CN114399367A - Insurance product recommendation method, device, equipment and storage medium - Google Patents
Insurance product recommendation method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114399367A CN114399367A CN202210079954.0A CN202210079954A CN114399367A CN 114399367 A CN114399367 A CN 114399367A CN 202210079954 A CN202210079954 A CN 202210079954A CN 114399367 A CN114399367 A CN 114399367A
- Authority
- CN
- China
- Prior art keywords
- service
- user
- feature
- characteristic
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Economics (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of artificial intelligence and discloses a method, a device, equipment and a storage medium for recommending insurance products. The method comprises the following steps: the method comprises the steps of obtaining vehicle service information of a target user, and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions; enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and analyzing the service preference of a target user based on the target dimension characteristic to obtain a relevancy score of the service preference; constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information; and recommending the product to the target user according to the product recommendation information. The invention realizes the recommendation of insurance products through the vehicle service information.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to an insurance product recommendation method, device, equipment and storage medium.
Background
With the popularization of smart phones, the mobile internet enters a high-speed development stage, and the insurance industry is about a new expansion tide. By means of the Internet financial and big data, the business of an insurance company is full-bloomed, some of the businesses cooperate with e-commerce to build a self-operated e-commerce platform, and some cooperate with the Internet company to embed the insurance into the Internet ecology, so that the deep touch screen of the insurance is realized. Therefore, complicated and complicated information is brought, the problem of information overload is caused, and the problem of how to mine useful information from a plurality of information is a key concern of insurance companies in assisting the marketing of insurance products.
Conventional insurance product recommendations use essentially historical sales data, which is not considered comprehensive enough. Through data analysis, the preference degree of the customer on vehicle-related services (fuel filling service, car washing service and the like) is found to influence the purchase insurance product to a certain degree. There is currently no method for artificial intelligence insurance product recommendations directly for vehicle fueling and other related services.
Disclosure of Invention
The invention mainly aims to solve the technical problem that the recommendation pertinence of the existing artificial intelligence product is low.
The invention provides an insurance product recommendation method in a first aspect, which comprises the following steps: the method comprises the steps of obtaining vehicle service information of a target user, and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions; enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and analyzing the service preference of a target user based on the target dimension characteristic to obtain a relevancy score of the service preference; constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information; and recommending the product to the target user according to the product recommendation information.
Optionally, in a first implementation manner of the first aspect of the present invention, extracting, according to a preset service dimension, a plurality of entity features in the vehicle service information includes: according to the preset service dimensionality, service indexes of the vehicle service information in each preset service scene are counted; respectively calculating the classification attribute values of the service indexes in each service scene; and identifying a plurality of entity characteristics corresponding to the service dimensionality in each service scene according to the classification attribute value.
Optionally, in a second implementation manner of the first aspect of the present invention, enhancing each entity feature according to a preset feature engineering to obtain a corresponding target dimension feature includes: establishing a feature matrix by adopting each entity feature, and performing feature decomposition on the feature matrix to obtain a basic feature and a feature fusion factor; performing feature reconstruction by using the basic features and the feature fusion factors to obtain fusion features, wherein the fusion features comprise single features and composite features; and selecting the composite feature in the fusion features as the target dimension feature after the enhancement of each entity feature.
Optionally, in a third implementation manner of the first aspect of the present invention, performing service preference analysis on a target user based on a target dimension feature, and obtaining a relevance score of the service preference includes: distinguishing target dimension characteristics of each user in target users to obtain a plurality of characteristic sets, and extracting preference commonality information among the characteristic sets; and calculating the similarity distance between the feature sets according to the preference commonality information, and calculating the relevancy score between the users according to the similarity distance.
Optionally, in a fourth implementation manner of the first aspect of the present invention, constructing the user service representation according to the relevancy score includes: constructing a user service portrait model according to the relevancy score, and initializing parameters in the user service portrait model to obtain distribution parameters; generating grouping feature vectors by using the user service portrait model according to the distribution parameters, grouping the grouping feature vectors, and reconstructing the grouping feature vectors; calculating grouping errors based on the grouping results, and calculating reconstruction probabilities corresponding to the grouping feature vectors based on the reconstruction results; and optimizing the user service portrait model by adopting a preset minimization model method until the sum of the grouping error and the reconstruction probability is less than a preset threshold value to obtain the user service portrait.
