Method and device for generating recommendation words

Abstract

The present application discloses a method and device for generating recommendation words, and relates to the field of information technology. The method includes: collecting historical association data between users and objects in different dimensions; determining the association weight between the user and the recommendation words of different dimensions of the object according to the historical association data; obtaining the target dimension whose association weight meets the preset conditions; and generating the recommendation words of the object for the user according to the target dimension. The present application can generate recommendation words under the target dimension that matches the user, so as to ensure that the generated recommendation words are more accurate and suitable for the user, thereby improving the selection rate of the object.

Description

Technical Field

[0001] The present application relates to the field of information technology, and in particular to a method and device for generating a recommendation. Background Art

[0002] With the rapid development of science and technology and the continuous improvement of living material standards, users have more and more demands for object selection. In order to facilitate users to select corresponding objects, it is necessary to add recommendations corresponding to the objects and display them to users so that users can quickly match the required objects according to the recommendations.

[0003] At present, object recommendations are usually generated based on the descriptions provided by the object publisher and displayed to users. However, the descriptions provided by the object publisher are mostly related to the object, but not to the user. Therefore, it is difficult to give accurate and appropriate recommendations from the user's perspective. In addition, the recommendation generated by this method has a relatively single dimension and does not attract the user's attention, resulting in low efficiency in object selection. Summary of the invention

[0004] In view of this, the present application provides a method and device for generating recommendation words, the main purpose of which is to generate recommendation words under target dimensions that match the user, so as to ensure that the generated recommendation words are more accurate and suitable for the user, thereby improving the selection rate of the object.

[0005] According to one aspect of the present application, a method for generating a recommendation is provided, the method comprising:

[0006] Collect historical association data between users and objects in different dimensions;

[0007] Determining, based on the historical association data, association weights between the user and the recommended terms of different dimensions of the object;

[0008] Obtaining a target dimension whose association weight meets a preset condition;

[0009] A recommendation of the object is generated for the user according to the target dimension.

[0010] Optionally, obtaining the target dimension whose association weight meets a preset condition includes:

[0011] Determine the basic weights of the recommendations for different dimensions;

[0012] Adding the basic weight and the associated weight to obtain a total weight between the recommendations of different dimensions of the user and the object;

[0013] The dimension whose total weight is greater than the preset weight is determined as the target dimension.

[0014] Optionally, the different dimensions include a first-level dimension, or the sum of a first-level dimension and a second-level dimension, wherein the second-level dimension is a sub-dimension corresponding to the first-level dimension.

[0015] Optionally, collecting historical association data between users and objects in different dimensions includes:

[0016] Collecting first behavior data between the user and the object in different dimensions;

[0017] Determine the target group to which the user belongs, and collect second behavior data between the target group and the object in different dimensions;

[0018] Determine similar objects corresponding to the object, and collect third behavior data between the user and the similar objects in different dimensions, and fourth behavior data between the target population and the similar objects in different dimensions;

[0019] The first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data are determined as the historical association data.

[0020] Optionally, determining the association weights between the user and the recommended terms of different dimensions of the object according to the historical association data includes:

[0021] respectively determining association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data;

[0022] The association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data are added together to obtain the association weights between the recommendations of different dimensions of the user and the object.

[0023] Optionally, generating a recommendation for the object for the user based on the target dimension includes:

[0024] Generate a recommendation word for the object for the user by using a preset recommendation word generation grammar corresponding to the target dimension.

[0025] Optionally, if the target dimension includes at least one dimension of weight, material, and taste corresponding to the object, then generating a recommendation for the object for the user using a preset recommendation generation grammar corresponding to the target dimension includes:

[0026] Obtaining raw data of at least one dimension of weight, material, and taste;

[0027] Generate a recommendation corresponding to the object for the user according to the original data.

[0028] Optionally, if the target dimension includes comment information, then generating a recommendation for the object for the user by using a preset recommendation grammar corresponding to the target dimension includes:

[0029] Obtaining comment information related to the object;

[0030] Filtering the review information using a preset positive word library to obtain filtered review information;

[0031] Using a preset grammar model to concatenate the filtered comment information to obtain a recommendation corresponding to the object; or,

[0032] The review words in the review information are located using a preset positive word library, and the located review words are concatenated to obtain a recommendation word corresponding to the object.

[0033] Optionally, if the target dimension includes at least one of ordering behavior, positive review behavior, and browsing behavior, then generating a recommendation for the object for the user using a preset recommendation generation grammar corresponding to the target dimension includes:

[0034] Performing statistics on raw data in at least one dimension of ordering behavior, reviewing behavior, and browsing behavior to obtain a first statistical result corresponding to the object;

[0035] Based on the first statistical result and a first preset grammar model, a recommendation word corresponding to the object is generated.

[0036] Optionally, if the target dimension includes at least one of regional ranking, in-store ranking and quality ranking, the generating of the recommendation words for the object for the user by using a preset recommendation word generation grammar corresponding to the target dimension includes:

[0037] Collecting statistics on the regional ranking, in-store ranking or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object;

[0038] Generate a recommendation corresponding to the object according to the second statistical result and a second preset grammar model.

[0039] Optionally, after generating a recommendation for the object for the user based on the target dimension, the method further includes:

[0040] If there are multiple target dimension recommendations for the object, the historical click volumes corresponding to the objects marked with the multiple target dimension recommendations are counted respectively;

[0041] Based on the historical click volume, the multiple target dimension recommendations are sorted, and the target dimension recommendations to be displayed are determined according to the sorting result.

[0042] Optionally, after generating a recommendation for the object for the user based on the target dimension, the method further includes:

[0043] According to the target dimension, determining whether the recommendation for the object needs to be highlighted;

[0044] If it is determined that the recommendation word of the object needs to be highlighted, a highlight mark corresponding to the recommendation word is added, and the recommendation word with the mark added is displayed.

[0045] According to another aspect of the present application, a device for generating a recommendation is provided, the device comprising:

[0046] A collection unit, used to collect historical association data between users and objects in different dimensions;

[0047] a determining unit, configured to determine, based on the historical association data, association weights between the user and the recommendation terms of different dimensions of the object;

[0048] An acquisition unit, used for acquiring a target dimension whose association weight meets a preset condition;

[0049] A generating unit is used to generate a recommendation for the object for the user based on the target dimension.

[0050] Optionally, the acquisition unit includes: a determination module and an addition module,

[0051] The determination module is used to determine the basic weights of the recommended terms for different dimensions of the object;

[0052] The adding module is used to add the basic weight and the associated weight to obtain the total weight between the recommendations of different dimensions of the user and the object;

[0053] The determination module is further used to determine the dimension whose total weight is greater than the preset weight as the target dimension.

