CN111079001A

CN111079001A - Decoration recommendation information generation method and device, storage medium and electronic equipment

Info

Publication number: CN111079001A
Application number: CN201911170048.6A
Authority: CN
Inventors: 蒋志颖; 范娇娇
Original assignee: Beike Technology Co Ltd
Current assignee: Beike Technology Co Ltd
Priority date: 2019-11-26
Filing date: 2019-11-26
Publication date: 2020-04-28

Abstract

The application discloses a decoration recommendation information generation method and device, a storage medium and electronic equipment, and relates to the field of computing science and technology. The specific scheme comprises the following steps: acquiring preference information of a target user and house attribute information of a house to be decorated; determining a target user type to which the target user belongs from at least one user type according to the preference information of the target user; determining a target recommendation model from at least one recommendation model obtained by pre-training according to the type of the target user; the target recommendation model is obtained by training at least one sample user belonging to the target user type; and inputting the house attribute information into the target recommendation model to obtain at least one decoration recommendation information. The method and the system can better combine the preference information and the house attribute information of the target user, can better exert the house type and the characteristics of the house, and can obtain the most reasonable decoration recommendation information meeting the real preference of the user.

Description

Decoration recommendation information generation method and device, storage medium and electronic equipment

Technical Field

The application relates to the field of computing science and technology, in particular to a decoration recommendation information generation method and device, a storage medium and electronic equipment.

Background

With the development and promotion of urbanization, the house decoration needs are more and more. When house decoration is carried out, user preferences are different, and physical characteristics of different houses are different greatly. In the prior art, the limited decoration effect drawings which are formed by the templates are provided for users to select, and the limited decoration templates completely depend on the personal experience of designers, so that the provided decoration templates are completely limited by the personal experience of the designers, the house types and the characteristics of houses are often difficult to be really played, and the most reasonable scheme which meets the real preferences of the users cannot be obtained.

Disclosure of Invention

In view of the above, a main object of the present application is to provide a decoration recommendation information generating method, which can get rid of the personal experience limitation of designers, better combine the preference information and the house attribute information of the target user, and can not only better exert the house type and the characteristics of the house, but also obtain the most reasonable decoration recommendation information that meets the real preferences of the user.

In order to achieve the purpose, the technical scheme provided by the application is as follows:

in a first aspect, an embodiment of the present application provides a decoration recommendation information generation method, including the following steps:

acquiring preference information of a target user and house attribute information of a house to be decorated;

determining a target user type to which the target user belongs from at least one user type according to the preference information of the target user;

determining a target recommendation model from at least one recommendation model obtained by pre-training according to the type of the target user; the target recommendation model is obtained by training samples belonging to the target user type;

and inputting the house attribute information into the target recommendation model to obtain at least one decoration recommendation information.

In a possible implementation manner, the step of determining the user type to which the target user belongs according to the preference information includes:

extracting preference features based on the preference information;

inputting the preference characteristics into a clustering model obtained by pre-training to obtain the type of the target user; the clustering model is obtained by adopting historical preference information of at least one sample user for pre-training.

In a possible implementation, before the step of extracting preference features based on the preference information, the method further includes a step of training the clustering model:

obtaining at least one sample user and respective historical preference information of each sample user;

respectively extracting historical preference features based on each piece of historical preference information;

determining at least one cluster, and randomly determining a respective cluster center of each cluster;

according to each historical preference feature, respectively determining sample users belonging to each cluster by using the cluster model;

respectively aiming at each cluster, updating the cluster center of the cluster according to the historical preference characteristics of the sample users belonging to the cluster;

judging whether iteration stopping conditions are met or not, and returning to the step of executing the step of determining the sample users belonging to each cluster by using the clustering model according to each historical preference feature when the iteration stopping conditions are not met;

and when the iteration stop condition is met, determining each obtained cluster as the user type.

In a possible implementation, before the step of extracting the house attribute feature based on the house attribute information, the method further includes a step of training the recommendation model:

acquiring at least one sample user belonging to each user type according to each user type, and acquiring historical house attribute information, historical decoration recommendation information, target historical decoration recommendation information actually selected by each sample user and historical grading information of each sample user on the historical decoration recommendation information;

respectively extracting historical house attribute features based on each piece of historical house attribute information, respectively extracting historical decoration recommendation features based on each piece of historical decoration recommendation information, determining target historical decoration recommendation features according to target historical decoration recommendation information, and respectively extracting historical grading information features based on each piece of historical grading information;

and training an initial recommendation model by using the historical house attribute characteristics, the historical decoration recommendation characteristics, the target historical decoration recommendation characteristics and the historical grading information characteristics to obtain the recommendation model of the user type.

In one possible embodiment, the historical decoration recommendation information includes color information; the color information comprises wall surface color information and/or floor color information;

the step of respectively extracting the historical decoration recommendation features based on each piece of historical decoration recommendation information comprises the following steps:

acquiring RGB color values of the color information;

and normalizing the RGB color values to obtain the historical decoration recommended features.

In one possible embodiment, the recommendation model is a neural network model comprising an input layer, an intermediate layer and an output layer,

the step of training an initial recommendation model by using the historical house attribute features, the historical decoration recommendation features and the historical scoring information features to obtain the recommendation model of the user type comprises the following steps:

connecting the historical house attribute features to the input layer, connecting the historical decoration recommendation features and the historical scoring information features to the output layer, and adjusting parameters of the middle layer of the initial recommendation model through back propagation training to obtain the recommendation model of the user type.

