AI-based interactive car recommendation methods, systems, and media
Through the interactive automotive quality recommendation method based on artificial intelligence, user needs are obtained, accessories information are judged, car recommendation groups and weight values are obtained, information is processed using learning models, and orderly recommendations are generated, which solves the problem of human recommendations neglecting personalized needs. Improved the accuracy and targetedness of recommendations.
Patent Information
- Application Number
- CN202410514415.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-26
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-04-26
AI Technical Summary
Existing automotive quality recommendations mainly rely on manual recommendations, which easily ignore users' personalized needs, resulting in reduced recommendation accuracy.
Adopt an interactive automotive quality recommendation method based on artificial intelligence. By obtaining the user's car purchase needs, we can determine whether the demand information is accessories information, obtain the car recommendation group and weight value, use the trained learning model to process the quality information and commission value, and generate the Sequential recommendation information to ensure the accuracy and personalization of recommendations.
It improves the accuracy of recommended high-quality automobiles, reduces the influence of salespeople's subjective awareness, can better respond to users' personalized needs, and enhances the targetedness and effectiveness of recommendations.
Smart Images

Figure CN118350838B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of premium automobile recommendation, and in particular to an interactive premium automobile recommendation method, system and medium based on artificial intelligence. Background Art
[0002] In recent years, as competition in the automobile market has become increasingly fierce, the profits of vehicle manufacturers' sales business have gradually declined, and the profit from selling a car is much lower than expected. Relatively speaking, the profit of premium automobiles is more lucrative. Therefore, in order to increase sales profits, stores gradually tend to sell premium automobiles.
[0003] At present, stores mainly sell premium automotive products through manual recommendations by salespeople. For example, salespeople recommend cars to users based on their own sales experience, so that salespeople can promptly understand the user's interest in each recommended vehicle. However, this method not only has high requirements on the salesperson's work experience, but also easily substitutes the salesperson's own subjective consciousness into the sales, ignoring the user's personalized needs, thereby reducing the accuracy of recommending premium automotive products to users. Summary of the invention
[0004] In order to improve the accuracy of recommending premium automobile products to users, the embodiments of the present application provide an interactive premium automobile product recommendation method, system and medium based on artificial intelligence.
[0005] In a first aspect, this embodiment provides an interactive car boutique recommendation method based on artificial intelligence, the method comprising:
[0006] Acquire the user's car purchase demand, wherein the car purchase demand includes at least one demand information;
[0007] Determine whether each piece of demand information is accessory information, and if not, obtain a corresponding car recommendation group from a preset database based on the demand information, wherein each car recommendation group includes a number of recommended car models;
[0008] Obtaining a weight value corresponding to each car recommendation group, and determining an initial recommendation value for each recommended car model based on the weight value and the corresponding car recommendation group;
[0009] Based on the car purchase demand, the boutique information and corresponding boutique commission value of each recommended car model are obtained, and the trained learning model is used to process each boutique information and the corresponding boutique commission value and the initial recommendation value to obtain ordered recommendation information for recommending boutique cars to users.
[0010] In some of the embodiments, obtaining the user's car purchase demand includes:
[0011] Obtaining historical browsing information of the user on the platform, and dividing the historical browsing information by attributes to obtain a plurality of historical browsing sub-information, wherein each historical browsing sub-information corresponds to a unique attribute, and each historical browsing sub-information includes historical attribute browsing sub-information corresponding to at least one time period;
[0012] Demand information corresponding to each attribute is determined based on the historical attribute browsing sub-information corresponding to the attribute, and all demand information corresponding to the same user is determined as the car purchase demand of the user.
[0013] In some embodiments, the preset database stores a plurality of storage units, each storage unit corresponds to a uniquely determined attribute, and each storage unit stores each car model and the corresponding attribute parameter, and obtaining the corresponding car recommendation group from the preset database based on the demand information includes:
[0014] Acquire a demand attribute corresponding to the demand information, and determine a corresponding demand storage unit in the preset database based on the demand attribute;
[0015] Selecting target attribute parameters that meet the demand information in the demand storage unit, and determining the car model corresponding to each target attribute parameter as a recommended car model that meets the demand information;
[0016] It is determined whether the demand attribute is an ordered attribute. If so, the corresponding recommended car models are packaged in ascending or descending order according to the target attribute parameter to obtain a car recommendation group.
[0017] In some embodiments, determining the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group includes:
[0018] Determine whether the car recommendation group belongs to the ordered attribute respectively. If so, derive a derived weight value corresponding to each recommended car model in the car recommendation group based on the weight value corresponding to the car recommendation group, and give the derived weight value to the weight value of the recommended car model in the car recommendation group.
[0019] If not, assigning a weight value corresponding to the car recommendation group to each recommended car model in the car recommendation group;
[0020] The weight values given to each recommended car model are added up respectively to obtain an initial recommendation value of the recommended car model.