[0054] Optionally, the different dimensions include a first-level dimension, or the sum of a first-level dimension and a second-level dimension, wherein the second-level dimension is a sub-dimension corresponding to the first-level dimension.

[0055] Optionally, the collection unit includes: a collection module and a determination module.

[0056] The collecting module is used to collect the first behavior data between the user and the object in different dimensions;

[0057] The collection module is further used to determine the target group to which the user belongs, and collect the second behavior data between the target group and the object in different dimensions;

[0058] The collection module is further used to determine similar objects corresponding to the object, and collect third behavior data between the user and the similar objects in different dimensions, and fourth behavior data between the target population and the similar objects in different dimensions;

[0059] The determination module is used to determine the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data as the historical association data.

[0060] Optionally, the determining unit includes: a determining module and an adding module,

[0061] The determination module is used to respectively determine the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data;

[0062] The adding module is used to add the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data, respectively, to obtain the association weights between the recommendations of different dimensions of the user and the object.

[0063] Optionally, the generating unit is specifically configured to generate a recommendation word for the object for the user by using a preset recommendation word generating grammar corresponding to the target dimension.

[0064] Optionally, if the target dimension includes at least one dimension of weight, material, and taste corresponding to the object, the generating unit includes: an acquisition module and a generation module,

[0065] The acquisition module is used to acquire raw data in at least one dimension of weight, material, and taste;

[0066] The generating module is used to generate a recommendation corresponding to the object for the user according to the original data.

[0067] Optionally, if the target dimension includes comment information, the generating unit includes: an acquisition module, a filtering module and a splicing module.

[0068] The acquisition module is used to acquire comment information related to the object;

[0069] The filtering module is used to filter the comment information using a preset positive word library to obtain filtered comment information;

[0070] The splicing module is used to splice the filtered comment information using a preset grammar model to obtain the recommendation words corresponding to the object; or to locate the comment words in the comment information using a preset positive word library, and splice the located comment words to obtain the recommendation words corresponding to the object.

[0071] Optionally, if the target dimension includes at least one of ordering behavior, positive review behavior and browsing behavior, the generating unit includes: a statistical module and a generating module,

[0072] The statistical module is used to collect statistics on the original data in at least one dimension of ordering behavior, positive review behavior and browsing behavior to obtain a first statistical result corresponding to the object;

[0073] The generating module is used to generate a recommendation corresponding to the object based on the first statistical result and a first preset grammar model.

[0074] Optionally, if the target dimension includes at least one of regional ranking, in-store ranking and quality ranking, the generating unit includes: a statistical module and a generating module,

[0075] The statistical module is used to collect statistics on the regional ranking, in-store ranking or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object;

[0076] The generating module is used to generate a recommendation corresponding to the object according to the second statistical result and a second preset grammar model.

[0077] Optionally, the device further comprises: a statistical unit,

[0078] The statistical unit is used for respectively counting the historical click volumes corresponding to the objects marked with the multiple target dimension recommendation words if the object has multiple target dimension recommendation words;

[0079] The determination unit is further configured to sort the multiple target dimension recommendations based on the historical click volume, and determine the target dimension recommendations to be displayed according to the sorting result.

[0080] Optionally, the device further comprises: a determination unit and a display unit,

[0081] The determination unit is used to determine whether the recommendation word of the object needs to be highlighted according to the target dimension;

[0082] The display unit is used to add a highlight mark corresponding to the recommendation word if it is determined that the recommendation word of the object needs to be highlighted, and display the recommendation word with the mark added.

[0083] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for generating the recommendation is implemented.

[0084] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned method for generating recommendation words when executing the computer program.

[0085] By means of the above technical scheme, the present application provides a method and device for generating recommendation words. Compared with the current method of generating recommendation words based on the description words provided by the publisher of the object, the present application collects the historical association data between the user and the object in different dimensions; and determines the association weights between the user and the recommendation words of different dimensions of the object based on the historical association data; at the same time, obtains the target dimension in which the association weight meets the preset conditions; and finally generates the recommendation words of the object for the user based on the target dimension, thereby collecting the historical association data between the user and the object in different dimensions, it is possible to generate the recommendation words in the target dimension that match the user, thereby ensuring that the generated recommendation words are closely associated with the user and more suitable for the user, thereby increasing the attractiveness of the object to the user and improving the selection rate of the object.

[0086] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0088] Figure 1 A schematic diagram showing a flow chart of a method for generating a recommendation provided in an embodiment of the present application;

[0089] Figure 2 A schematic diagram showing a flow chart of another method for generating a recommendation provided in an embodiment of the present application;

[0090] Figure 3 A schematic diagram of a recommendation provided in an embodiment of the present application is shown;

[0091] Figure 4 Another schematic diagram of recommendation provided in an embodiment of the present application is shown;

[0092] Figure 5 A schematic diagram of the structure of a device for generating a recommendation provided in an embodiment of the present application is shown;

[0093] Figure 6 A schematic diagram of the structure of another device for generating recommendation words provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0094] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0095] In order to solve the problem that the recommendation words generated in the prior art are not highly relevant to the user and have a relatively single dimension. This embodiment provides a method for generating recommendation words, such as Figure 1 As shown, the method includes:

[0096] Step 101: Collect historical association data between users and objects in different dimensions.

[0097] The executor of this embodiment is a client device or server device that can generate recommendation words. The client side includes applications (Application, APP) with transaction functions installed on smart terminals such as smartphones and tablets, or mini-programs, or public accounts, or WEB applications, etc.

[0098] The object may be a dish, or other types of goods, commodities or services. The different dimensions may include a primary dimension, or the sum of a primary dimension and a secondary dimension. The secondary dimension is a sub-dimension corresponding to the primary dimension. The primary dimension may include dimensions such as object attributes, user evaluation, user behavior and list ranking. The secondary dimension corresponding to the object attributes specifically includes various attribute labels of the object.

[0099] For example, if the object is a dish, and the first-level dimension corresponding to the dish is the dish attribute, then the second-level dimension corresponding to the dish includes the quality, calories, raw materials, cuisine, preparation method, taste, etc. of the dish; the second-level dimension corresponding to user behavior includes the user's ordering behavior, positive review behavior, and browsing behavior; the second-level dimension corresponding to the list ranking includes regional list data, in-store list data and other dimensions, among which the next-level dimension corresponding to the regional list data may include the sales ranking list, popularity ranking list, and positive review ranking list corresponding to the object.

