recommended, and information about these items to be recommended may be pushed to the user.

[0072] Step S205: Determine the items preferred by the user according to the historical behavior data, and push information of the items preferred by the user to the user.

[0073] When the amount of historical behavior data of a user is sufficient, the historical behavior data of the user may be analyzed to determine the items preferred by the user so as to make personalized recommendations to the user.

[0074] In the embodiment of the present invention, the above steps can improve the recommendation effect of the recommendation system, solving the problem that the system has poor recommendation effect or even cannot make recommendations when there are new users or relatively little user behavior data.

[0075] The following combination Figure 3a , 3b 4a and 4b further illustrate the distribution of BI scores under the same category and the normalization effect of Bi scores in the embodiments of the present invention. Figure 3aIt is a schematic diagram of BI score distribution of items in a category (specifically, the category of thermal underwear) according to an embodiment of the present invention. Figure 3b FIG. 4 is a schematic diagram of the BI score distribution of items in another category (specifically, the category of computer accessories) according to an embodiment of the present invention. Figure 3a As shown in the figure, in the third-level category of thermal underwear, the BI score of the 90% quantile (i.e., 90% of the items in this category) is 44.452, the BI score of the 95% quantile is 91.5292, and the BI score of the 99% quantile is 414.9577; Figure 3b As shown, in the third-level category of computer accessories, the BI score of the 90% percentile is 89.0096, the BI score of the 95% percentile is 178.3715, and the BI score of the 99% percentile is 1093.2305.

[0076] Figure 4a Specifically, it is a schematic diagram of the normalized BI scores of the entire range under a certain category. Figure 4b Specifically, it is a schematic diagram of the normalization of BI scores below the 90% quantile in a certain category. Figure 4a , 4b As shown in the figure, the normalized BI score ensures the differentiation while avoiding the situation where the BI score of long-tail products is too low, thereby ensuring the recommendation effect of long-tail products.

[0077] Figure 5 FIG. 1 is a schematic diagram of the main modules of the information recommendation device according to the third embodiment of the present invention. Figure 5 As shown, the information recommendation device 500 according to the embodiment of the present invention includes: a judgment module 501 and a first recommendation module 502 .

[0078] The judgment module 501 is used to judge whether the amount of historical behavior data of the user on the network platform is insufficient in response to a preset trigger condition.

[0079] For example, the preset trigger condition may be detecting that a user enters a certain page on a network platform, or detecting that a user scrolls the screen on a certain page, or detecting that a user stays on a certain page for a long time, etc. In addition, the preset trigger condition may also be that the current time meets the time requirements of scheduled recommendation or periodic recommendation, or other trigger conditions. After the preset trigger condition is met, the processing flow of the information recommendation method of the embodiment of the present invention may be executed.

[0080] In an optional implementation, the judgment module 501 may first obtain the data volume information of the user's historical behavior data on the network platform; then the judgment module 501 compares the data volume of the user's historical behavior data on the network platform with the preset data volume threshold; if the data volume of the user's historical behavior data on the network platform is greater than or equal to the preset data volume threshold, the judgment module 501 confirms that the user's historical behavior data on the network platform is sufficient; if the data volume of the user's historical behavior data on the network platform is less than the preset data volume threshold, the judgment module 501 confirms that the data volume of the user's historical behavior data on the network platform is insufficient. Among them, the user's historical behavior data on the network platform can be the user's browsing, searching, clicking, and / or ordering behavior data on the network platform (such as a shopping platform). In addition, without affecting the implementation of the present invention, the judgment module 501 can also use other methods to determine whether the data volume of the user's historical behavior data on the network platform is insufficient.

[0081] The first recommendation module 502 is used to determine the rating values ​​of multiple candidate items when the amount of historical behavior data of the user on the network platform is insufficient.

[0082] Among them, the score value can be used to measure the sales of the item and reflect the activity of the item. In this sense, it can also be called "activity evaluation value" or "heat evaluation value". In an optional embodiment, the first recommendation module 502 can calculate the corresponding score value in real time based on the attribute data of each candidate item. Exemplarily, the attribute data used in calculating the score value of the candidate item may include one or more of the candidate item's click volume, number of comments, good review rate, bad review rate, 3-day average sales volume, 7-day average sales volume, 3-day average sales amount, 7-day average sales amount, PV (page view) conversion rate, UV (unique visitor number) conversion rate, return rate, offline discount and other indicators.

[0083] Preferably, the first recommendation module 502 can calculate the score value of the candidate item according to the following method: the first recommendation module 502 calculates the original score of the candidate item according to the values ​​of multiple attributes of the candidate item; the first recommendation module 502 compares the original score of the candidate item with a preset threshold, and selects a corresponding normalization model according to the comparison result; the first recommendation module 502 normalizes the original score of the candidate item based on the corresponding normalization model to obtain the score value of the candidate item.