[0021] In some embodiments, each recommended car model corresponds to at least one boutique information, and obtaining the boutique information and the corresponding boutique commission value of each recommended car model based on the car purchase demand includes:
[0022] Determine whether the car purchase demand includes the demand information of accessories, and if not, match a number of recommended accessories groups for each recommended car model;
[0023] If included, determining a recommended accessory group corresponding to each recommended automobile model based on the accessory information;
[0024] Each recommended car model and each corresponding recommended accessories group constitute a high-quality information;
[0025] The car commission and the parts commission included in each boutique information are obtained, and the car commission and the parts commission corresponding to the same boutique information are added together to obtain the boutique commission value corresponding to the boutique information.
[0026] In some embodiments, the trained learning model includes a first learning model and a second learning model, and the use of the trained learning model to process each boutique information and the corresponding boutique commission value and initial recommendation value includes:
[0027] Processing the boutique commission value and the initial recommendation value using the first learning model to obtain a revised recommendation value for each car model that represents a balance between commission and user demand;
[0028] The modified recommendation value and the boutique information are processed using a second learning model to obtain ordered recommendation information for recommending premium automobile products to the user.
[0029] In some of the embodiments, the actual automobile boutique selected by the user and the position sequence number of the actual automobile boutique in the ordered recommendation information are obtained;
[0030] Determine whether the position sequence number falls within a preset qualified sequence number range, and if so, retain the learning model;
[0031] If it does not fall into the range, the information difference between the actual automobile boutique and the ordered recommendation information is obtained, and the learning model is adjusted based on the information difference.
[0032] In some embodiments, adjusting the learning model based on the information difference includes:
[0033] The information difference is sent to a first learning model, and the first learning model is adjusted based on the information difference.
[0034] In a second aspect, this embodiment provides an interactive automobile boutique recommendation system based on artificial intelligence, the system comprising: an information acquisition module, a processing module, a calculation module and a recommendation module; wherein:
[0035] The information acquisition module is used to acquire the user's car purchase demand, wherein the car purchase demand includes at least one demand information;
[0036] The processing module is used to determine whether each piece of demand information is accessory information, and if not, obtain a corresponding car recommendation group from a preset database based on the demand information, wherein each car recommendation group includes a number of recommended car models;
[0037] The calculation module is used to obtain a weight value corresponding to each car recommendation group, and determine an initial recommendation value for each recommended car model based on the weight value and the corresponding car recommendation group;
[0038] The recommendation module is used to obtain the boutique information and corresponding boutique commission value of each recommended car model based on the car purchase demand, and use the trained learning model to process each boutique information and corresponding boutique commission value and initial recommendation value to obtain ordered recommendation information for recommending automobile boutiques to users.
[0039] In a third aspect, this embodiment provides a computer-readable storage medium having stored thereon a computer program that can be run on a processor, and when the computer program is executed by the processor, the interactive automobile boutique recommendation method based on artificial intelligence as described in the first aspect is implemented.
[0040] By adopting the above method, the present application first obtains the user's car purchase demand, wherein the car purchase demand includes at least one demand information. Then, it is judged whether each demand information is accessory information. If not, the corresponding car recommendation group is obtained from the preset database based on the demand information, wherein each car recommendation group includes several recommended car models; the weight value corresponding to each car recommendation group is obtained, and the initial recommendation value of each recommended car model is determined based on the weight value and the corresponding car recommendation group. In this way, the degree of compliance of each recommended car model determined above with the non-accessory information in the user's car purchase demand is obtained in a targeted manner, which plays a certain guiding and reference role in the subsequent final determination of the recommended information that meets the user's car purchase demand. Then, based on the car purchase demand, the boutique information and corresponding boutique commission value of each recommended car model are obtained, and the trained learning model is used to process each boutique information and the corresponding boutique commission value and the initial recommendation value to obtain the ordered recommendation information for recommending car boutiques to users. Therefore, the premium information of each recommended car model obtained based on the user's car purchasing needs and the corresponding premium commission value and initial recommendation value are substituted into the trained learning model for processing to obtain ordered recommendation information that can recommend premium cars to users, and the information is sent to the salesperson, so that the salesperson can make recommendations to users based on the ordered recommendation information without having to substitute his or her own subjective consciousness, and can respond to the user's personalized needs, thereby improving the accuracy of recommending premium cars to users. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 This is a flow chart of an interactive car boutique recommendation method based on artificial intelligence provided in an embodiment of the present application.
[0042] Figure 2 This is a flow chart of obtaining a corresponding car recommendation group from a preset database based on demand information provided by this embodiment.
[0043] Figure 3 This is a flow chart of determining the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group provided by this embodiment.
[0044] Figure 4 This is a flow chart for obtaining premium information and corresponding premium commission value for each recommended car model based on car purchase needs provided by this embodiment.
[0045] Figure 5 This is a framework diagram of an interactive automobile boutique recommendation system based on artificial intelligence provided in an embodiment of the present application. DETAILED DESCRIPTION
[0046] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. However, it should be understood by those of ordinary skill in the art that the present application can be implemented without these details. For those of ordinary skill in the art, it is obvious that various changes can be made to the embodiments disclosed in the present application, and without departing from the principles and scope of the present application, the general principles defined in the present application can be applied to other embodiments and application scenarios. Therefore, the present application is not limited to the embodiments shown, but conforms to the broadest scope consistent with the scope claimed for protection in the present application.