[0100] For this embodiment, when a user browses an object on the client or the client recommends an object to a user, a recommendation message corresponding to the object will be generated and displayed to the user so that the user can quickly match the object he or she needs based on the recommendation message, thereby improving the object selection rate. In order to ensure that the generated recommendation message is closely related to the user, it is necessary to first collect the historical association data between the user and the object in the above-mentioned different dimensions.

[0101] The historical association data may include: the behavior data between the user and the object in different dimensions, the behavior data between the user's group and the object in different dimensions, the behavior data between the user and the same type of objects in different dimensions, and the behavior data between the user's group and the same type of objects in different dimensions, etc. These behavior data may specifically include the user's add-to-cart behavior data, repurchase behavior data, order behavior data, and click behavior data, etc., and the association data between the user and the object in different dimensions can be determined based on the above behavior data.

[0102] For example, the object may be a dish. When recommending dishes to users, the historical association data between the user and the dishes marked with dish attributes, user evaluation, user behavior, and list ranking data recommendation words are collected respectively. When specifically collecting the historical association data between the user and the dishes marked with dish attribute recommendation words, the user's purchase behavior data, repurchase behavior data, order behavior data, and click behavior data for the dishes marked with dish attribute recommendation words can be collected first, and then the user's group can be determined to which the user belongs, and the purchase behavior data, repurchase behavior data, order behavior data, and click behavior data for the dishes marked with dish attribute recommendation words are collected from this group of people, and then the category of the dish can be determined to which the dish belongs, and the purchase behavior data, repurchase behavior data, order behavior data, and click behavior data for the dishes marked with dish attribute recommendation words are collected from the user's purchase behavior data, repurchase behavior data, order behavior data, and click behavior data for the dishes marked with dish attribute recommendation words are collected from the user, and finally the purchase behavior data, repurchase behavior data, order behavior data, and click behavior data for the dishes marked with dish attribute recommendation words are collected from this group of people.

[0103] Based on the behavioral data collected above, the historical association data between the user and the dishes marked with dish attribute recommendation words can be determined. Further, according to the above method, the historical association data between the user and the dishes marked with user evaluation recommendation words, the historical association data between the user and the dishes marked with user behavior recommendation words, and the historical association data between the user and the dishes marked with list ranking data recommendation words can be determined respectively, so as to determine the association weights between the user and the recommendations of different dimensions of the dishes based on the historical association data.

[0104] Step 102: Determine the association weights between the user and the recommended terms of different dimensions of the object based on the historical association data.

[0105] For this embodiment, in order to generate the recommendation words that best match the user and ensure the relevance between the user and the generated recommendation words, it is necessary to calculate the relevance weights between the user and the recommendation words of different dimensions of the object.

[0106] Specifically, the behavioral data between the user and objects in different dimensions, the behavioral data between the user's group and objects in different dimensions, the behavioral data between the user and similar objects in different dimensions, and the behavioral data between the user's group and similar objects in different dimensions are collected; according to the size of the collected behavioral data, the association weight corresponding to each behavioral data is determined; the association weights corresponding to each behavioral data are added together to obtain the association weights between the recommendations of different dimensions between the user and the object.

[0107] For example, when determining the association weight between a user and a user behavior recommendation for a dish, first determine the association weight corresponding to the first behavior data based on the size of the first behavior data between the user and the dish marked with the user behavior recommendation.

[0108] Specifically, the corresponding association weight can be determined according to the data range of the first behavior data. For example, when the first behavior data is greater than 0 and less than 10, the association weight corresponding to the first behavior data is determined to be 1; when the first behavior data is greater than or equal to 10 and less than 20, the association weight corresponding to the first behavior data is determined to be 2; when the first behavior data is greater than or equal to 20 and less than 30, the association weight corresponding to the first behavior data is determined to be 3. If the cumulative sum of the number of purchase behaviors, repurchase behaviors, order behaviors, and click behaviors of the user in the collected first behavior data is 25, then the association weight corresponding to the first behavior data is determined to be 3. Similarly, the association weight corresponding to the second behavior data can be determined according to the size of the second behavior data between the user's group and the dish marked with the user behavior recommendation, and then the association weight corresponding to the third behavior data can be determined according to the size of the third behavior data between the user and the same dish marked with the user behavior recommendation, and finally the association weight corresponding to the fourth behavior data can be determined according to the size of the fourth behavior data between the user's group and the same dish marked with the user behavior data recommendation, and the association weight corresponding to each behavior data is added to obtain the association weight between the user and the user behavior recommendation of the dish.

[0109] Therefore, the association weight between the recommendation words of different dimensions of the user and the object can be determined in the above manner, so as to determine the target dimension that best matches the user based on the size of the association weight, so that the recommendation words generated under the target dimension are most suitable for the user and closely associated with the user.

[0110] Step 103: Obtain the target dimension whose association weight meets the preset conditions.

[0111] Among them, the target dimension may include at least one dimension. For this embodiment, after calculating the association weights between the recommended terms of different dimensions of the user and the object, the various association weights can be sorted according to the calculated association weights, and the dimensions corresponding to the association weights ranked within a preset ranking range can be determined as the target dimensions, or the dimensions corresponding to the association weights ranked highest can be determined as the target dimensions, or the dimensions corresponding to the association weights within a preset weight range can be determined as the target dimensions. This embodiment does not make any specific restrictions on the specific method of determining the target dimension.

[0112] Step 104: Generate a recommendation for the object for the user based on the target dimension.

[0113] For this embodiment, after determining the primary target dimension that matches the user, the secondary target dimension that matches the user is further determined based on the primary target dimension. Specifically, the historical association data between the user and the object in different secondary dimensions is collected, and the association weights between the user and the object's different secondary dimension recommendations are determined based on the historical association data, and then the secondary target dimensions whose association weights meet the preset conditions are obtained, and the object recommendations are generated for the user based on the secondary target dimensions.

[0114] For example, the first-level target dimension may be user behavior, and the second-level dimensions corresponding to user behavior include the user's ordering behavior, positive review behavior, and browsing behavior, respectively collecting the historical association data between the user and the object marked with ordering behavior recommendations, the historical association data between the user and the object marked with positive reviews as recommendations, and the historical association data between the user and the object marked with browsing behavior recommendations.

[0115] The historical association data specifically includes: the behavior data between the user and the object in different secondary dimensions, the behavior data between the user's group and the object in different secondary dimensions; the behavior data between the user and such objects in different secondary dimensions, and the behavior data between the user's group and such objects in different secondary dimensions. Then, based on each behavior data, the association weight between the user and the object's different secondary dimension recommendation words is determined, and the secondary target dimension whose association weight meets the preset conditions is determined. The determination process of the secondary target dimension is the same as the determination process of the primary dimension. Finally, based on the determined secondary target dimension, a recommendation word is generated for the user. This will result in different recommendations for the same object for different users, which can ensure that the generated recommendation word has a high degree of matching and association with the user.