In a possible embodiment, after the step of obtaining at least one decoration recommendation information, the method further comprises:

and sending the at least one decoration recommendation information to the target user, and recording the grading information of each decoration recommendation information of the target user.

In a second aspect, an embodiment of the present application further provides a decoration recommendation information generating device, including:

the acquisition module is used for acquiring preference information of a target user and house attribute information of a house to be decorated;

the user type determining module is used for determining a target user type to which the target user belongs from at least one user type according to the preference information of the target user;

the recommendation model determining module is used for determining a target recommendation model from at least one recommendation model obtained by pre-training according to the target user type; the target recommendation model is obtained by training samples belonging to the target user type;

and the recommendation information determining module is used for inputting the house attribute information into the target recommendation model to obtain at least one decoration recommendation information.

In a possible implementation manner, the user type determining module further includes:

the preference feature extraction module is used for extracting preference features based on the preference information;

the clustering model is used for inputting the preference characteristics to obtain the type of the target user; the clustering model is obtained by adopting historical preference information of at least one sample user for pre-training.

In a possible implementation manner, the decoration recommendation information generating apparatus further includes a clustering model training module, where the clustering model training module includes:

the historical preference information acquisition module is used for acquiring at least one sample user and the respective historical preference information of each sample user;

the historical preference feature extraction module is used for respectively extracting historical preference features based on each piece of historical preference information;

the cluster initialization module is used for determining at least one cluster and randomly determining the respective cluster center of each cluster;

the initial clustering model is used for respectively determining sample users belonging to each cluster according to each historical preference feature;

the cluster updating module is used for respectively aiming at each cluster and updating the cluster center of the cluster according to the historical preference characteristics of the sample users belonging to the cluster;

and the judging module is used for judging whether the iteration stopping condition is met.

In a possible implementation manner, the decoration recommendation information generating apparatus further includes a recommendation model training module, including:

the system comprises a sample information acquisition module, a storage module and a display module, wherein the sample information acquisition module is used for acquiring at least one sample user belonging to each user type according to each user type, and acquiring historical house attribute information, historical decoration recommendation information and target historical decoration recommendation information actually selected by each sample user of the sample user, and historical scoring information of each sample user on the historical decoration recommendation information;

the sample feature extraction module is used for respectively extracting historical house attribute features based on each piece of historical house attribute information, respectively extracting historical decoration recommendation features based on each piece of historical decoration recommendation information, determining target historical decoration recommendation features according to the target historical decoration recommendation information, and respectively extracting historical grading information features based on each piece of historical grading information;

and the model training module is used for training an initial recommendation model by using the historical house attribute characteristics, the historical decoration recommendation characteristics, the target historical decoration recommendation characteristics and the historical scoring information characteristics to obtain the recommendation model of the user type.

a sample feature extraction module further to:

acquiring RGB color values of the color information;

connecting the historical house attribute features to the input layer, and connecting the historical decoration recommendation features and the historical scoring information features to the output layer;

the intermediate layer is specifically configured to adjust parameters of the initial recommendation model through back propagation training to obtain the recommendation model of the user type.

In one possible embodiment, the decoration recommendation information generating apparatus further includes:

the sending module is used for sending the at least one decoration recommendation information to the target user;

and the storage module is used for recording the grading information of each decoration recommendation information of the target user.

In a third aspect, an embodiment of the present application further provides a computer-readable storage medium. The specific scheme is as follows:

a computer readable storage medium storing computer instructions which, when executed by a processor, may implement the steps of any one of the possible embodiments of the first aspect and the first aspect.

In a fourth aspect, an embodiment of the present application further provides an electronic device. The specific scheme is as follows:

an electronic device, comprising:

a processor;

a memory for storing the processor-executable instructions;

the processor is configured to read the executable instructions from the memory and execute the instructions to implement the steps of any one of the possible implementations of the first aspect and the first aspect.

In summary, the present application provides a decoration recommendation information generating method, device, storage medium, and electronic device. The method and the device for recommending the decoration by the user type better combine preference information and house attribute information of the target user, classify the users according to the preference information of the target user, respectively train the sample users of each user type to obtain recommendation models corresponding to the user types, determine the target user type of the target user, determine the target recommendation model according to the target user type, and obtain the decoration recommendation information according to the house attribute information by using the target recommendation model. The obtained decoration recommendation information is determined by combining the preference information of the target user and the house attribute information, so that the house type and the characteristics of the house can be better played, and the most reasonable decoration recommendation information meeting the real preference of the user can be obtained.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive labor.

Fig. 1 is a schematic flow chart of a decoration recommendation information generation method according to an embodiment of the present application;

fig. 2 is a schematic flow chart of another decoration recommendation information generation method according to an embodiment of the present application;

FIG. 3 is a schematic flow chart of a clustering model training step in the decoration recommendation information generation method;

FIG. 4 is a schematic diagram of a recommendation model in the decoration recommendation information generation method;

FIG. 5 is a system architecture diagram of another decoration recommendation information generation method provided in the embodiment of the present application;

fig. 6 is a schematic structural diagram of a decoration recommendation information generation apparatus according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of another decoration recommendation information generation apparatus according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The terms "first," "second," "third," "fourth," and the like in the description and in the claims of the present application and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprising" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, method, article, or apparatus.