[0047] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.
[0048] Figure 1 : is a flow chart of an interactive car boutique recommendation method based on artificial intelligence provided by an embodiment of the present application. Figure 1 As shown, the interactive automobile boutique recommendation method based on artificial intelligence includes the following steps:
[0049] Step S100, obtaining the user's car purchase demand, wherein the car purchase demand includes at least one demand information.
[0050] The above-mentioned car purchase demand represents some demand information of users for car purchase. Each demand information included in the car purchase demand corresponds to a uniquely determined attribute. For example, a car purchase demand may include at least one of the following demand information: demand information of price attributes, demand information of energy consumption attributes, demand information of vehicle model attributes, demand information of power attributes, demand information of brand attributes, demand information of car film attributes, demand information of foot pad attributes, demand information of seat cover attributes, demand information of charging pile attributes, demand information of maintenance attributes, demand information of configuration attributes, etc. Car film attributes, foot pad attributes, seat cover attributes, charging pile attributes, and maintenance attributes all belong to the attribute of accessories.
[0051] Among them, obtaining the user's car purchase needs includes the following steps:
[0052] Step S101, obtain the user's historical browsing information on the platform, and divide the historical browsing information by attributes to obtain a number of historical browsing sub-information, wherein each historical browsing sub-information corresponds to a unique attribute, and each historical browsing sub-information includes historical browsing sub-information corresponding to at least one time period.
[0053] Step S102 , determining demand information corresponding to each attribute based on the historical attribute browsing sub-information corresponding to each attribute, and determining all demand information corresponding to the same user as the user's car purchase demand.
[0054] The method for recommending premium automobiles disclosed in the embodiment of the present application is applied to a platform available to any user, such as a mobile terminal such as a mobile phone, a tablet, a laptop, or a desktop computer. Users can use the platform when they have the idea, intention, or inclination to buy a car, and view the information about the vehicle provided by the platform by inputting or clicking, wherein the same user needs to input relevant information representing his or her identity each time he or she uses the platform, so that when it is necessary to recommend premium automobiles to the user, the user's historical browsing information on the platform can be obtained by viewing all the information about the vehicle viewed by the corresponding user on the platform, or all the information about the vehicle viewed by the user on the platform in the log.
[0055] The historical browsing information for the same user includes historical browsing sub-information of at least one attribute, and each historical browsing sub-information includes historical attribute browsing sub-information corresponding to at least one time period. Keyword recognition technology or voice recognition technology can be used in combination with information extraction technology to target the historical browsing information of this user, so as to obtain several historical attribute browsing sub-information in this historical browsing sub-information, and each historical attribute browsing sub-information corresponds to a time period. By viewing each historical attribute browsing sub-information, the historical attribute browsing sub-information belonging to the same attribute is put together, so as to obtain several historical browsing sub-information, and each historical attribute browsing sub-information contained in the same historical browsing sub-information belongs to the same attribute, but belongs to different time periods. For example, a user viewed SUV-type electric-powered cars with a price range of A to B through the platform in the first time period, and viewed SUV-type oil-powered cars with a price range of A to C through the platform in the second time period. Then the user's historical browsing information on the platform includes electric-powered SUV cars with prices between A and B and gasoline-powered SUV cars with prices between A and C. There are three historical browsing sub-information. The first historical browsing sub-information is cars with prices between A and B and cars with prices between A and C, and its attribute is price; the second historical browsing sub-information is SUV cars, and its attribute is car model; the third historical browsing sub-information is electric-powered cars and gasoline-powered cars, and its attribute is power.
[0056] By checking the historical browsing sub-information corresponding to each attribute, the historical attribute browsing sub-information corresponding to the attribute is obtained, and then the historical attribute browsing sub-information corresponding to the attribute is arranged in order according to the time period from far to near from the present time, and it is judged whether the historical attribute browsing sub-information at the back is opposite to the historical attribute browsing sub-information at the front. If not, all the attribute browsing sub-information corresponding to the attribute is determined as the corresponding demand information. If so, it is judged in turn whether the historical attribute browsing sub-information before the historical attribute browsing sub-information at the back is opposite to the historical attribute browsing sub-information at the back. If not, it is judged whether the historical attribute browsing sub-information at the front two is opposite to the historical attribute browsing sub-information at the back until the opposite result is obtained, and the above judgment is stopped, and the number of judgments is determined. If the number is not less than the preset number, the historical attribute browsing sub-information at the back is determined as the corresponding demand information. If the number is less than the preset number, all the attribute browsing sub-information corresponding to the attribute is determined as the corresponding demand information.
[0057] Finally, all the demand information corresponding to the same user is regarded as the user's car purchase demand. In this way, the user's car purchase demand can be directly obtained only through the user's browsing on the platform, without the need to use other additional methods to obtain the user's car purchase demand, which reduces the communication cost with the user and reduces the disturbance to the user, so that the user can actively inform the seller of his or her car purchase demand.
[0058] Step S200 , determining whether each piece of demand information is accessory information, and if not, obtaining a corresponding automobile recommendation group from a preset database based on the demand information, wherein each automobile recommendation group includes a number of recommended automobile models.