[0084] Among them, the first recommendation module 502 compares the original score of the candidate item with a preset threshold, and selects the corresponding normalization model according to the comparison result, which may specifically include: comparing the candidate item with the threshold corresponding to the category to which it belongs (the threshold may be called "heat threshold", "activity threshold", or "BI score threshold"); when the original score of the candidate item is greater than or equal to the threshold, selecting the first normalization model; when the original score of the candidate item is less than the threshold, selecting the second normalization model.

[0085] The first normalization model and the second normalization model together constitute a segmented normalization model. Further, the segmented normalization model can be specifically set as:

[0086]

[0087] Among them, y represents the normalized BI score (i.e., heat evaluation value), represents the first normalized model, represents the second normalized model, bi represents the BI score, and α represents the threshold.

[0088] Preferably, the threshold α can be determined in the following manner: determine the distribution of the original scores of the items in the same category; determine the threshold α corresponding to the category according to the distribution of the original scores of the items in the same category. Taking the third-level category of thermal underwear as an example, the BI score of each product in the category can be calculated first, and then the distribution of the BI score of each product in the category can be statistically analyzed. For example, a BI score distribution graph can be drawn with the BI score of each product in the category as the horizontal axis and the number of items as the vertical axis, and then the threshold α corresponding to the category can be determined according to the BI score distribution graph. For example, for the BI score distribution graph of the third-level category of thermal underwear, the BI score of the 90% quantile (i.e., 90% of the items in the category) differs by an order of magnitude from the BI score of the 99% quantile, and the BI scores of the 90% quantile are very concentrated, so the BI score corresponding to the 90% quantile can be used as the threshold α of the category.

[0089] In the embodiment of the present invention, by selecting a segmented normalization model to normalize the BI score, it is possible to ensure the discrimination while avoiding the situation where the BI score of long-tail products is too low, thereby ensuring the recommendation effect of long-tail products. In addition, compared with the normalization method of "directly dividing the BI score by the maximum BI score under the category", the problem that the normalized BI score of long-tail products is still low and the discrimination of the BI score of long-tail products is still poor in the normalization method is solved, and the normalization effect is improved, which in turn helps to improve the effect of subsequent recommendation of item information based on the normalized BI score.

[0090] In another optional implementation, the ratings of the items on the network platform may be calculated and stored in advance based on the attribute data of the candidate items. In this way, the first recommendation module 502 may directly obtain the ratings of the multiple candidate items by querying the storage module.

[0091] The first recommendation module 502 is further configured to filter out items to be recommended from the multiple candidate items according to the scoring values, and push information of the items to be recommended to the user.

[0092] In an optional embodiment, the first recommendation module 502 may select M (e.g., 5, 10 or other numerical values) candidate items with the largest score values ​​as items to be recommended, and push the information of these items to be recommended to the user. In another optional embodiment, the first recommendation module 502 may also select candidate items with score values ​​greater than a preset threshold (the preset threshold may be referred to as a "score threshold" to distinguish it from a data volume threshold and an activity threshold) as items to be recommended, and push the information of these items to be recommended to the user.

[0093] Furthermore, the information recommendation device of the embodiment of the present invention may further include: a second recommendation module, configured to determine items preferred by the user according to the historical behavior data, and push information of the items preferred by the user to the user.

[0094] Specifically, when the amount of historical behavior data of the user is sufficient, the second recommendation module can analyze the historical behavior data of the user to determine the items preferred by the user, so as to make personalized recommendations to the user.

[0095] In an embodiment of the present invention, a judgment module responds to a preset trigger condition to determine whether the amount of historical behavior data of a user on the network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, a first recommendation module determines the score values ​​of multiple candidate items, and selects items to be recommended from the multiple candidate items according to the score values, and pushes the information of the items to be recommended to the user, which can improve the recommendation effect of the recommendation system and solve the problem that the system recommendation effect is poor or even unable to make recommendations in the case of new users or relatively little user behavior data.

[0096] Figure 6 An exemplary system architecture 600 is shown to which the information recommendation method or information recommendation device according to the embodiment of the present invention can be applied.

[0097] like Figure 6As shown, system architecture 600 may include terminal devices 601, 602, 603, network 604 and server 605. Network 604 is used to provide a medium for communication links between terminal devices 601, 602, 603 and server 605. Network 604 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0098] The user can use the terminal devices 601, 602, 603 to interact with the server 605 through the network 604 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 601, 602, 603, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0099] The terminal devices 601 , 602 , and 603 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0100] The server 605 may be a server that provides various services, such as a backend management server that provides support for shopping websites browsed by users using terminal devices 601, 602, and 603. The backend management server may respond to a user's trigger operation and recommend item information to the user according to a preset recommendation algorithm.