The core invention points of the application are as follows: classifying users according to preference information of target users, respectively training by using sample users of each user type as samples to obtain recommendation models corresponding to the user types, determining the target user type of the target user, determining to obtain a target recommendation model according to the target user type, and obtaining decoration recommendation information according to house attribute information by using the target recommendation model. The obtained decoration recommendation information is determined by combining the preference information of the target user and the house attribute information, so that the house type and the characteristics of the house can be better played, and the most reasonable decoration recommendation information meeting the real preference of the user can be obtained.

The decoration recommendation information generation method provided by the embodiment of the application is generally realized in electronic equipment of a server, obtains the preference information and the house attribute information of the target user sent by a client, or obtains the preference information and the house attribute information of the target user stored in a memory, and sends the decoration recommendation information to the client after generating the decoration recommendation information so as to be checked and referred by the user.

In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention are described in detail below with specific embodiments. Several of the following embodiments may be combined with each other and some details of the same or similar concepts or processes may not be repeated in some embodiments.

Example one

Fig. 1 is a schematic flow chart of a first embodiment of the present application, as shown in fig. 1, the first embodiment mainly includes:

s101: and acquiring preference information of a target user and house attribute information of a house to be decorated.

Here, the target user is usually an owner or user who needs to finish the house to be finished, and the generated finishing recommendation information is provided to the target user, so that the target user can determine the finishing scheme of the house to be finished by referring to the finishing recommendation information. In order to generate decoration recommendation information meeting the real preferences of a user, preference information of a target user needs to be acquired, wherein the preference information represents the preferences of the target user for visual effects presented by decoration.

The house to be decorated can be one or more houses, one or more sets of houses, the house attribute information is the physical attribute information of the house to be decorated, which can affect the decoration effect, the house attribute information is the physical attribute information describing the internal space and the external environment of the house to be decorated, and exemplarily, the house attribute information can comprise the orientation, the floor, the house type, the size, the depth, the region, the price and the like of the house.

S102: and determining the target user type to which the target user belongs from at least one user type according to the preference information of the target user.

Classifying the users according to the preference information, determining at least one user type in advance, and determining the target user type to which the target user belongs from at least one user type obtained in advance according to the preference information of the target user when determining decoration recommendation information of the house to be decorated of the target user.

S103: determining a target recommendation model from at least one recommendation model obtained by pre-training according to the type of the target user; and the target recommendation model is obtained by training at least one sample user belonging to the target user type.

In order to better combine preference information of a target user and house attribute information of a house to be decorated, and enable decoration recommendation information to be obtained through determination, the house type and the characteristics of the house to be decorated can be better played, and real preferences of the target user can be met. Therefore, the target recommendation model corresponding to the target user type can be determined and obtained from the recommendation model according to the target user type, the target recommendation model is obtained by training sample users belonging to the target user type, the recommendation model is trained by using the sample users belonging to the same user type as the sample, interference of the sample users of different user types on the recommendation model training can be avoided, pertinence and accuracy of the decoration recommendation information determined by the recommendation model for the user of the user type are improved, and compared with the method of training the recommendation model without distinguishing the user type, the decoration recommendation information obtained by determination can better meet real preference of the target user on the basis of exerting the house type and characteristics of a house to be decorated.

S104: and inputting the house attribute information into the target recommendation model to obtain at least one decoration recommendation information.

And the target recommendation model determines to obtain at least one decoration recommendation information according to the house attribute information of the house to be decorated. When the decoration recommendation information is determined to be obtained, the obtained decoration recommendation information can be presented to a target user, and the target user determines whether to adopt the decoration recommendation information; when more than two decoration recommendation information are determined to be obtained, each obtained decoration recommendation information can be presented to a target user, and the target user selects from the determined decoration recommendation information.

Example two

Further, in order to improve the implementation effect in the specific implementation, an embodiment of the present application further provides a decoration recommendation information generating method, as shown in fig. 2, including:

s201: and acquiring preference information of a target user and house attribute information of a house to be decorated.

Preferably, the answer to the questionnaire of the target user can be obtained according to a predetermined preference questionnaire; and determining the preference information according to the questionnaire answers. The preference questionnaire can be in the form of choice questions, for example, in the form of single choice questions and/or multiple choice questions, and the preference questionnaire generated in the form of choice questions can facilitate the limitation of the answer range of the questionnaire, the determination of preference information, and further, the extraction of preference characteristics. In specific implementation, the answers to the questionnaire can be directly used as preference information, or the preference information can be obtained by analyzing and determining the answers to the questionnaire.

S202: and determining the target user type to which the target user belongs from at least one user type according to the preference information of the target user.

Preferably, the method determines the type of the target user to which the target user belongs by using a clustering model obtained by pre-training, and specifically may determine the type of the user to which the target user belongs by using the following steps 1 and 2:

and 1, extracting preference characteristics based on the preference information.

When the preference information is determined in the form of a user questionnaire, particularly, the preference questionnaire is in the form of a choice question, and answers to the questionnaire are directly used as the preference information, preference features can be directly extracted according to the answers to the questionnaire.

For example, assuming that the preference questionnaire includes 9 questions, and each question has four ABCD options, each question may be abstracted into a four-dimensional question feature, and the question features of each question are spliced to obtain a 36-dimensional preference feature. For example, when the target user answers a certain question, two options a and C are selected, a four-dimensional question feature [1, 0, 1, 0] can be extracted according to the questionnaire answers to the question by the target user, and the four-dimensional question features of 9 questions are spliced to obtain a complete 36-dimensional preference feature.