[0059] Since accessory information needs to be matched according to the car, the car will have certain restrictions on the recommendation of accessory information. Therefore, it is necessary to first obtain the recommended car models that can be recommended based on the non-accessory information in the demand information. After obtaining the user's car purchase needs, you can check whether the attributes corresponding to each demand information belong to the category of accessory attributes to determine whether each demand information is accessory information. If the attribute corresponding to the demand information belongs to the category of accessory attributes, then the demand information is accessory information; if the attribute corresponding to the demand information does not belong to the category of accessory attributes, then the demand information is not accessory information. For example, the demand information of car film attributes, the demand information of foot pad attributes, the demand information of seat cover attributes, the demand information of charging pile attributes, and the demand information of maintenance attributes all belong to accessory information.
[0060] In the case where the demand information is for accessories, the demand information may not be processed, that is, a corresponding car recommendation group will not be obtained from the preset database based on the demand information. In the case where the demand information is not for accessories, a corresponding car recommendation group needs to be obtained from the preset database based on the demand information. Among them, each car recommendation group includes several recommended car models. The preset database stores several storage units, each storage unit corresponds to a uniquely determined attribute, and each storage unit stores each car model and the corresponding attribute parameters. For example, the preset database stores storage units for price attributes, energy consumption attributes, vehicle attributes, power attributes, brand attributes, car film attributes, foot pad attributes, seat cover attributes, charging pile attributes, etc., which contain storage units corresponding to the attributes of each user's car purchase demand on the market. Each car model and the corresponding price are stored in the price storage unit, and each car model and the corresponding energy consumption are stored in the energy consumption storage unit, and so on. Storage units with different attributes have no intersection in the storage space, so that each storage unit stores each car model and the corresponding attribute parameters. If a new car is newly launched on the market, then each storage unit in the preset database will add the car model and corresponding attribute parameters of the new car accordingly. If the car is discontinued and out of stock, then each storage unit in the preset database will delete the car model and corresponding attribute parameters of the car accordingly.
[0061] Figure 2 This is a flow chart of obtaining a corresponding car recommendation group from a preset database based on demand information provided by this embodiment. Figure 2 As shown, obtaining the corresponding car recommendation group from the preset database based on the demand information includes the following steps:
[0062] Step S201: acquiring demand attributes corresponding to demand information, and determining corresponding demand storage units in a preset database based on the demand attributes.
[0063] Step S202 : selecting target attribute parameters that meet the demand information in the demand storage unit, and determining the car model corresponding to each target attribute parameter as a recommended car model that meets the demand information.
[0064] Step S203, determining whether the demand attribute is an ordered attribute, and if so, packaging the corresponding recommended car models in ascending or descending order of the target attribute parameter to obtain a car recommendation group.
[0065] Whenever a corresponding car recommendation group is obtained based on a demand information, the attribute to which the demand information belongs is first determined as the corresponding demand attribute, and then the storage unit in the preset database is selected to obtain the corresponding car recommendation group through the obtained demand attribute, that is, the demand storage unit corresponding to the demand information is determined.
[0066] Each storage unit is either an ordered storage unit or an unordered storage unit. Among them, whether the storage unit is an ordered storage unit or an unordered storage unit can be determined according to the attributes corresponding to the storage unit. Among them, price attributes, energy consumption attributes, power attributes, and maintenance attributes are ordered attributes, and vehicle model attributes, brand attributes, car film attributes, foot mat attributes, seat cover attributes, and charging pile attributes are unordered attributes. For an ordered storage unit, when the storage unit stores the car model and the corresponding attribute parameters, the car model and the corresponding attribute parameters are stored in ascending or descending order of the attribute parameters. For an unordered storage unit, when the storage unit stores the car model and the corresponding attribute parameters, the car model and the corresponding attribute parameters are stored in a random order.
[0067] After obtaining the demand storage unit corresponding to the demand information, if the demand storage unit is an ordered storage unit, then the demand information is compared with the direction from the first attribute parameter to the last attribute parameter in the demand storage unit to find two critical attribute parameters that can meet the demand information, then all attribute parameters located at these two critical attribute parameters are target attribute parameters that meet the demand information. If the demand storage unit is an unordered storage unit, then the demand information is compared with each attribute parameter in the demand storage unit one by one to obtain the target attribute parameters that meet the demand information in the demand storage unit. Among them, the car model corresponding to each target attribute parameter is the recommended car model that meets the demand information.
[0068] Whether the demand attribute is an ordered attribute can be determined by checking whether the demand storage unit is an ordered attribute. If the demand storage unit is an ordered attribute, then the demand attribute also belongs to an ordered attribute. If the demand storage unit is an unordered attribute, then the demand attribute also belongs to an unordered attribute. In the case where the demand attribute belongs to an ordered attribute, the corresponding recommended car models can be sorted and packaged in ascending or descending order according to the target attribute parameters to obtain the car recommendation group corresponding to the demand information, that is, all the attribute parameters located at the two critical attribute parameters obtained above are directly used as the car recommendation group corresponding to the demand information. In the case where the demand attribute does not belong to an ordered attribute, the position of each target attribute parameter can be arbitrarily determined to obtain the car recommendation group corresponding to the demand information. In this way, each storage unit corresponding to each attribute is provided in the preset database. Although the memory occupied by the preset database will increase, since the respective storage units are established by using the attribute division method, it is convenient to quickly obtain the corresponding car recommendation group from the preset database based on the demand information, and in the case of new cars being launched or discontinued, it is convenient to adjust and maintain the preset database.