Optionally, in a fifth implementation manner of the first aspect of the present invention, based on the user service representation, performing product matching on the target user by using a preset collaborative filtering algorithm, and obtaining the product recommendation information includes: according to the user service portrait, matching interested products and corresponding interested degrees of all users in the target users; calculating cosine similarity between the interested products according to the interested degree, and determining recommendation scores between the interested products and the users according to the cosine similarity; and adjusting the interested products corresponding to the users according to the recommendation scores to obtain product recommendation information.
A second aspect of the present invention provides an insurance product recommendation apparatus, including: the extraction module is used for acquiring the vehicle service information of a target user and extracting a plurality of entity characteristics in the vehicle service information according to a preset service dimension; the analysis module is used for enhancing the entity characteristics according to the preset characteristic engineering to obtain corresponding target dimension characteristics, and performing service preference analysis on the target user based on the target dimension characteristics to obtain the relevancy score of the service preference; the matching module is used for constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information; and the recommending module is used for recommending the product to the target user according to the product recommending information.
Optionally, in a first implementation manner of the second aspect of the present invention, the extraction module includes: the statistical unit is used for counting the service indexes of the vehicle service information in each preset service scene according to the preset service dimensionality; the first calculation unit is used for respectively calculating the classification attribute values of the service indexes in each service scene; and the identification unit is used for identifying a plurality of entity characteristics corresponding to the service dimensionality in each service scene according to the classification attribute value.
Optionally, in a second implementation manner of the second aspect of the present invention, the analysis module includes: the decomposition unit is used for establishing a characteristic matrix by adopting each entity characteristic and decomposing the characteristic matrix to obtain a basic characteristic and a characteristic fusion factor; the reconstruction unit is used for reconstructing the features by using the basic features and the feature fusion factors to obtain fusion features, wherein the fusion features comprise single features and composite features; and the selecting unit is used for selecting the composite feature in the fusion features as the target dimension feature after the enhancement of each entity feature.
Optionally, in a third implementation manner of the second aspect of the present invention, the analysis module further includes: the distinguishing unit is used for distinguishing target dimension characteristics of each user in the target users to obtain a plurality of characteristic sets and extracting preference commonality information among the characteristic sets; and the second calculating unit is used for calculating the similarity distance among the feature sets according to the preference commonality information and calculating the relevancy score among the users according to the similarity distance.
Optionally, in a fourth implementation manner of the second aspect of the present invention, the matching module includes: the construction unit is used for constructing a user service portrait model according to the relevancy score and carrying out initialization processing on parameters in the user service portrait model to obtain distribution parameters; the grouping unit is used for generating grouping characteristic vectors by using the user service portrait model according to the distribution parameters, grouping the grouping characteristic vectors and reconstructing the grouping characteristic vectors; a third calculation unit, configured to calculate a grouping error based on a result of the grouping, and calculate a reconstruction probability corresponding to the grouping feature vector based on a result of the reconstruction; and the optimization unit is used for optimizing the user service portrait model by adopting a preset minimization model method until the sum of the grouping error and the reconstruction probability is less than a preset threshold value, so as to obtain the user service portrait.
Optionally, in a fifth implementation manner of the second aspect of the present invention, the matching module further includes: the matching unit is used for matching the interesting products and the corresponding interesting degrees of all users in the target users according to the user service portraits; the determining unit is used for calculating cosine similarity among the interested products according to the interested degree and determining recommendation scores between the interested products and the users according to the cosine similarity; and the adjusting unit is used for adjusting the interested products corresponding to the users according to the recommendation scores to obtain the product recommendation information.
A third aspect of the present invention provides an insurance product recommendation apparatus including: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes instructions in the memory to cause the insurance product recommendation device to perform the insurance product recommendation method described above.
A fourth aspect of the present invention provides a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to execute the insurance product recommendation method described above.
According to the technical scheme provided by the invention, vehicle service information is analyzed from the perspective of preset service dimensions, a plurality of entity characteristics are extracted and subjected to characteristic enhancement so as to be used for carrying out service preference analysis and calculation similarity on a target user, a user service portrait is constructed according to the relevance score, and finally a collaborative filtering algorithm is utilized to carry out product matching on the target user by adopting the user portrait so as to recommend the target user, so that the artificial intelligent recommendation of products for the user aiming at the relevant service data of the vehicle is realized.