Furthermore, because the preference features obtained by extracting the preference features according to the answers of the questionnaire belong to high-dimensional feature information, the extracted high-dimensional feature information has high redundancy, and the effect of directly adopting the high-dimensional feature information to determine the type of the target user is not good, preferably, the intermediate preference features are extracted based on the preference information; and then, reducing the dimension of the intermediate preference feature to obtain the preference feature. In specific implementation, any one of commonly used algorithms may be used to perform dimension reduction on the intermediate preference feature, for example, algorithms such as Principal Component Analysis (PCA) and self-encoding algorithm (AutoEncode) may be used to perform dimension reduction on the intermediate preference feature to obtain the preference feature.

Step 2, inputting the preference characteristics into a clustering model obtained by pre-training to obtain the type of the target user; the clustering model is obtained by adopting historical preference information of at least one sample user for pre-training.

And inputting the preference characteristics into a clustering model obtained by pre-training to obtain the type of the target user. Before the target user type is obtained by using the clustering model, the clustering model needs to be trained in advance, and specifically, the clustering model can be trained by adopting the steps shown in fig. 3:

s301: and acquiring at least one sample user and respective historical preference information of each sample user.

Preferably, the respective historical preference information of the sample users is obtained, and the clustering model is subjected to unsupervised training according to the historical preference information. The historical preference information can also be acquired by adopting a predetermined preference questionnaire, a certain number of users are required to collect information before the clustering model is trained, a plurality of users fill in the preference questionnaire, and the questionnaire answers of each user are collected and stored. Each user who fills in the preference questionnaire may be taken as a sample user, or two or more users may be randomly drawn from each user who fills in the preference questionnaire as sample users. And taking the stored questionnaire answers of the sample users as historical preference information, or determining to obtain the respective historical preference information of each sample user according to the stored questionnaire answers of the sample users.

S302: and respectively extracting historical preference features based on each piece of historical preference information.

The step of extracting the historical preference feature based on each piece of the historical preference information is similar to the step of extracting the preference feature based on the preference information in the step 1. And the intermediate historical preference features can be extracted respectively based on each piece of historical preference information, and then dimension reduction is carried out on the intermediate historical preference features to obtain the historical preference features.

S303: at least one cluster is determined, and a respective cluster center for each of the clusters is randomly determined.

Firstly, the number of clusters is determined, and the number of clusters is at least one, namely at least one cluster is determined. And placing each extracted historical preference feature in a vector space with the same dimension as the feature, and randomly determining the respective cluster center of each cluster in the vector space. Assuming that k clusters are determined, that is, users are divided into k user types, then k cluster centers are randomly determined in a vector space, and a coordinate value in the vector space is generally used to represent each cluster center.

S304: and according to each historical preference characteristic, respectively determining sample users belonging to each cluster by using the cluster model.

And according to each historical preference characteristic, respectively determining sample users belonging to each cluster by using a clustering model. Specifically, according to each historical preference feature, a clustering model is used for determining the distance between the historical preference feature and each clustering center, and sample users corresponding to the historical preference feature are classified into the clustering with the closest distance to the clustering center, so that the sample users belonging to each clustering are determined and obtained.

S305: and respectively aiming at each cluster, updating the cluster center of the cluster according to the historical preference characteristics of the sample users belonging to the cluster.

For example, the coordinate value of the cluster center of each cluster may be updated to the average of the historical preference features of the sample users belonging to the cluster.

S306: and judging whether the iteration stop condition is met.

And returning to the step S304 when the iteration stop condition is not met, and executing the step of respectively determining the sample users belonging to each cluster by using the cluster model according to each historical preference feature. When the iteration stop condition is satisfied, step S307 is executed.

S307: and respectively determining each obtained cluster as the user type.

For example, the iteration stop condition may be:

aiming at each cluster, when the difference between the updated coordinate value of the cluster center of the cluster and the coordinate value of the cluster center of the cluster obtained by the previous iteration is smaller than a preset difference threshold value;

or the iteration number is larger than a preset number threshold.

When the iteration stop condition is met, each cluster can be respectively determined as a user type, and the target user type to which the target user belongs can be determined and obtained by utilizing the cluster model according to the preference information of the target user.

S203: and determining a target recommendation model from at least one recommendation model obtained by pre-training according to the type of the target user.

And determining an obtained target recommendation model according to the type of the target user, and training historical house attribute information and historical decoration recommendation information of a respective sample house of at least one sample user belonging to the type of the target user, the target historical decoration recommendation information actually selected by each sample user, and historical scoring information of the historical decoration recommendation information by each sample user.

During pre-training, aiming at each user type, respectively training the recommendation models of the user type, wherein one user type corresponds to one pre-trained recommendation model.

Specifically, the following steps i to iii may be adopted to train the recommendation model:

and i, acquiring at least one sample user belonging to each user type according to each user type, and acquiring historical house attribute information, historical decoration recommendation information, target historical decoration recommendation information actually selected by each sample user and historical grading information of each sample user on the historical decoration recommendation information.