[0116] In the process of generating recommendations based on the secondary target dimension, different secondary target dimensions have different methods for generating recommendations. For example, if the secondary target dimension is the raw materials of a dish, the dish attribute information entered by the merchant is directly obtained, and the raw material information in the dish attribute information is extracted as a recommendation to be displayed to the user.

[0117] For another example, if the secondary target dimension is positive review behavior, the review information for the dish is obtained, the user's positive review information is determined by locating the associated words, and the number of positive review information is accumulated. Based on the accumulated results and a specific grammatical model, a user behavior recommendation for the dish is generated for the user.

[0118] For another example, if the secondary target dimension is user evaluation, then the review information for the dish is obtained, and by filtering and reorganizing the review information, a user evaluation recommendation for the dish is generated for the user.

[0119] For example, if the secondary target dimension is regional list data, the regional sales ranking, regional popularity ranking and regional positive review ranking corresponding to the dishes can be counted, and based on the statistical results and a specific grammatical model, a regional list data recommendation for the dish can be generated for the user.

[0120] The present embodiment provides a method for generating recommendation words. Compared with the current method of generating recommendation words based on description words provided by the publisher of an object, the present embodiment collects historical association data between the user and the object in different dimensions; and determines the association weights between the user and the recommendation words of different dimensions of the object based on the historical association data; at the same time, obtains the target dimension in which the association weight meets the preset conditions; and finally generates the recommendation words of the object for the user based on the target dimension. By collecting the historical association data between the user and the object in different dimensions, the recommendation words in the target dimension matching the user can be generated, thereby ensuring that the generated recommendation words are closely associated with the user and more suitable for the user, thereby increasing the attractiveness of the object to the user and improving the selection rate of the object.

[0121] Further, as a refinement and extension of the above-mentioned specific implementation method, another method for generating recommendation words is provided, such as Figure 2 As shown, the method includes:

[0122] Step 201: Collect historical association data between users and objects in different dimensions.

[0123] For this embodiment, in order to determine the target dimension that best matches the object, it is necessary to collect historical association data between the user and the object in different dimensions. For the collection process of the historical association data, as an optional implementation method, step 201 specifically includes: collecting first behavior data between the user and the object in different dimensions; determining the target population to which the user belongs, and collecting second behavior data between the target population and the object in different dimensions; determining similar objects corresponding to the object, and collecting third behavior data between the user and the similar objects in different dimensions, and fourth behavior data between the target population and the similar objects in different dimensions; determining the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data as the historical association data.

[0124] For example, different dimensions are user behavior and list ranking data, which collect historical association data between users and dishes marked with user behavior recommendations, as well as historical association data between users and dishes marked with list ranking data recommendations. When specifically collecting historical association data between users and dishes marked with user behavior recommendations, first collect the user's first behavior data for dishes marked with user behavior recommendations, then collect the second behavior data of the user's group for dishes marked with user behavior recommendations, then collect the user's third behavior data for dishes marked with user behavior recommendations, and finally collect the user's fourth behavior data for dishes marked with user behavior recommendations. The first behavior data, the second behavior data, the third behavior data, and the fourth behavior data respectively include add-to-purchase behavior data, repeat purchase behavior data, ordering behavior data, and click behavior data. Then, based on at least one of the first behavior data, the second behavior data, the third behavior data, and the fourth behavior data, determine the historical association data between the user and the dishes marked with user behavior recommendations. Similarly, the historical association data between the user and the dishes marked with list ranking data can be determined in the above manner, so that based on the historical association data, the association weights between the user and recommendations of different dimensions of the dishes can be determined.

[0125] Step 202: Determine the association weights between the user and the recommended terms of different dimensions of the object based on the historical association data.

[0126] For this embodiment, in order to determine the association weights between the recommended terms of different dimensions of the user and the object, step 202 specifically includes: respectively determining the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data; adding the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data, respectively, to obtain the association weights between the recommended terms of different dimensions of the user and the object. In determining the association weights corresponding to the recommended terms of different dimensions, the association weights corresponding to the recommended terms of different dimensions can be determined based on the association weights corresponding to any behavior data, or the association weights corresponding to any of the first behavior data, the second behavior data, the third behavior data, and the fourth behavior data can be added to obtain the association weights corresponding to the recommended terms of different dimensions.

[0127] For example, when determining the association weight between a user and a dish's ranking data recommendation, first determine the association weight corresponding to the first behavior data based on the size of the first behavior data between the user and the dish marked with the regional ranking data recommendation. Specifically, the corresponding association weight can be determined based on the data range in which the first behavior data is located. For example, when the first behavior data is greater than 0 and less than 10, determine the association weight corresponding to the first behavior data as 1; when the first behavior data is greater than or equal to 10 and less than 20, determine the association weight corresponding to the first behavior data as 2; when the first behavior data is greater than or equal to 20 and less than 30, determine the association weight corresponding to the first behavior data as 3. If the cumulative sum of the user's add-to-cart behavior times, repeat purchase behavior times, order behavior times, and click behavior times in the collected first behavior data is 25, then determine the association weight corresponding to the first behavior data as 3. Similarly, the association weight corresponding to the second behavior data can be determined based on the size of the second behavior data between the user's population and the dishes marked with regional list data recommendations. Then, the association weight corresponding to the third behavior data can be determined based on the size of the third behavior data between the user and similar dishes marked with regional list data recommendations. Finally, the association weight corresponding to the fourth behavior data can be determined based on the size of the fourth behavior data between the user's population and similar dishes marked with regional list data recommendations. The association weights corresponding to each behavior data are added together to obtain the association weight between the user and the regional list data recommendation for the dish.

[0128] Step 203: Obtain the target dimension whose association weight meets the preset conditions.

[0129] For this embodiment, if the user is a new user, it is impossible to collect the historical association data of the user, and therefore it is impossible to determine the target dimension that matches the user based on the historical association data. In order to also generate corresponding recommendations for new users, when determining the target dimension, the target dimension that matches the user can be determined based on the basic weight of the object's recommendations for different dimensions, and then generate recommendations for the new user; if the user is not a new user, the target dimension that matches the user can be determined based on the calculated association weight and basic weight, and then generate corresponding recommendations for the user. Based on this, step 203 specifically includes: determining the basic weight of the object's recommendations for different dimensions; adding the basic weight and the association weight to obtain the total weight between the user and the object's recommendations for different dimensions; determining the dimension whose total weight is greater than the preset weight as the target dimension.