Drawings
FIG. 1 is a schematic diagram of a first embodiment of an insurance product recommendation method in an embodiment of the invention;
FIG. 2 is a schematic diagram of a second embodiment of an insurance product recommendation method in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of a third embodiment of an insurance product recommendation method in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of an embodiment of an insurance product recommendation device in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of another embodiment of an insurance product recommendation device in accordance with an embodiment of the present invention;
FIG. 6 is a schematic diagram of an embodiment of an insurance product recommendation device in an embodiment of the invention.
Detailed Description
The embodiment of the invention provides a method, a device, equipment and a storage medium for recommending insurance products, which are used for acquiring vehicle service information of a target user and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions; enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and analyzing the service preference of a target user based on the target dimension characteristic to obtain a relevancy score of the service preference; constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information; and recommending the product to the target user according to the product recommendation information. The invention realizes the recommendation of insurance products through the vehicle service information.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
For easy understanding, the following describes a specific process of an embodiment of the present invention, and referring to fig. 1, a first embodiment of an insurance product recommendation method according to an embodiment of the present invention includes:
101. the method comprises the steps of obtaining vehicle service information of a target user, and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions;
it is to be understood that the executing subject of the present invention may be an insurance product recommendation device, and may also be a terminal or a server, which is not limited herein. The embodiment of the present invention is described by taking a server as an execution subject.
In this embodiment, the target user may include a plurality of users, and the vehicle service information refers to related services performed by the user on the vehicle, including a refueling service, a car washing service, a high-speed passing service, a number of times of using the vehicle, and the like; the method comprises the steps of presetting a plurality of service dimensions, such as a time dimension, a space dimension, a user dimension, a merchant dimension, a consumption dimension and the like, then extracting a plurality of entity characteristics in vehicle service information according to the preset service dimensions, wherein the entity characteristics refer to the summary of actual service data of a vehicle by a target user, and the analysis and the characteristic extraction can be carried out from three entities, namely a merchant, a customer and a service.
102. Enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and analyzing the service preference of a target user based on the target dimension characteristic to obtain a relevancy score of the service preference;
in this embodiment, the step of enhancing each entity feature by using a preset feature engineering means that each entity feature is fused according to an approximation degree to obtain a high-dimensional fused feature, and then un-fused entity features are screened out, so that a target dimension feature with more accurate semantic expression can be obtained. And subsequently, analyzing the service preferences of one or more users in the target users according to the target dimension characteristics obtained by dimension increasing, and determining the preference similarity degree between the users through the obtained relevancy scores, for example, among a plurality of users, the user scores with similar service preferences a are close to the standard reference score corresponding to the service preferences a, and the user scores with similar service preferences b are close to the standard reference score corresponding to the service preferences b.
103. Constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information;
in this embodiment, product tags of different insurance products (hereinafter referred to as products) are stored in a database, a collaborative filtering algorithm is used for recommending products, the product tags of the stored products are matched through a user service portrait to represent interesting products of each user, the interest degree of each user in the interesting products is calculated, and the products corresponding to each user are matched to obtain product recommendation information.
In this embodiment, the user service representation model refers to a calculation framework of a user service representation, and at this time, the user service representation model does not have a representation of user service features, and parameters in the user service representation model include: user service grouping variable parameter, user service distribution variable parameter, service object grouping variable parameter and service object distribution variable parameter.
104. And recommending the product to the target user according to the product recommendation information.
In this embodiment, each user may be recommended finally according to the calculated interesting product.
In the embodiment of the invention, from the perspective of preset service dimensionality, vehicle service information is analyzed, a plurality of entity characteristics are extracted and subjected to characteristic enhancement so as to be used for carrying out service preference analysis and calculation similarity on a target user, a user service portrait is constructed according to the relevance score, and finally, a collaborative filtering algorithm is utilized to carry out product matching on the target user by adopting the user portrait so as to recommend the target user, so that the artificial intelligent recommendation of products for the user aiming at the relevant service data of the vehicle is realized.