[0101] It should be noted that the information recommendation method provided in the embodiment of the present invention may be executed by the server 605 , and accordingly, the information recommendation device is generally disposed in the server 605 .

[0102] It should be understood that Figure 6 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0103] Reference below Figure 7 , which shows a schematic diagram of the structure of a computer system 700 of an electronic device suitable for implementing an embodiment of the present invention. Figure 7 The computer system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0104] like Figure 7As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage part 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the system 700 are also stored. The CPU 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0105] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage section 708 as needed.

[0106] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the central processing unit (CPU) 701, the above-mentioned functions defined in the system of the present invention are executed.

[0107] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0108] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0109] The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor. For example, they may be described as: a processor includes an acquisition module and a first recommendation module. The names of these modules do not, in some cases, constitute a limitation on the modules themselves. For example, the acquisition module may also be described as a "module for acquiring user behavior data."

[0110] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device executes the following process: in response to a preset trigger condition, determine whether the amount of historical behavior data of the user on the network platform is insufficient; when the amount of historical behavior data of the user on the network platform is insufficient, determine the score values ​​of multiple candidate items; filter out the items to be recommended from the multiple candidate items according to the score values, and push the information of the items to be recommended to the user.

[0111] According to the technical solution of the embodiment of the present invention, the recommendation effect of the recommendation system can be improved, and the problem that the system recommendation effect is poor or even unable to make recommendations when there are new users or relatively little user behavior data is solved.

[0112] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An information recommendation method, characterized in that: The method comprises: In response to a preset trigger condition, determining whether the amount of historical behavior data of the user on the network platform is insufficient; When the amount of historical behavior data of the user on the network platform is insufficient, determining the ratings of multiple candidate items; wherein the ratings are used to measure the sales of items and reflect the activity of items; Filter out items to be recommended from the multiple candidate items according to the score values, and push information of the items to be recommended to the user.

2. The method according to claim 1, characterized in that Determining the scoring values ​​of the plurality of candidate items includes: A storage module is queried according to the identifier of the candidate item to obtain a score value of the corresponding candidate item; wherein the score value of the candidate item is obtained in advance according to the following processing method: an original score of the candidate item is calculated according to the values ​​of multiple attributes of the candidate item; the original score of the candidate item is compared with a preset threshold, and a corresponding normalization model is selected according to the comparison result; and the original score of the candidate item is normalized based on the corresponding normalization model to obtain the score value of the candidate item.

3. The method according to claim 2, characterized in that The threshold is determined as follows: Determine the distribution of original scores of items in the same category; and determine a threshold corresponding to the category according to the distribution of original scores of items in the same category.

4. The method according to claim 3, characterized in that The comparing the original score of the candidate item with a preset threshold and selecting a corresponding normalization model according to the comparison result includes: The candidate item is compared with a threshold value corresponding to the category to which it belongs; when the original score of the candidate item is greater than or equal to the threshold value, a first normalization model is selected; when the original score of the candidate item is less than the threshold value, a second normalization model is selected.

5. The method according to claim 2, characterized in that: The calculating the original score of the candidate item according to the values ​​of the multiple attributes of the candidate item includes: Determine a first weight and a second weight of each of the multiple attributes; calculate a comprehensive weight of the attributes based on the first weight and the second weight; and calculate an original score of the candidate item based on the values ​​of the multiple attributes of the candidate item and the comprehensive weight of each attribute.

6. The method according to claim 1, characterized in that The method further comprises: When the amount of historical behavior data of the user on the network platform is sufficient, the items preferred by the user are determined according to the historical behavior data of the user, and the information of the items preferred by the user is pushed to the user.

7. An information recommendation device, characterized in that: The device comprises: A judgment module, used to judge whether the amount of historical behavior data of the user on the network platform is insufficient in response to a preset trigger condition; The first recommendation module is used to determine the rating values ​​of multiple candidate items when the amount of historical behavior data of the user on the network platform is insufficient; it is also used to filter out the items to be recommended from the multiple candidate items according to the rating values, and push the information of the items to be recommended to the user; wherein the rating values ​​are used to measure the sales of items and reflect the activity of items.

8. The device according to claim 7, characterized in that The first recommendation module determines the score values ​​of the plurality of candidate items including: The first recommendation module queries the storage module according to the identifier of the candidate item to obtain the score value of the corresponding candidate item; wherein the score value of the candidate item is obtained in advance according to the following processing method: calculating the original score of the candidate item according to the values ​​of multiple attributes of the candidate item; comparing the original score of the candidate item with a preset threshold, and selecting a corresponding normalization model according to the comparison result; normalizing the original score of the candidate item based on the corresponding normalization model to obtain the score value of the candidate item.

9. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

10. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images