[0130] Specifically, when determining the target dimension corresponding to a new user, since it is impossible to collect the historical association data of the user, the historical association data between the same group of people and the object in different dimensions can be collected, and the basic weights of the recommendations for the object in different dimensions can be determined based on the historical association data corresponding to the same group of people. For example, if the new user is a fat loss user, the historical association data between the fat loss group of people and the object in different dimensions can be collected, and the basic weights of the recommendations for the object in different dimensions can be determined based on the historical association data corresponding to the fat loss group of people, and the target dimension whose basic weight meets the preset conditions can be obtained. For example, if the basic weight corresponding to the list ranking data is the largest, the list ranking data is determined as the target dimension. For another example, if the basic weights corresponding to the list ranking data and user behavior are within the preset weight range, the list ranking data and user behavior are determined as the target dimensions.

[0131] Furthermore, if the user is not a new user, the associated weight can be added to the corresponding basic weight to obtain the total weight between the user and the recommendations of different dimensions of the object. For example, the basic weight corresponding to the list ranking data and the associated weight between the user and the recommendations of the list ranking data can be added to obtain the total weight between the user and the list ranking data. Similarly, the total weight between the user and the user behavior and the total weight between the user and the product attributes can be obtained, and the dimension corresponding to the maximum total weight can be determined as the target dimension, and the corresponding recommendation is generated for the user based on the target dimension. In this way, according to the above method, not only can recommendation words be generated for old users, but also for new users.

[0132] Step 204: Generate a recommendation for the object for the user using a preset recommendation generation grammar corresponding to the target dimension.

[0133] Among them, different target dimensions correspond to different preset recommendation generation grammars. For this embodiment, after determining the primary target dimension that matches the user, it is necessary to continue to determine the secondary target dimension that matches the user. The specific determination process of the secondary target dimension is the same as the determination process of the primary target dimension, and will not be repeated again.

[0134] In a specific application scenario, if the secondary target dimension includes at least one of the weight, material, and taste dimensions corresponding to the object, then the preset recommendation word generation grammar corresponding to the target dimension is used to generate a recommendation word for the object for the user, including: obtaining original data under at least one of the weight, material, and taste dimensions; and generating a recommendation word corresponding to the object for the user based on the original data.

[0135] For example, if the first-level target dimension is the dish attribute and the second-level target dimension is the material corresponding to the dish, then the dish description information published by the merchant can be obtained, and the dish description information can be used as the original data to extract the material corresponding to the dish from the original data to generate the corresponding recommendation for the user. For example, if tomato scrambled eggs include tomatoes and eggs, the material label "tomato and eggs" can be displayed to the user as a recommendation. Furthermore, the user's historical behavior data for tomato scrambled eggs can be collected, including the number of times the user has added purchases, repurchased, ordered, and clicked on tomato scrambled eggs. By analyzing the user's historical behavior data for tomato scrambled eggs, if it is determined that when the material label "tomato" is displayed first, the user's add-on purchase, repurchase, order, and click operations are more frequent, indicating that the user has a higher degree of interest in tomatoes, then the material label "tomato" can be displayed to the user first, followed by "eggs".

[0136] In a specific application scenario, if the target secondary dimension includes comment information, the generation of the recommendation words for the object for the user through the preset recommendation word generation grammar corresponding to the target dimension includes: obtaining comment information related to the object; filtering the comment information using a preset positive word library to obtain filtered comment information; splicing the filtered comment information using a preset grammar model to obtain the recommendation words corresponding to the object; or locating the comment words in the comment information using a preset positive word library, and splicing the located comment words to obtain the recommendation words corresponding to the object. The preset positive word library includes commendatory words.

[0137] like Figure 3 As shown, the secondary target dimension is the review information, and the dish is Miao Xiang Grilled Fish. The review information related to Miao Xiang Grilled Fish is collected, and the positive review words in the review information are located using the preset positive vocabulary, and the sentences containing the positive review words are extracted, and then the extracted sentences are spliced ​​using the preset grammar model.

[0138] For example, the extracted sentences include "Miao Xiang Grilled Fish", "Delicious", and "I will order it again next time", which are combined using the preset grammar model to generate a recommendation for the user: "Miao Xiang Grilled Fish, delicious, I will order it again next time". In addition, the located positive comment words can be directly extracted and spliced. For example, the extracted positive comment words include "good words" and "very tender". After splicing the comment with the name of the dish, a recommendation is generated for the user: "Miao Xiang Grilled Fish, very delicious, very tender". In this way, the comment information recommendation can be generated in the above two ways and displayed to the user, so that the user can quickly find the required dishes according to the promotion words, thereby increasing the order rate of the dishes.

[0139] In a specific application scenario, if the secondary target dimension includes at least one of ordering behavior, positive review behavior, and browsing behavior, then the preset recommendation word generation grammar corresponding to the target dimension is used to generate a recommendation word for the object for the user, including: performing statistics on the original data under at least one dimension of ordering behavior, positive review behavior, and browsing behavior to obtain a first statistical result corresponding to the object; and generating a recommendation word corresponding to the object based on the first statistical result and a first preset grammar model.

[0140] like Figure 3 As shown, if the secondary target dimension is ordering behavior, a large number of users' ordering behavior data for the object are collected, and the ordering behavior data are counted. For example, if the number of users who have placed orders for the object within 7 days is 100, the ordering behavior recommendation phrase "100 people placed orders within 7 days" is generated for the user according to the first preset grammar model. For another example, if the secondary target dimension is favorable comment behavior, a large number of users' favorable comment behavior data for the object are collected, and the favorable comment behavior data are counted. For example, if 200 people have given favorable comments within 1 month, the favorable comment behavior recommendation phrase "200 people have given favorable comments recently" is generated for the user according to the first preset grammar model. For another example, if the secondary target dimension is user recommendation, a large number of users' recommendation behavior data for the object are collected, and the recommendation behavior data are counted. When the recommended behavior data exceeds the preset number, the comment recommendation phrase "recommended by commenting netizens" is generated for the user according to the first preset grammar model.

[0141] In a specific application scenario, if the secondary target dimension includes at least one of the dimensions of regional ranking, in-store ranking, and quality ranking, the generation of the recommendation word for the object for the user through the preset recommendation word generation grammar corresponding to the target dimension includes: performing statistics on the regional ranking, in-store ranking, or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object; and generating the recommendation word corresponding to the object according to the second statistical result and the second preset grammar model. Among them, the regional ranking includes the sales ranking, popularity ranking, and favorable comment ranking of the object in the region, and the in-store ranking includes the sales ranking, popularity ranking, and favorable comment ranking of the object in the store. The store in this embodiment is a store that can conduct online transactions through a combination of software and hardware.