Referring to fig. 2, a second embodiment of the insurance product recommendation method according to the embodiment of the invention includes:
201. acquiring vehicle service information of a target user, and counting service indexes of the vehicle service information in each preset service scene according to preset service dimensions;
202. respectively calculating the classification attribute values of the service indexes in each service scene;
203. identifying a plurality of entity characteristics corresponding to the service dimensionality in each service scene according to the classification attribute value;
in this embodiment, the service indexes may be counted from three time dimensions of accumulation, monthly and weekly, the service scene may count the service indexes in the vehicle service information from the perspective of the merchant, the perspective of the user, and the perspective of the service, and the service indexes counted from the perspective of the merchant may include: the total number of refueling orders, the farthest radiation distance (the wider the range of the influence of the merchant is, the higher the level of the merchant is), the number of provided vehicle services (the more vehicle services can be provided, the more active the merchant is), the service conditions of different services and the like; the service indicators statistically derived from the user perspective may include: the number of the gas stations for visiting, the number of the obtained vehicle services, the times of performing the vehicle services on weekends, the farthest distance for traveling and the like; the service indicators statistically derived from the service perspective may include: the number of times of getting various services, the number of users getting the services, the number of merchants issuing the services, and the like are considered.
In this embodiment, different service indicators have corresponding weights in different service scenarios, and weighting is performed on the service indicators obtained through statistics, so that corresponding entity features can be obtained, where a specific calculation formula may be: sij＝ai*WijWherein S isijClass attribute value, a, for ith service index at jth service scenarioiIs the ith service index, WijAnd the weight corresponding to the ith service index in the jth scene. After the classification attribute value is obtained through calculation, a plurality of entity characteristics on the time dimension, the merchant angle, the user angle and the service angle can be obtained.
204. Establishing a feature matrix by adopting each entity feature, and performing feature decomposition on the feature matrix to obtain a basic feature and a feature fusion factor;
205. performing feature reconstruction by using the basic features and the feature fusion factors to obtain fusion features, wherein the fusion features comprise single features and composite features;
206. selecting composite features in the fusion features as target dimension features after the enhancement of the entity features, and performing service preference analysis on a target user based on the target dimension features to obtain a relevancy score of the service preference;
in this embodiment, if the total number of the obtained entity features is IJ, the total number of the feature classes is D, and the total number of the service scenarios is J, the IJ samples are divided into D classes, each class includes K entity features, that is, D entity features with different dimensions, that is, J feature matrices with D × K dimensions can be obtained, and each feature matrix includes an entity feature in one service scenario.
In this embodiment, after the feature matrix is obtained by the establishment, a non-negative matrix decomposition method of a gradient projection method may be adopted to perform feature decomposition processing on the feature matrix to obtain a basic feature matrix composed of a preset number r of basic features
And the feature fusion factor matrix of the IJ r dimension
Wherein, the characteristic matrix of the d-th characteristic and the j-th category obtains a corresponding basic characteristic matrix and a fusion coefficient matrix which are expressed by the following formula,
and decomposing the features of the feature matrix to obtain the basic features and the fusion features.
In this embodiment, when the basic features and the feature fusion factors are subjected to feature reconstruction, the decomposed basic features and the feature fusion factors may be fused by using a random index method to obtain corresponding fusion features. The number of fusion features desired to be reconstructed can be preset; then randomly generating a fusion characteristic reconstruction index; then, selecting a feature fusion factor and a basic feature corresponding to the feature fusion factor according to the reconstruction index, multiplying the feature fusion factor and the basic feature and accumulating the feature fusion factor and the basic feature to obtain a fusion feature; and performing feature reconstruction in the same way until the reconstructed fusion features reach the preset number. And finally, selecting the composite features subjected to fusion as target dimension features. In the feature fusion, the following formula can be specifically adopted:
wherein RIJ E [1, pk]Is an integer value and represents a random index, formed by a function
According to
Is generated at random by the dimension of (a),
representing a base feature matrix
The number b of the first row (c),
representing a feature fusion factor matrix
Row a, column RIJ,
representing the nth fusion feature of the kth class sample.
207. Constructing a user service portrait according to the relevancy score, and performing product matching on a target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information;
208. and recommending the product to the target user according to the product recommendation information.
In the embodiment of the invention, the entity characteristics of the vehicle service information in three aspects of users, businesses and services are counted from the time dimension, then the entity characteristics are utilized to carry out characteristic fusion and characteristic screening, the target dimension characteristics related to product recommendation are selected, and the relevance score is used for classifying the service preference of each user so as to be used for carrying out product matching and recommendation according to the preference category of the user.