The sample user relies on the work provided in the previous data preparation and collection phase, and the sample user has collected and classified historical preference information according to the historical preference information. The method comprises the steps that when historical preference information of sample users is collected, the house attribute information and historical decoration recommendation information of respective sample houses of the sample users, target historical decoration recommendation information actually selected by each sample user and historical grading information of the historical decoration recommendation information of each sample user are collected. Specifically, the questionnaire answers of each user, the house attribute information of the house to be decorated of each user, the decoration recommendation information actually selected by each user, and the grading information of the decoration recommendation information by each user are collected and stored, and when the clustering model and the recommendation model need to be trained, each stored user is taken as a sample user, or more than two users are randomly extracted from each stored user to be taken as sample users. The method comprises the steps of taking questionnaire answers of stored sample users as preference information, taking house attribute information of houses to be decorated of the stored sample users as historical house attribute information, taking decoration recommendation information selected by the stored sample users as historical decoration recommendation information and target historical decoration recommendation information actually selected by each sample user, and taking grading information of the decoration recommendation information of the stored sample users as historical grading information of the historical decoration recommendation information of each sample user.

And ii, respectively extracting historical house attribute features based on each piece of historical house attribute information, respectively extracting historical decoration recommendation features based on each piece of historical decoration recommendation information, determining target historical decoration recommendation features according to the target historical decoration recommendation information, and respectively extracting historical grading information features based on each piece of historical grading information.

The historical property features of the house also include physical property information of the house to be decorated, which may affect the decoration effect, and illustratively, the historical property information of the house may include orientation, floor, house type, size, depth, region, price, and the like of the house. And extracting historical house attribute features, historical decoration recommendation features and historical grading information features respectively based on the historical house attribute information, the historical decoration recommendation information and the historical grading information.

And determining target historical decoration recommended features according to the target historical decoration recommended information, and specifically determining the historical decoration recommended features extracted from the target historical decoration recommended information actually selected by the sample user as the target historical decoration recommended features.

And iii, training an initial recommendation model by using the historical house attribute characteristics, the historical decoration recommendation characteristics, the target historical decoration recommendation characteristics and the historical grading information characteristics to obtain the recommendation model of the user type.

Specifically, calculating decoration recommendation information is determined and obtained by using an initial recommendation model according to the historical house attribute characteristics, and parameters of the initial recommendation model are adjusted according to the calculating decoration recommendation information, the historical decoration recommendation characteristics, the target historical decoration recommendation characteristics, the historical scoring information characteristics and respective weights, so that the recommendation model of the user type is obtained.

In actual implementation, the historical grading information of the sample user on the historical decoration recommended information is emphasized, so that the weight of the historical grading information characteristic is higher than the weight of the calculated decoration recommended information, the weight of the target historical decoration recommended characteristic and the weight of the historical decoration recommended characteristic.

The training process can also be carried out for more than two rounds, and the initial recommendation model is subjected to more than two rounds of iterative training to obtain the recommendation model of the user type. The training of the recommendation model is supervised training, the historical house attribute characteristics are input quantity of the model, the historical decoration recommendation characteristics, the target historical decoration recommendation characteristics and the historical grading information characteristics are output supervision quantity of the model, and a user can adjust all parameters of the recommendation model, optimize a target function and finally obtain the recommendation model.

Illustratively, an important link in house decoration is to select the wall color and the floor color, and the matching of the wall color and the floor color of the house decoration can lay a good foundation for the visual effect of the house decoration and even play a role in determining the style of the house decoration. Therefore, when the decoration recommendation information includes color information, the historical decoration recommendation information obtained during model training also includes color information, where the color information may include wall color information and/or floor color information.

Assuming that the color information in the decoration recommendation information includes wall color information and floor color information, determining that the obtained decoration recommendation information needs to include the wall color information and the floor color information by using the target recommendation model. At this time, it is preferable that the wall color information and the floor color information are represented by RGB color values. The step of extracting the historical decoration recommendation characteristics based on the historical decoration recommendation information includes: acquiring RGB color values of the color information; and normalizing the RGB color values to obtain the historical decoration recommended features. Since the RGB color value is usually a value between 0 and 255, the RGB color value can be divided by 255, i.e. the RGB color value can be normalized to obtain the recommended characteristics for history decoration.

Similarly, each piece of historical scoring information may be normalized to obtain the historical scoring information features. Specifically, the total score can be used for each historical scoring information, that is, each historical scoring information can be normalized respectively to obtain the historical scoring information characteristics.

S204: and inputting the house attribute information into the target recommendation model to obtain at least one decoration recommendation information.

Specifically, house attribute features are extracted based on the house attribute information, and the house attribute features are input into the target recommendation model to obtain the at least one decoration recommendation information.

For example, when the decoration recommendation information includes color information, a neural network model may be preferably used as the recommendation model. Assuming a case where the target recommendation model obtains decoration recommendation information according to the house attribute information, the adopted recommendation model may be as shown in fig. 4.

The recommended model shown in fig. 4 is a neural network model including an input layer, an intermediate layer and an output layer, the first layer is the input layer, and the input layer accesses the house attribute features extracted from the house attribute information; the second layer is an intermediate layer and is used for adjusting intermediate weight and intermediate parameters; the third layer is an output layer, the output layer comprises RGB color numerical values of wall surface color information and RGB color numerical values of floor color information, and the output layer further comprises scoring information of a user for supervised training.