[0142] Specifically, within a preset time period, the sales data, sales growth data and positive review data of the object in a certain area are counted and ranked according to the statistical results. If the ranking rank corresponding to the object is within a preset range, a corresponding recommendation is generated according to the ranking result.

[0143] For example, the sales data of all dishes in Putuo District within a month are counted, and the dishes are ranked from high to low according to the sales data. For the top 5 dishes, the corresponding regional list data recommendation is generated for the user according to their corresponding ranking results, such as Figure 4 As shown, the sales data of potato beef ranks 5th in Putuo District. According to the ranking result, a recommendation is generated for the user, "Putuo District potato beef orders rank 5th". Similarly, the sales growth data of the dishes within a week are counted and sorted from high to low according to the sales growth data. For dishes ranked within the preset rankings, according to their corresponding ranking results, corresponding regional list data recommendations are generated for users, such as "Putuo District potato beef popularity soars to No. 1". In addition, the ranking can also be based on the statistical results of the positive reviews data. For dishes ranked within the preset rankings, according to their corresponding sorting results, corresponding regional list data recommendations are generated for users, such as "Putuo District potato beef positive reviews list No. 2".

[0144] Similarly, within a preset time period, the sales data, sales growth data and positive review data of the object in the store are counted and ranked according to the statistical results. If the ranking rank of the object is within a preset range, a corresponding recommendation is generated based on the ranking result, for example, "Our store ranks first in terms of order volume", "Our store ranks second in terms of positive reviews", "Our store ranks first in terms of popularity".

[0145] Furthermore, it is also possible to determine whether the object belongs to the high-quality category based on the sales data and favorable comment data corresponding to the object. Based on the determination result, a recommendation corresponding to the object is generated and displayed to the user. Specifically, objects whose sales data is greater than the preset sales data and whose favorable comment rate is greater than the preset favorable comment rate can be screened out. If the sales ranking and favorable comment rate ranking corresponding to the object can be within the preset ranking range, the object is determined to be a high-quality category. When recommending the object to the user, a recommendation corresponding to the object can be generated, such as "high-quality pasta product". In this way, the list ranking data recommendation can be generated in the above manner, so that users can quickly match the objects they need through the list ranking data recommendation, thereby improving the selection rate of the objects.

[0146] Step 205: Determine whether the recommendation for the object needs to be highlighted based on the target dimension.

[0147] For this embodiment, in order to facilitate users to distinguish between the recommended words and other information of the object, the generated recommended words can be displayed in a different way from other information. For example, the recommended words can be displayed in the form of red text on a yellow background. In addition, in order to further attract the attention of users, the recommended words of certain dimensions need to be highlighted. For example, if the generated recommended words are regional list data recommended words, the recommended words need to be highlighted to increase user attention.

[0148] Step 206: If it is determined that the recommendation word of the object needs to be highlighted, a highlight mark corresponding to the recommendation word is added, and the marked recommendation word is displayed.

[0149] For this embodiment, if there are multiple target dimension recommendations for the object, the historical click volumes corresponding to the objects marked with multiple target dimension recommendations are counted respectively; based on the historical click volumes, the multiple target dimension recommendations are sorted, and the target dimension recommendations to be displayed are determined according to the sorting results.

[0150] For example, an object has both regional list data recommendations and in-store list data recommendations. If the historical click volumes of the object marked with regional list data recommendations and in-store list data recommendations are greater than the preset click volumes, the regional list recommendations and in-store list recommendations are sorted according to the size of the historical click volumes, and only the recommendation ranked first is displayed to the user. The regional list recommendations and in-store list recommendations can also be arranged and displayed in ranking order. Furthermore, if the historical click volumes of the object marked with regional list data recommendations and in-store list data recommendations are less than or equal to the preset click volumes, the regional list recommendations or the in-store list recommendations can be randomly selected and displayed to the user. The regional list recommendations and the in-store list recommendations can also be randomly arranged and the arranged recommendations can be displayed to the user.

[0151] Another method for generating recommendation words provided by this embodiment, compared with the current method of generating recommendation words based on the description words provided by the publisher of the object, this embodiment collects historical association data between the user and the object in different dimensions; and determines the association weights between the user and the recommendation words of different dimensions of the object based on the historical association data; at the same time, obtains the target dimension in which the association weight meets the preset conditions; and finally generates the recommendation words of the object for the user based on the target dimension, thereby collecting the historical association data between the user and the object in different dimensions, it is possible to generate recommendation words in the target dimension that match the user, thereby ensuring that the generated recommendation words are closely associated with the user and are more suitable for the user, thereby increasing the attractiveness of the object to the user and improving the selection rate of the object.

[0152] Further, as Figure 1 and Figure 2 The specific implementation of the method shown in the embodiment provides a device for generating a recommendation, such as Figure 5 As shown, the device includes: a collecting unit 31, a determining unit 32, an acquiring unit 33 and a generating unit 34.

[0153] The collecting unit 31 may be used to collect historical association data between users and objects in different dimensions.

[0154] The determining unit 32 may be configured to determine, based on the historical association data, association weights between the user and the recommended terms of different dimensions of the object.

[0155] The acquisition unit 33 may be used to acquire the target dimension whose associated weight meets a preset condition.

[0156] The generating unit 34 may be configured to generate a recommendation for the object for the user based on the target dimension.

[0157] In specific application scenarios, in order to determine the target dimension, such as Figure 6 As shown, the acquisition unit 33 includes a determination module 331 and an addition module 332 .

[0158] The determination module 331 may be used to determine the basic weights of the recommended terms for different dimensions.

[0159] The adding module 332 may be configured to add the basic weight and the associated weight to obtain a total weight between the recommendations of different dimensions of the user and the object.

[0160] The determination module 331 may also be used to determine the dimension whose total weight is greater than a preset weight as a target dimension.

[0161] Furthermore, in order to collect historical behavior data, the collecting unit 31 includes: a collecting module 311 and a determining module 312 .

[0162] The collecting module 311 may be used to collect first behavior data between the user and the object in different dimensions.

[0163] The collection module 311 may also be used to determine the target group to which the user belongs, and collect second behavior data between the target group and the object in different dimensions.

[0164] The collection module 311 may also be used to determine similar objects corresponding to the object, and collect third behavior data of the user with the similar objects in different dimensions, and fourth behavior data of the target population with the similar objects in different dimensions.

[0165] The determination module 312 may be configured to determine the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data as the historical association data.

[0166] Furthermore, in order to determine the association weights between the recommendation terms of different dimensions of the user and the object, the determination unit 32 includes: a determination module 321 and an addition module 322 .

[0167] The determination module 321 may be used to respectively determine the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data.