[0069] Step S300, obtaining a weight value corresponding to each car recommendation group, and determining an initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group.
[0070] The above-mentioned weight value represents the score that can be attached to the recommended car model in the car recommendation group corresponding to the car recommendation group of the attribute. A weight value is marked in each storage unit in the preset database, and the weight value marked by the storage unit corresponding to the same attribute can be used as the weight value corresponding to the car recommendation group, so as to obtain the weight value corresponding to each car recommendation group. Then, the initial recommendation value of each recommended car model is determined based on the weight value and the corresponding car recommendation group. Among them, the initial recommendation value represents the recommendation value corresponding to each car obtained only based on the user's non-accessory demand information. The larger the initial recommendation value, the more the corresponding recommended car model meets the user's non-accessory demand information. Figure 3 1 is a flow chart of determining the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group provided by this embodiment. Figure 3 As shown, determining the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group includes the following steps:
[0071] Step S301, determine whether the car recommendation group belongs to the ordered attribute. If so, derive the derived weight value corresponding to each recommended car model in the car recommendation group based on the weight value corresponding to the car recommendation group, and give the derived weight value to the weight value of the recommended car model in the car recommendation group.
[0072] Step S302: If not, give the weight value corresponding to the car recommendation group to each recommended car model in the car recommendation group.
[0073] Step S303 , respectively adding up the weight values given to each recommended car model to obtain an initial recommendation value of the recommended car model.
[0074] By checking the attribute corresponding to each car recommendation group, it can be determined whether the car recommendation group belongs to the ordered attribute. If the attribute corresponding to the car recommendation group does not belong to the ordered attribute, then the car recommendation group does not belong to the ordered attribute. If the attribute corresponding to the car recommendation group belongs to the ordered attribute, then the car recommendation group belongs to the ordered attribute.
[0075] In the case where the car recommendation group belongs to an ordered attribute, in order to be able to more specifically distinguish the degree to which different recommended car models meet the demand information corresponding to the attribute, it is also possible to derive according to the weight value corresponding to the car recommendation group to obtain a derived weight value corresponding to each recommended car model. Specifically, the weight value can be reduced in a fixed value to obtain a number of derived weight values, wherein the smallest derived weight value is not less than a preset percentage of the weight value. The weight value can also be reduced in an equal ratio to obtain a number of derived weight values, wherein the equal ratio represents that the ratio between the difference between the attribute parameters corresponding to each two adjacent recommended car models and the difference between the corresponding derived weight values is equal. The smallest derived weight value is not less than a preset percentage of the weight value. Thus, the obtained several derived weight values are sequentially given to the weight value of the corresponding recommended car model in the car recommendation group, and each recommended car model corresponds to a uniquely determined derived weight value.
[0076] When the car recommendation group does not have an ordered attribute, it is impossible to determine which recommended car model better meets the demand information in the above manner. At this time, the weight value corresponding to the car recommendation group is given to each recommended car model in the car recommendation group.
[0077] Finally, all weight values given to each recommended car model are added together to obtain the initial recommended value of each recommended car model. In this way, the degree of compliance of each recommended car model determined in step S200 with the non-accessory information in the user's car purchase needs can be obtained in a targeted manner, which plays a certain guiding role in the subsequent final determination of recommended information that meets the user's car purchase needs.
[0078] Step S400, based on the car purchase demand, obtain the boutique information and corresponding boutique commission value of each recommended car model, use the trained learning model to process each boutique information and corresponding boutique commission value, and initial recommendation value to obtain ordered recommendation information for recommending car boutiques to users.
[0079] In line with the principle of not only recommending premium cars that meet users' car purchase needs as much as possible, but also earning a certain degree of commission, after obtaining the initial recommendation value of each recommended car model, it is necessary to further combine the premium information of each recommended car model and the corresponding premium commission value to obtain the ordered recommendation information of premium cars that can be recommended to users. Among them, each recommended car model corresponds to at least one premium information. Figure 4 This is a flow chart of obtaining the premium information and corresponding premium commission value of each recommended car model based on car purchase needs provided by this embodiment. Figure 4 As shown, obtaining the boutique information and corresponding boutique commission value of each recommended car model based on the car purchase demand includes the following steps:
[0080] Step S401 , determining whether the car purchase demand includes demand information for accessories information, and if not, matching a plurality of recommended accessory groups for each recommended car model.
[0081] Step S402, if included, determines a recommended accessory group corresponding to each recommended automobile model based on the accessory information.
[0082] Step S403: Each recommended automobile model and each corresponding recommended accessory group are combined into a set of high-quality information.
[0083] Step S404, obtaining the automobile commission and the parts commission contained in each boutique information, and adding the automobile commission and the parts commission corresponding to the same boutique information to obtain the boutique commission value corresponding to the boutique information.