Referring to fig. 3, a third embodiment of the insurance product recommendation method according to the embodiment of the invention includes:
301. the method comprises the steps of obtaining vehicle service information of a target user, and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions;
302. enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, distinguishing the target dimension characteristic of each user in the target users to obtain a plurality of characteristic sets, and extracting preference commonality information among the characteristic sets;
303. calculating the similarity distance between the feature sets according to the preference commonality information, and calculating the relevancy score between the users according to the similarity distance;
in this embodiment, the target dimension features of different users are identified by corresponding codes, the target dimension features to which each user belongs can be distinguished through identification, then preference commonality information between each user is calculated, and differences between each feature set are measured by a similar distance such as a Jaccard distance, so as to calculate a relevancy score between each user. Wherein, the similarity distance can be calculated by adopting the following formula:
and M (H, L) is the similar distance between the two feature sets H and L, and then the relevance scores of the feature sets H and L before corresponding to the two users can be obtained through 1-M (H, L).
304. Constructing a user service portrait model according to the relevancy score, and initializing parameters in the user service portrait model to obtain distribution parameters;
305. generating grouping feature vectors by using the user service portrait model according to the distribution parameters, grouping the grouping feature vectors, and reconstructing the grouping feature vectors;
306. calculating grouping errors based on the grouping results, and calculating reconstruction probabilities corresponding to the grouping feature vectors based on the reconstruction results;
307. optimizing the user service portrait model by adopting a preset minimization model method until the sum of the grouping error and the reconstruction probability is less than a preset threshold value to obtain a user service portrait;
in this embodiment, the user service representation model refers to a calculation framework of a user service representation, and at this time, the user service representation model does not have a representation of user service features, and parameters in the user service representation model include: user service grouping variable parameter, user service distribution variable parameter, service object grouping variable parameter and service object distribution variable parameter.
In this embodiment, when initializing the user service portrait model, the method includes initializing a user service grouping variable parameter and a user service distribution variable parameter to obtain a user service grouping distribution parameter, and initializing a service object grouping variable parameter and a service object distribution variable parameter to obtain a service object distribution variable parameter, where the distribution parameter includes the user service grouping distribution parameter and the service object distribution variable parameter. Wherein, the user service grouping distribution parameter and the service object distribution variable parameter conform to Beta distribution.
Specifically, the grouping division probability of the user service characteristic vector is generated according to the user service grouping distribution parameters to serve as a first probability, and meanwhile, a group corresponding to the first probability which is larger than a first preset probability threshold is selected and set; and generating a grouping feature vector of the user according to the group, namely a user service portrait. In addition, the grouping division probability of the service object features is generated according to the service object distribution variable parameters as a second probability, a group with the second probability larger than a second preset probability threshold is selected, and further, the grouping feature vector of the user, namely the service object feature vector, is generated according to the group.
In this embodiment, the first label corresponding to the grouping determination result and the second label corresponding to the preset grouping may be compared and calculated by using a probability-based classification algorithm, and the grouping error between the grouping determination result and the set grouping may be determined according to the comparison result. In addition, the user service characteristic vector and the service object characteristic vector are reconstructed to generate a service matrix, and the probability that the behavior matrix generated by reconstruction is a matrix constructed by the vehicle service information is calculated, namely the reconstruction probability.
And evaluating the user service portrait model by using a minimization model method, and updating parameters in the model so as to optimize the model. The optimization process achieves the purposes of model evaluation and optimization updating by updating parameters to minimize an objective function with distance limits. The specific description is as follows:
wherein KL (II) represents distribution difference, p0 represents the real distribution of vectors in a matrix after the grouped feature vectors are reconstructed, q represents approximate distribution which cannot be reconstructed, R represents a matrix obtained by constructing vehicle service information, phi represents a parameter set by decomposing model variables of the reconstructed matrix, and psi represents a parameter set by grouping discrimination variables.