The RGB color numerical value of the wall surface color information comprises three color numerical values, the RGB color numerical value of the floor color information comprises three color numerical values, the grading information of the user is added, and the output layer is a seven-dimensional vector. Here, the reason for adding the rating information of the user to the output layer is to train the recommendation model using the historical rating information features, and does not mean that the recommendation model determines to obtain the rating information of the user according to the house attribute features. Before training the recommendation model shown in fig. 4, the RGB color values of the wall color information, the RGB color values of the floor color information, and the historical score information of the sample user are normalized respectively to obtain the historical decoration recommendation features and the historical score information features. When the recommendation model shown in fig. 4 is trained, the historical house attribute features are connected to the input layer, the historical decoration recommendation features and the historical scoring information features are connected to the output layer, and parameters of the middle layer of the initial recommendation model are adjusted through back propagation training to obtain the recommendation model of the user type. Preferably, the weight of the historical scoring information features is greater than that of the historical decoration recommendation features, and when the training recommendation model is propagated reversely, the training gradient of the historical scoring information features is greater than that of the historical decoration recommendation features.

After the recommendation model is obtained through training, the house attribute information of the house to be decorated is input into the target recommendation model corresponding to the target user type of the target user, so that the RGB color numerical value of the wall surface color information and the RGB color numerical value of the floor color information can be obtained, and the RGB color numerical value of the wall surface color information and the RGB color numerical value of the floor color information are the decoration recommendation information.

S205: and sending the at least one decoration recommendation information to the target user, and recording the grading information of each decoration recommendation information of the target user.

Meanwhile, the obtained decoration recommendation information and the questionnaire answers of the target user can be recorded at the same time. The recorded grading information, decoration recommendation information and questionnaire answers of the target user to each decoration recommendation information can be used as historical grading information, historical decoration recommendation information and historical preference information at regular time, and the clustering model and the recommendation model are trained, so that the effects of the clustering model and the recommendation model are improved.

For example, the architecture diagram of the decoration recommendation information generation method provided in the embodiment of the present application is shown in fig. 5, in the data preparation stage, preference information of a user, decoration recommendation information actually selected by the user, house attribute information of a house to be decorated, and rating information of the decoration recommendation information by the user are collected through a preference questionnaire, in the data preparation stage, decoration recommendation information actually selected by the user, house attribute information of the house to be decorated, and rating information of the decoration recommendation information by the user may also be obtained in the form of a questionnaire, and the information is stored in a database. And then carrying out unsupervised training on the clustering model, specifically comprising the steps of acquiring preference information of the user stored in a database as historical preference information of the sample user, extracting historical preference characteristics according to the historical preference information, and training the initial clustering model by adopting the historical preference characteristics to obtain at least one user type and the clustering model obtained by pre-training. And carrying out supervised training on the recommendation model, specifically, for each user type, obtaining respective sample information of sample users belonging to the user type, wherein the sample information comprises historical house attribute information of respective sample houses of the sample users, historical decoration recommendation information, target historical decoration recommendation information actually selected by each sample user, and historical rating information of each sample user on the historical decoration recommendation information, and training the initial recommendation model of the user type to obtain the recommendation model of the user type.

When decoration recommendation information is determined for a target user, firstly, questionnaire answers of preference questionnaires filled by the target user are obtained, preference information of the target user is determined according to the questionnaire answers, the type of the target user to which the target user belongs is determined based on the preference information of the target user, a target recommendation model is determined from the recommendation model according to the type of the target user, house attribute information of a house to be decorated is input into the target recommendation model, at least one decoration recommendation information is determined, the decoration recommendation information is sent to the target user, and grading information of the target user on each decoration recommendation information is obtained. And storing the questionnaire answers of the target users, the house attribute information of the house to be decorated, the confirmed decoration recommendation information and the grading information of the target users on each decoration recommendation information into a database to perfect a data set.

The decoration recommendation information generation method provided by the embodiment of the application can better exert the house type and the characteristics of a house, so that the most reasonable decoration recommendation information meeting the real preference of a user is obtained.

Based on the same design concept, the embodiment of the application also provides a decoration recommendation information generation device, a storage medium and electronic equipment.

EXAMPLE III

As shown in fig. 6, a decoration recommendation information generating apparatus 600 according to an embodiment of the present application includes:

the obtaining module 601 is configured to obtain preference information of a target user and house attribute information of a house to be decorated;

a user type determining module 602, configured to determine, according to the preference information of the target user, a target user type to which the target user belongs from at least one user type;

a recommendation model determining module 603, configured to determine a target recommendation model from at least one recommendation model obtained through pre-training according to the target user type; the target recommendation model is obtained by training at least one sample user belonging to the target user type;

and a recommendation information determining module 604, configured to input the house attribute information into the target recommendation model to obtain at least one piece of decoration recommendation information.

The obtaining module 601 is connected to the user type determining module 602 and the recommendation information determining module 604, and the obtaining module 601 sends the obtained preference information of the target user to the user type determining module 602. The user type determining module 602 is connected to the recommendation model determining module 603, and the user type determining module 602 determines the target user type according to the preference information of the target user, and sends the target user type to the recommendation model determining module 603. The recommendation model determining module 603 is connected to the recommendation information determining module 604, the recommendation model determining module 603 determines a target recommendation model according to the type of the target user, and sends the target recommendation model to the recommendation information determining module 604, and the recommendation information determining module 604 obtains at least one decoration recommendation information according to the target recommendation model determined by the recommendation model determining module 603 and the house attribute information of the house to be decorated, which is obtained by the obtaining module 601.