[0084] In the case where there is no accessory information about accessories in the user's car purchase needs, it means that the user has no specific requirements for the vehicle accessories. At this time, in order to facilitate the user to use the car more comfortably, the built-in matching rules can be used to match several recommended accessory groups for this recommended car model. Among them, the built-in matching rules are provided by experienced sales staff and have been stored and set. For example, non-fuel-consuming cars will generate at least one recommended accessory group containing charging piles, all cars will generate a blank recommended accessory group, and cars without special requirements will generate recommended accessory groups involving various combinations of foot pads, car films and maintenance times. For example, if the recommended car model is an electric-consuming car of brand A, then the recommended car model will be matched with a recommended accessory group containing only charging piles, a recommended accessory group containing charging piles and foot pads, a recommended accessory group containing charging piles and car films, a recommended accessory group containing charging piles, car films and foot pads, a recommended accessory group containing only foot pads, a recommended accessory group containing only car films, a recommended accessory group containing foot pads and car films, and a recommended accessory group without any accessories, that is, a blank recommended accessory group.
[0085] When the user's car purchase needs include accessory information about accessories, it indicates that the user has specific requirements for vehicle accessories. In order to better meet the user's needs, the recommended accessory group corresponding to the recommended car model can be determined based on the accessory information. That is, each recommended accessory group obtained includes at least the accessories corresponding to the user's vehicle accessory information. For example, if the user's car purchase needs include the accessory information of a car film, then each recommended accessory group corresponding to the recommended car model determined based on the accessory information includes at least a car film.
[0086] After obtaining several recommended accessory groups corresponding to each recommended car model, by combining each recommended accessory group corresponding to the same recommended car model with the recommended car model, several boutique information corresponding to the recommended car model can be obtained, and each boutique information includes a recommended car model and a recommended accessory group. Among them, each recommended car model corresponds to a car commission, and each recommended accessory group corresponds to an accessory commission. The sum of the commissions obtained by adding the car commission and the accessory commission corresponding to the same boutique information is the boutique commission value corresponding to the boutique information.
[0087] Finally, the boutique information of each recommended car model obtained based on the user's car purchase needs and the corresponding boutique commission value and initial recommendation value are substituted into the trained learning model for processing to obtain ordered recommendation information that can recommend premium cars to users, and sent to the salesperson, so that the salesperson can make recommendations to users based on the ordered recommendation information without having to substitute his or her own subjective consciousness, and can respond to the user's personalized needs, thereby improving the accuracy of recommending premium cars to users. Among them, the trained learning model includes a first learning model and a second learning model, and using the trained learning model to process each boutique information and the corresponding boutique commission value and initial recommendation value includes the following steps:
[0088] Step S405 , using the first learning model to process the premium commission value and the initial recommendation value to obtain a revised recommendation value for each car model that represents a balance between commission and user demand.
[0089] Step S406: Use the second learning model to process the revised recommendation value and the premium product information to obtain ordered recommendation information for recommending premium automobile products to the user.
[0090] Among them, the input information of the first learning model is the boutique commission value and the initial recommendation value corresponding to a certain recommended car model, and the output information is the revised recommendation value of this recommended car model. The revised recommendation value is the recommended value when the salesperson finally recommends the car boutique corresponding to the recommended car model to the user. The first learning model is mainly used to balance the commission that the salesperson can get and the degree of satisfying the user's car purchase needs, so that the revised recommendation value output by the first learning model can be more in line with the positioning of the car boutique promotion. The first learning model is specifically a neural network deep learning model, which is obtained through training with the revised recommendation values and boutique commission values and initial recommendation values corresponding to a large amount of successful recommendation information.
[0091] The input information of the second learning model is an input information group composed of all the modified recommendation values and corresponding boutique information output by the first learning model, wherein the input information group is a two-dimensional array, the first line is the modified recommendation value, the second line is the boutique information, and the modified recommendation value and boutique information in each column correspond to each other. The output information is an ordered recommendation information, and the ordered recommendation information is arranged in ascending or descending order according to the modified recommendation value. The second learning model is mainly used to sort the modified recommendation values output by the first learning model and to sort the boutique information corresponding to it, so that the second learning model outputs an ordered recommendation information that can be used to recommend automobile boutiques to users. In this way, by dividing the learning model into the first learning model and the second learning model, compared with using only one learning model, the learning model can be simplified, the time and cost spent on training the learning model can be shortened, and the update of the learning model can be quickly completed when it needs to be updated later.
[0092] In addition, after obtaining the ordered recommendation information of the user recommending the automobile boutique to the user, the following steps are also included:
[0093] Step S500, obtaining the actual automobile boutique selected by the user and the position number of the actual automobile boutique in the ordered recommendation information.
[0094] Step S600, determine whether the position serial number falls within a preset qualified serial number range, if so, retain the learning model.
[0095] Step S700: If it does not fall into the category, obtain the information difference between the actual automobile boutique and the ordered recommendation information, and adjust the learning model based on the information difference.