[0168] The adding module 322 can be used to add the association weights corresponding to the first behavior data, and / or the second behavior data, and / or the third behavior data, and / or the fourth behavior data, respectively, to obtain the association weights between the recommendations of different dimensions of the user and the object.

[0169] In a specific application scenario, the generating unit 34 may be specifically configured to generate a recommendation word for the object for the user by using a preset recommendation word generating grammar corresponding to the target dimension.

[0170] Further, if the target dimension includes at least one dimension of weight, material, and taste corresponding to the object, the generation unit 34 includes: an acquisition module 341 and a generation module 342 .

[0171] The acquisition module 341 can be used to acquire original data in at least one dimension of the weight, material, and taste.

[0172] The generating module 342 may be configured to generate a recommendation corresponding to the object for the user based on the original data.

[0173] Furthermore, if the target dimension includes comment information, the generating unit 34 includes: an acquiring module 341 , a filtering module 343 and a splicing module 344 .

[0174] The acquisition module 341 may be used to acquire comment information related to the object.

[0175] The filtering module 343 may be used to filter the comment information using a preset positive word library to obtain filtered comment information.

[0176] The splicing module 344 may be used to splice the filtered comment information using a preset grammar model to obtain a recommendation corresponding to the object.

[0177] The splicing module 344 may also be used to locate the comment words in the comment information using a preset positive word library, and to splice the located comment words to obtain the recommendation words corresponding to the object.

[0178] Further, if the target dimension includes at least one dimension of ordering behavior, positive review behavior and browsing behavior, the generating unit 34 includes: a statistical module 345 and a generating module 346 .

[0179] The statistical module 345 can be used to collect statistics on the original data in at least one dimension of ordering behavior, positive review behavior and browsing behavior to obtain a first statistical result corresponding to the object.

[0180] The generating module 346 may be configured to generate a recommendation corresponding to the object based on the first statistical result and a first preset grammar model.

[0181] Further, if the target dimension includes at least one of regional ranking, in-store ranking and quality ranking, the generating unit 34 includes: a statistical module 345 and a generating module 346 .

[0182] The statistical module 345 may be used to collect statistics on the regional ranking, in-store ranking or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object.

[0183] The generating module 346 may be configured to generate a recommendation corresponding to the object according to the second statistical result and a second preset grammar model.

[0184] Furthermore, to determine the target dimension recommendation words for display, the device also includes a statistical unit 35.

[0185] The statistical unit 35 may be configured to respectively count the historical click volumes corresponding to the objects marked with the multiple target dimension recommendation words if the object has multiple target dimension recommendation words.

[0186] The determining unit 32 may also be configured to sort the plurality of target dimension recommendations based on the historical click volume, and determine the target dimension recommendations to be displayed according to the sorting result.

[0187] Furthermore, in order to highlight the recommendation, the device further includes a determination unit 36 ​​and a display unit 37 .

[0188] The determination unit 36 ​​may be configured to determine whether the recommendation for the object needs to be highlighted according to the target dimension.

[0189] The display unit 37 may be configured to add a highlight mark corresponding to the recommendation word if it is determined that the recommendation word of the object needs to be highlighted, and display the recommendation word with the mark added.

[0190] It should be noted that for other corresponding descriptions of the functional units involved in the device for generating a recommendation provided in this embodiment, reference can be made to Figure 1 and Figure 2 The corresponding description in will not be repeated here.

[0191] Based on the above Figure 1 and Figure 2 The method shown in the embodiment of the present application is accordingly provided with a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned Figure 1 and Figure 2 The method for generating the recommendation shown.

[0192] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0193] Based on the above Figure 1 and Figure 2 The method shown, and Figure 5 and Figure 6 In order to achieve the above-mentioned purpose, the embodiment of the present application further provides a computer device, which can be a tablet computer, a smart phone, a smart watch, a smart bracelet, or other network devices, etc. The computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 and Figure 2 The method for generating the recommendation shown.

[0194] Optionally, the above-mentioned physical devices may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display screen (Display), an input unit such as a keyboard (Keyboard), etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0195] Those skilled in the art will appreciate that the client device structure provided in this embodiment does not constitute a limitation on such physical device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0196] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the two physical devices mentioned above, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to realize communication between the components inside the storage medium, and communication with other hardware and software in the information processing physical device.

[0197] Through the description of the above implementation methods, technical personnel in this field can clearly understand that by collecting historical association data between the user and the object in different dimensions; and determining the association weight between the user and the recommendation words in different dimensions of the object based on the historical association data; at the same time, obtaining the target dimension in which the association weight meets the preset conditions; finally generating the recommendation words of the object for the user based on the target dimension, thereby collecting historical association data between the user and the object in different dimensions, it is possible to generate the recommendation words in the target dimension that match the user, thereby ensuring that the generated recommendation words are closely associated with the user and more suitable for the user, thereby increasing the attractiveness of the object to the user and improving the selection rate of the object.

[0198] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple submodules.

[0199] The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be thought of by technicians in this field should fall within the scope of protection of this application.

Claims

1. A method for generating a recommendation, characterized in that: include: Collecting historical association data between users and objects in different dimensions, the historical association data includes one or more of the following: behavior data between users and objects marked with recommendation words in different dimensions, behavior data between groups to which users belong and objects marked with recommendation words in different dimensions, behavior data between users and objects of the same category marked with recommendation words in different dimensions, and behavior data between groups to which users belong and objects of the same category marked with recommendation words in different dimensions, wherein the behavior data includes the user's add-to-cart behavior data, repeat purchase behavior data, ordering behavior data, and click behavior data; Determine, based on the historical association data, an association weight between the user and the recommendation words of different dimensions of the object, wherein the association weight is determined based on the cumulative sum of the number of behaviors of the add-to-purchase behavior data, the repeat purchase behavior data, the order behavior data, and the click behavior data; Obtain the target dimension whose association weight meets the preset conditions, wherein the basic weight of the object's recommendation words for different dimensions is determined; the basic weight and the association weight are added to obtain the total weight between the user and the object's recommendation words for different dimensions; the dimension whose total weight is greater than the preset weight is determined as the target dimension; wherein the basic weight is the weight determined by the historical association data corresponding to the same group of people as the new user; A recommendation of the object is generated for the user according to the target dimension.

2. The method according to claim 1, characterized in that The different dimensions include a primary dimension, or the sum of a primary dimension and a secondary dimension, wherein the secondary dimension is a sub-dimension corresponding to the primary dimension.

3. The method according to claim 1, characterized in that Generating a recommendation for the object for the user according to the target dimension includes: Generate a recommendation word for the object for the user by using a preset recommendation word generation grammar corresponding to the target dimension.