In a possible implementation manner, the decoration recommendation information generating apparatus 600 further includes a central control module and a storage module, each module of the decoration recommendation information generating apparatus 600 is connected to the central control module, the storage module is also connected to the central control module, the central control module schedules execution sequence and data interaction between each module, stores each intermediate data in an internal storage or a memory, and reads or writes the intermediate data from or into the internal storage or the memory through the central control module.

As shown in fig. 7, for another decoration recommendation information generating apparatus 700 provided in this embodiment of the application, the decoration recommendation information generating apparatus 700 also includes an obtaining module 701, a user type determining module 702, a recommendation model determining module 703 and a recommendation information determining module 704.

In a possible implementation, the user type determining module 702 further includes:

a preference feature extraction module 7021 configured to extract a preference feature based on the preference information;

a user type determining module 7022, configured to input the preference feature to obtain the target user type; the clustering model is obtained by adopting historical preference information of at least one sample user for pre-training.

The preference feature extraction module 7021 is connected to the acquisition module 701 and the user type determination module 7022, extracts the preference feature according to the preference information acquired by the acquisition module 701, and sends the preference feature to the user type determination module 7022. The user type determining module 7022 is obtained by training in advance, and determines the target user type according to the preference characteristics.

In a possible implementation manner, the decoration recommendation information generating apparatus 600 further includes a clustering model training module 705, where the clustering model training module 705 includes:

a historical preference information obtaining module 7051, configured to obtain at least one sample user and respective historical preference information of each sample user;

a history preference feature extraction module 7052, configured to extract history preference features based on each piece of history preference information, respectively;

a cluster initialization module 7053, configured to determine at least one cluster, and randomly determine a cluster center of each cluster;

the initial clustering model 7054 is used for determining sample users belonging to each cluster respectively by using the clustering model according to each historical preference feature;

a cluster updating module 7055, configured to update, for each cluster, a cluster center of the cluster according to a historical preference feature of a sample user belonging to the cluster;

a judging module 7056, configured to judge whether an iteration stop condition is met, and when the iteration stop condition is not met, return to the initial clustering model 7054 to perform the step of determining, according to each historical preference feature, a sample user belonging to each cluster by using the clustering model;

and when the iteration stop condition is met, ending the iteration, and respectively determining each obtained cluster as the user type.

The clustering model training module 705 is connected to the user type determining module 7022, and the initial clustering model 7054 is trained to obtain the user type determining module 7022. In the cluster model training module 705, the historical preference information obtaining module 7051 is connected to the historical preference feature extracting module 7052, and sends the obtained at least one sample user and the respective historical preference information of each sample user to the historical preference feature extracting module 7052, so that the historical preference feature extracting module 7052 extracts the historical preference features. The cluster initialization module 7053 is connected to the initial cluster model 7054, and sends the determined initial clusters and cluster centers to the initial cluster model 7054. The initial clustering model 7054 is connected to the historical preference feature extraction module 7052, the clustering initialization module 7053, the clustering update module 7055, and the judgment module 7056, and continuously updates the clustering center under the driving of the judgment module 7056 until an iteration stop condition is satisfied, and then determines each obtained cluster as the user type.

In a possible implementation manner, the decoration recommendation information generating apparatus 700 further includes a recommendation model training module 706, including:

the sample information obtaining module 7061 is configured to obtain, for each user type, at least one sample user belonging to the user type, and obtain historical house attribute information and historical decoration recommendation information of a sample house of each sample user, target historical decoration recommendation information actually selected by each sample user, and historical scoring information of each sample user on the historical decoration recommendation information;

a sample feature extraction module 7062, configured to extract historical house attribute features based on each piece of historical house attribute information, extract historical decoration recommendation features based on each piece of historical decoration recommendation information, determine target historical decoration recommendation features according to target historical decoration recommendation information, and extract historical score information features based on each piece of historical score information;

and the model training module 7063 is configured to train an initial recommendation model by using the historical house attribute features, the historical decoration recommendation features, the target historical decoration recommendation features, and the historical scoring information features, so as to obtain the recommendation model of the user type.

The recommendation model training module 706 performs training of the recommendation model according to user type. For each user type, the sample information obtaining module 7061 is connected to the sample feature extracting module 7062, and sends the obtained sample information of the user type to the sample feature extracting module 7062, so that the sample feature extracting module 7062 extracts the sample features. The sample feature extraction module 7062 is connected to the model training module 7063, and trains the initial recommendation model according to the extracted sample features to obtain the recommendation model of the user type.

sample feature extraction module 7062 is further configured to:

acquiring RGB color values of the color information;

In one possible embodiment, as shown in fig. 4, the recommendation model is a neural network model comprising an input layer, an intermediate layer and an output layer,

In one possible embodiment, the decoration recommendation information generating apparatus 700 further includes:

a sending module 707, configured to send the at least one decoration recommendation information to the target user;

the storage module 708 is configured to record scoring information of each piece of decoration recommendation information by the target user.

The recommendation information determining module 704 is respectively connected to the sending module 707 and the storage module 708, and sends the decoration recommendation information determined by the recommendation information determining module 704 to the target user. The storage module 708 records the rating information and may also record the decoration recommendation information determined by the recommendation information determination module 704.