[0096] After each transaction of a user is completed, the actual car boutique selected by the user will be fed back to the back-end control end to obtain the actual car boutique selected by the user, and the actual car boutique will be compared with the obtained ordered recommendation information to obtain the position number of the actual car boutique in the ordered recommendation information. Then the obtained position number is compared with the preset qualified number range. If the obtained position number is not less than the minimum value of the qualified number range and not greater than the maximum value of the qualified number range, then the position number falls within the preset qualified number range. Otherwise, the position number does not fall within the preset qualified number range.
[0097] When the position serial number falls within the preset qualified serial number range, it indicates that the current learning model can still accurately output ordered recommendation information for recommending high-quality automobiles to users. At this time, there is no need to adjust the learning model, that is, the learning model is retained to handle the car purchase needs of subsequent users.
[0098] If the position number does not fall within the preset qualified range, it indicates that the current learning model cannot accurately output the ordered recommendation information for recommending premium automobiles to users, and the learning model needs to be adjusted. That is, the actual premium automobiles are compared with the premium information corresponding to the highest recommendation value in the ordered recommendation information to obtain the difference between the two, thereby obtaining the information difference.
[0099] Adjusting the learning model through information difference includes: sending the information difference to the first learning model, and adjusting the first learning model based on the information difference.
[0100] The information difference includes the difference in one or several attributes between the actual premium automobile product and the premium information corresponding to the highest recommendation value in the ordered recommendation information. The information difference is used to increase the weight of the initial recommendation value corresponding to the premium information of the attributes involved in the information difference, thereby achieving the adjustment of the first learning model, that is, the adjustment of the learning model. In this way, when the existing update model cannot determine the ordered recommendation information well, a new learning model can be updated in time to replace the existing learning model, which also serves the purpose of improving the accuracy of recommending premium automobile products to users.
[0101] Figure 5 is a framework diagram of an interactive car boutique recommendation system based on artificial intelligence provided by an embodiment of the present application. Figure 5 As shown, the interactive automobile boutique recommendation system based on artificial intelligence includes: an information acquisition module, a processing module, a calculation module and a recommendation module.
[0102] Among them, the information acquisition module is used to obtain the user's car purchase needs, wherein the car purchase needs include at least one demand information. The processing module is used to determine whether each demand information is accessory information. If not, the corresponding car recommendation group is obtained from the preset database based on the demand information, wherein each car recommendation group includes several recommended car models. The calculation module is used to obtain the weight value corresponding to each car recommendation group, and determine the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group. The recommendation module is used to obtain the boutique information and corresponding boutique commission value of each recommended car model based on the car purchase needs, and use the trained learning model to process each boutique information and the corresponding boutique commission value and initial recommendation value to obtain ordered recommendation information for recommending car boutiques to users.
[0103] The other functions performed by the above-mentioned information acquisition module, processing module, calculation module and recommendation module as well as the technical details of each function are the same or similar to the corresponding features in the interactive automobile boutique recommendation method based on artificial intelligence described above, so they will not be repeated here.
[0104] An embodiment of the present application also provides a computer storage medium on which a computer program is stored. When the computer program is run on a computer, the computer can execute the steps of the interactive automobile boutique recommendation method based on artificial intelligence described above.
[0105] It should be understood that, although the steps in the flowchart of the accompanying drawings are shown in sequence according to the instructions of the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless explicitly stated in this document, the execution of these steps is not strictly limited in order and can be performed in other orders.
[0106] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.
Claims
1. An interactive automobile boutique recommendation method based on artificial intelligence, characterized in that: The method comprises: Acquire the user's car purchase demand, wherein the car purchase demand includes at least one demand information; Determine whether each piece of demand information is accessory information, and if not, obtain a corresponding car recommendation group from a preset database based on the demand information, wherein each car recommendation group includes a number of recommended car models; Obtaining a weight value corresponding to each car recommendation group, and determining an initial recommendation value for each recommended car model based on the weight value and the corresponding car recommendation group; Based on the car purchase demand, obtain the boutique information and corresponding boutique commission value of each recommended car model, and use the trained learning model to process each boutique information and corresponding boutique commission value and initial recommendation value to obtain ordered recommendation information for recommending automobile boutiques to the user; Wherein, obtaining the user's car purchase demand includes: Obtaining historical browsing information of the user on the platform, and dividing the historical browsing information by attributes to obtain a plurality of historical browsing sub-information, wherein each historical browsing sub-information corresponds to a unique attribute, and each historical browsing sub-information includes historical attribute browsing sub-information corresponding to at least one time period; Determine the demand information corresponding to each attribute based on the historical attribute browsing sub-information corresponding to the attribute, and determine all demand information corresponding to the same user as the car purchase demand of the user; The preset database stores a plurality of storage units, each storage unit corresponds to a uniquely determined attribute, and each storage unit stores each car model and the corresponding attribute parameter. The corresponding car recommendation group is obtained from the preset database based on the demand information, including: Acquire a demand attribute corresponding to the demand information, and determine a corresponding demand storage unit in the preset database based on the demand attribute; Selecting target attribute parameters that meet the demand information in the demand storage unit, and determining the car model corresponding to each target attribute parameter as a recommended car model that meets the demand information; It is determined whether the demand attribute is an ordered attribute. If so, the corresponding recommended car models are packaged in ascending or descending order according to the target attribute parameter to obtain a car recommendation group.