4. The method according to claim 3, characterized in that If the target dimension includes at least one dimension of weight, material, and taste corresponding to the object, then generating a recommendation for the object for the user by using a preset recommendation generation grammar corresponding to the target dimension includes: Obtaining raw data of at least one dimension of weight, material, and taste; Generate a recommendation corresponding to the object for the user according to the original data.

5. The method according to claim 3, characterized in that: If the target dimension includes comment information, then generating a recommendation for the object for the user by using a preset recommendation grammar corresponding to the target dimension includes: Obtaining comment information related to the object; Filtering the review information using a preset positive word library to obtain filtered review information; Using a preset grammar model to concatenate the filtered comment information to obtain a recommendation corresponding to the object; or, The review words in the review information are located using a preset positive word library, and the located review words are concatenated to obtain a recommendation word corresponding to the object.

6. The method according to claim 3, characterized in that If the target dimension includes at least one of ordering behavior, reviewing behavior, and browsing behavior, then generating a recommendation for the object for the user by using a preset recommendation generation grammar corresponding to the target dimension includes: Performing statistics on raw data in at least one dimension of ordering behavior, reviewing behavior, and browsing behavior to obtain a first statistical result corresponding to the object; Based on the first statistical result and a first preset grammar model, a recommendation word corresponding to the object is generated.

7. The method according to claim 3, characterized in that If the target dimension includes at least one of regional ranking, in-store ranking and quality ranking, generating a recommendation for the object for the user by using a preset recommendation generation grammar corresponding to the target dimension includes: Collecting statistics on the regional ranking, in-store ranking or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object; Generate a recommendation corresponding to the object according to the second statistical result and a second preset grammar model.

8. The method according to claim 1, characterized in that: After generating a recommendation for the object for the user based on the target dimension, the method further includes: If there are multiple target dimension recommendations for the object, the historical click volumes corresponding to the objects marked with the multiple target dimension recommendations are counted respectively; Based on the historical click volume, the multiple target dimension recommendations are sorted, and the target dimension recommendations to be displayed are determined according to the sorting result.

9. The method according to claim 1, characterized in that: After generating a recommendation for the object for the user based on the target dimension, the method further includes: According to the target dimension, determining whether the recommendation for the object needs to be highlighted; If it is determined that the recommendation word of the object needs to be highlighted, a highlight mark corresponding to the recommendation word is added, and the recommendation word with the mark added is displayed.

10. A device for generating a recommendation, characterized in that: include: A collecting unit, configured to collect historical association data between a user and an object in different dimensions, wherein the historical association data includes one or more of behavior data between the user and an object marked with recommendation words in different dimensions, behavior data between a group of people to which the user belongs and an object marked with recommendation words in different dimensions, behavior data between the user and objects of the same type marked with recommendation words in different dimensions, and behavior data between a group of people to which the user belongs and objects of the same type marked with recommendation words in different dimensions, wherein the behavior data includes the user's add-to-cart behavior data, repeat purchase behavior data, ordering behavior data, and click behavior data; a determination unit, configured to determine, based on the historical association data, an association weight between the user and the recommendation terms of different dimensions of the object, wherein the association weight is determined based on a cumulative sum of the number of behaviors of the add-to-purchase behavior data, the repeat purchase behavior data, the order behavior data, and the click behavior data; An acquisition unit is used to acquire the target dimension whose associated weight meets the preset conditions, wherein the acquisition unit includes: a determination module and an addition module, wherein the determination module is used to determine the basic weight of the object's recommendation terms for different dimensions; the addition module is used to add the basic weight and the associated weight to obtain the total weight between the user and the object's recommendation terms for different dimensions; the determination module is further used to determine the dimension whose total weight is greater than the preset weight as the target dimension, wherein the basic weight is the weight determined by the historical association data corresponding to the same group of people as the new user; A generating unit is used to generate a recommendation for the object for the user based on the target dimension.

11. The device according to claim 10, characterized in that The different dimensions include a primary dimension, or the sum of a primary dimension and a secondary dimension, wherein the secondary dimension is a sub-dimension corresponding to the primary dimension.

12. The device according to claim 10, characterized in that The generating unit is specifically configured to generate a recommendation word for the object for the user by using a preset recommendation word generating grammar corresponding to the target dimension.

13. The device according to claim 12, characterized in that If the target dimension includes at least one dimension of weight, material, and taste corresponding to the object, the generating unit includes: an acquisition module and a generation module, The acquisition module is used to acquire raw data in at least one dimension of weight, material, and taste; The generating module is used to generate a recommendation corresponding to the object for the user according to the original data.

14. The device according to claim 12, characterized in that If the target dimension includes comment information, the generating unit includes: an acquisition module, a filtering module and a splicing module. The acquisition module is used to acquire comment information related to the object; The filtering module is used to filter the comment information using a preset positive word library to obtain filtered comment information; The splicing module is used to splice the filtered comment information using a preset grammar model to obtain the recommendation words corresponding to the object; or to locate the comment words in the comment information using a preset positive word library, and splice the located comment words to obtain the recommendation words corresponding to the object.

15. The device according to claim 12, characterized in that If the target dimension includes at least one of ordering behavior, positive review behavior and browsing behavior, the generating unit includes: a statistical module and a generating module, The statistical module is used to collect statistics on the original data in at least one dimension of ordering behavior, positive review behavior and browsing behavior to obtain a first statistical result corresponding to the object; The generating module is used to generate a recommendation corresponding to the object based on the first statistical result and a first preset grammar model.

16. The device according to claim 12, characterized in that If the target dimension includes at least one of regional ranking, in-store ranking and quality ranking, the generation unit includes: a statistical module and a generation module, The statistical module is used to collect statistics on the regional ranking, in-store ranking or quality ranking of the object within a preset time period to obtain a second statistical result corresponding to the object; The generating module is used to generate a recommendation corresponding to the object according to the second statistical result and a second preset grammar model.

17. The device according to claim 10, characterized in that The device further comprises: a statistical unit, The statistical unit is used for respectively counting the historical click volumes corresponding to the objects marked with the multiple target dimension recommendation words if the object has multiple target dimension recommendation words; The determination unit is further configured to sort the multiple target dimension recommendations based on the historical click volume, and determine the target dimension recommendations to be displayed according to the sorting result.

18. The device according to claim 10, characterized in that The device further comprises: a determination unit and a display unit, The determination unit is used to determine whether the recommendation word of the object needs to be highlighted according to the target dimension; The display unit is used to add a highlight mark corresponding to the recommendation word if it is determined that the recommendation word of the object needs to be highlighted, and display the recommendation word with the mark added.

19. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

20. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 9 is implemented.

Images