In a possible embodiment, the decoration recommendation information generating device 700 further includes a central control module and a storage module, each module of the decoration recommendation information generating device 700 is connected with the central control module, the storage module is also connected with the central control module, the central control module schedules execution sequence and data interaction between each module, stores each intermediate data in an internal storage or a memory, and reads or writes the intermediate data from or into the internal storage or the memory through the central control module.

The decoration recommendation information generation device provided by the embodiment of the application better combines the preference information and the house attribute information of the target user, can better exert the house type and the characteristics of a house, and can obtain the most reasonable decoration recommendation information meeting the real preferences of the user.

Example four

A computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the steps of any one of the fitment recommendation information generation methods provided by embodiments of the present application. In practical applications, the computer readable medium may be included in the apparatus/device/system described in the above embodiments, or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement the steps of performing any one of the decoration recommendation information generation methods provided by the embodiments of the present application according to any one of the decoration recommendation information generation apparatuses provided by referring to the embodiments of the present application.

According to embodiments disclosed herein, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example and without limitation: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing, without limiting the scope of the present disclosure. In the embodiments disclosed herein, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The method steps described herein may be implemented in hardware, for example, logic gates, switches, Application Specific Integrated Circuits (ASICs), programmable logic controllers, embedded microcontrollers, etc., in addition to data processing programs. Such hardware capable of implementing the methods described herein may also constitute the present application.

EXAMPLE five

The embodiment of the application further provides an electronic device which can be a computer or a server, wherein any one of the decoration recommendation information generation devices provided by the embodiment of the application can be integrated. As shown in fig. 8, an electronic device 800 provided by an embodiment of the application is shown.

The electronic device may include one or more processors 801 of a processing core, one or more memories 802 for storing instructions executable by the processors 801. The electronic device may further include a power supply 803, an input-output unit 804. Those skilled in the art will appreciate that fig. 8 does not constitute a limitation of the electronic device and may include more or fewer components than illustrated, or some components may be combined, or a different arrangement of components.

Wherein:

the processor 801 is a control portion of the electronic device, and connects various portions by using various interfaces and lines, reads the executable instructions from the memory 802, and executes or executes the instructions stored in the memory 802 to implement any one of the steps of the decoration recommendation information generation methods provided by the embodiments of the present application.

The memory 802 may be used to store a software program, that is, a program involved in any one of the decoration recommendation information generation methods provided by the embodiments of the present application.

The processor 801 executes various functional applications and data processing by running software programs stored in the memory 802. The memory 802 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data or the like used according to the needs of the electronic device. Further, the memory 802 may include high speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory 802 may also include a memory controller to provide the processor 801 access to the memory 802.

The electronic device further comprises a power supply 803 for supplying power to each component, and preferably, the power supply 803 can be logically connected with the processor 801 through a power management system, so that functions of charging, discharging, power consumption management and the like can be managed through the power management system. The power supply 803 may also include one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and any like components.

The server may also include an input-output unit 804, such as may be used to receive input numeric or character information, and to generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control; such as various graphical user interfaces that may be used to display information entered by or provided to the user, as well as the server, which may be composed of graphics, text, icons, video, and any combination thereof.

Any decoration recommendation information generation method, device, storage medium and electronic equipment provided by the embodiments of the present application are based on the same design concept, and the technical means in any embodiment of the present application can be freely combined, and the combined technical means is still within the protection scope of the present application.

The flowchart and block diagrams in the figures of the present application illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments disclosed herein. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

It will be appreciated by a person skilled in the art that various combinations and/or combinations of features described in the various embodiments and/or claims of the present application are possible, even if such combinations or combinations are not explicitly described in the present application. In particular, the features recited in the various embodiments and/or claims of the present application may be combined and/or coupled in various ways, all of which fall within the scope of the present disclosure, without departing from the spirit and teachings of the present application.

The principle and implementation of the present application are explained by applying specific embodiments in the present application, and the above description of the embodiments is only used to help understanding the method and the core idea of the present application, and is not used to limit the present application. It will be appreciated by those skilled in the art that changes may be made in this embodiment and its broader aspects and without departing from the principles, spirit and scope of the invention, and that all such modifications, equivalents, improvements and equivalents as may be included within the scope of the invention are intended to be protected by the claims.

Claims

1. A decoration recommendation information generation method is characterized by comprising the following steps:

2. The method according to claim 1, wherein the step of determining the user type to which the target user belongs according to the preference information comprises:

extracting preference features based on the preference information;

3. The method of claim 2, wherein the step of extracting preferred features based on the preference information is preceded by the method further comprising the step of training the clustering model:

4. The method of claim 1, wherein the step of obtaining the preference information of the target user and the house property information of the house to be finished is preceded by the step of training the recommendation model by:

5. The method of claim 4, wherein the historical finish recommendation information includes color information; the color information comprises wall surface color information and/or floor color information;

acquiring RGB color values of the color information;

6. The method of claim 5, wherein the recommendation model is a neural network model comprising an input layer, an intermediate layer, and an output layer,

7. The method of claim 1, wherein after the step of obtaining at least one fitment recommendation, the method further comprises:

8. A decoration recommendation information generating device is provided,it is composed ofCharacterized in that it comprises:

9. A computer readable storage medium storing computer instructions which, when executed by a processor, implement the steps of the method of any one of claims 1 to 7.

10. An electronic device, comprising:

a processor;

a memory for storing the processor-executable instructions;

the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method of any one of claims 1-7.