2. The method according to claim 1, characterized in that: The determining of the initial recommendation value of each recommended car model based on the weight value and the corresponding car recommendation group includes: Determine whether the car recommendation group belongs to the ordered attribute respectively. If so, derive a derived weight value corresponding to each recommended car model in the car recommendation group based on the weight value corresponding to the car recommendation group, and give the derived weight value to the weight value of the recommended car model in the car recommendation group. If not, assigning a weight value corresponding to the car recommendation group to each recommended car model in the car recommendation group; The weight values given to each recommended car model are added up respectively to obtain an initial recommendation value of the recommended car model.
3. The method according to claim 1, characterized in that Each recommended car model corresponds to at least one piece of boutique information, and the boutique information and corresponding boutique commission value of each recommended car model obtained based on the car purchase demand include: Determine whether the car purchase demand includes the demand information of accessories, and if not, match a number of recommended accessories groups for each recommended car model; If included, determining a recommended accessory group corresponding to each recommended automobile model based on the accessory information; Each recommended car model and each corresponding recommended accessories group constitute a high-quality information; The car commission and the parts commission included in each boutique information are obtained, and the car commission and the parts commission corresponding to the same boutique information are added together to obtain the boutique commission value corresponding to the boutique information.
4. The method according to claim 1, characterized in that: The trained learning model includes a first learning model and a second learning model, and the use of the trained learning model to process each boutique information and the corresponding boutique commission value and initial recommendation value includes: Processing the boutique commission value and the initial recommendation value using the first learning model to obtain a revised recommendation value for each car model that represents a balance between commission and user demand; The modified recommendation value and the boutique information are processed using a second learning model to obtain ordered recommendation information for recommending premium automobile products to the user.
5. The method according to claim 4, characterized in that After obtaining the ordered recommendation information for recommending automobile products to the user: Acquire the actual automobile boutique selected by the user and the position sequence number of the actual automobile boutique in the ordered recommendation information; Determine whether the position sequence number falls within a preset qualified sequence number range, and if so, retain the learning model; If it does not fall into the range, the information difference between the actual automobile boutique and the ordered recommendation information is obtained, and the learning model is adjusted based on the information difference.
6. The method according to claim 5, characterized in that The adjusting the learning model based on the information difference comprises: The information difference is sent to a first learning model, and the first learning model is adjusted based on the information difference.
7. An interactive car boutique recommendation system based on artificial intelligence, characterized by: The system includes: an information acquisition module, a processing module, a calculation module and a recommendation module; wherein, The information acquisition module is used to acquire the user's car purchase demand, wherein the car purchase demand includes at least one demand information; The processing module is used to determine whether each piece of demand information is accessory information, and if not, obtain a corresponding car recommendation group from a preset database based on the demand information, wherein each car recommendation group includes a number of recommended car models; The calculation module is used to obtain a weight value corresponding to each car recommendation group, and determine an initial recommendation value for each recommended car model based on the weight value and the corresponding car recommendation group; The recommendation module is used to obtain the boutique information and corresponding boutique commission value of each recommended car model based on the car purchase demand, and use the trained learning model to process each boutique information and the corresponding boutique commission value and initial recommendation value to obtain ordered recommendation information for recommending automobile boutiques to users; Wherein, obtaining the user's car purchase demand includes: Obtaining historical browsing information of the user on the platform, and dividing the historical browsing information by attributes to obtain a plurality of historical browsing sub-information, wherein each historical browsing sub-information corresponds to a unique attribute, and each historical browsing sub-information includes historical attribute browsing sub-information corresponding to at least one time period; Determine the demand information corresponding to each attribute based on the historical attribute browsing sub-information corresponding to the attribute, and determine all demand information corresponding to the same user as the car purchase demand of the user; The preset database stores a plurality of storage units, each storage unit corresponds to a uniquely determined attribute, and each storage unit stores each car model and the corresponding attribute parameter. The corresponding car recommendation group is obtained from the preset database based on the demand information, including: Acquire a demand attribute corresponding to the demand information, and determine a corresponding demand storage unit in the preset database based on the demand attribute; Selecting target attribute parameters that meet the demand information in the demand storage unit, and determining the car model corresponding to each target attribute parameter as a recommended car model that meets the demand information; It is determined whether the demand attribute is an ordered attribute. If so, the corresponding recommended car models are packaged in ascending or descending order according to the target attribute parameter to obtain a car recommendation group.
8. A computer-readable storage medium having stored thereon a computer program that can be run on a processor, characterized in that: When the computer program is executed by the processor, the interactive automobile boutique recommendation method based on artificial intelligence as described in any one of claims 1 to 6 is implemented.
Citation Information
Patent Citations
Systems and methods of linear regression models and machine learning models for vehicles
CA3193024A1
Vehicle-mounted control module and matching method thereof
CN104699686A
Vehicle intelligent recommendation method, device, device and storage medium
CN109300018A
Automobile sales management method and device and computer storage medium
CN109741136A
Vehicle model recommendation method, method and system for obtaining vehicle model recommendation list, medium and electronic equipment
CN112507241A