308. According to the user service portrait, matching interested products and corresponding interested degrees of all users in the target users;
309. calculating cosine similarity between the interested products according to the interested degree, and determining recommendation scores between the interested products and the users according to the cosine similarity;
310. according to the recommendation scores, adjusting the interesting products corresponding to the users to obtain product recommendation information;
in this embodiment, product tags of different products are stored in the database, so that a collaborative filtering algorithm is used to recommend the products, the product tags of the stored products are matched in advance through the user service representation, so as to represent the interesting products of each user, and the interesting degree of each user to the interesting products is calculated. And then evaluating the similarity between each product through cosine similarity, and subsequently recommending the interested products among users. The cosine similarity between the interested products of any two users can be calculated by adopting the following formula:
wherein, N (u) represents the interest degree of the u product of the first user, and N (v) represents the interest degree of the v product of the second user.
311. And recommending the product to the target user according to the product recommendation information.
In the embodiment of the invention, a user service portrait model is pre-constructed through similarity scores to determine a calculation framework of a user service portrait, parameters in the model are initialized, grouped and reconstructed, meanwhile, a minimization model method is adopted to optimize the user service portrait model, the optimization effect is measured through grouping errors and reconstruction probability, and finally, the user service portrait is obtained for visualizing user service preference and improving the pertinence of subsequent product recommendation.
With reference to fig. 4, the method for recommending an insurance product in an embodiment of the present invention is described above, and an insurance product recommending apparatus in an embodiment of the present invention is described below, where an embodiment of the insurance product recommending apparatus in an embodiment of the present invention includes:
the extraction module 401 is configured to obtain vehicle service information of a target user, and extract a plurality of entity features in the vehicle service information according to a preset service dimension;
an analysis module 402, configured to enhance each entity feature according to a preset feature engineering to obtain a corresponding target dimension feature, and perform service preference analysis on a target user based on the target dimension feature to obtain a relevance score of the service preference;
the matching module 403 is configured to construct a user service representation according to the relevancy score, and perform product matching on a target user by using a preset collaborative filtering algorithm based on the user service representation to obtain product recommendation information;
and the recommending module 404 is configured to recommend a product to the target user according to the product recommending information.
In the embodiment of the invention, from the perspective of preset service dimensionality, vehicle service information is analyzed, a plurality of entity characteristics are extracted and subjected to characteristic enhancement so as to be used for carrying out service preference analysis and calculation similarity on a target user, a user service portrait is constructed according to the relevance score, and finally, a collaborative filtering algorithm is utilized to carry out product matching on the target user by adopting the user portrait so as to recommend the target user, so that the artificial intelligent recommendation of products for the user aiming at the relevant service data of the vehicle is realized.
Referring to fig. 5, another embodiment of the insurance product recommendation device according to the embodiment of the present invention includes:
the extraction module 401 is configured to obtain vehicle service information of a target user, and extract a plurality of entity features in the vehicle service information according to a preset service dimension;
an analysis module 402, configured to enhance each entity feature according to a preset feature engineering to obtain a corresponding target dimension feature, and perform service preference analysis on a target user based on the target dimension feature to obtain a relevance score of the service preference;
the matching module 403 is configured to construct a user service representation according to the relevancy score, and perform product matching on a target user by using a preset collaborative filtering algorithm based on the user service representation to obtain product recommendation information;
and the recommending module 404 is configured to recommend a product to the target user according to the product recommending information.
Specifically, the extraction module 401 includes:
the statistical unit 4011 is configured to count service indexes of the vehicle service information in each preset service scenario according to preset service dimensions;
a first calculating unit 4012, configured to calculate classification attribute values of the service index in each service scenario respectively;
the identifying unit 4013 is configured to identify, according to the classification attribute value, a plurality of entity features corresponding to the service dimension in each service scenario.
Specifically, the analysis module 402 includes:
the decomposition unit 4021 is configured to establish a feature matrix by using each entity feature, and perform feature decomposition on the feature matrix to obtain a basic feature and a feature fusion factor;
a reconstruction unit 4022, configured to perform feature reconstruction by using the basic features and the feature fusion factors to obtain fusion features, where the fusion features include single features and composite features;
a selecting unit 4023, configured to select a composite feature in the fusion features as a target dimension feature after enhancement of each entity feature.
Specifically, the analysis module 402 further includes:
the distinguishing unit 4024 is configured to distinguish target dimension features to which each user belongs from target users to obtain a plurality of feature sets, and extract preference commonality information between the feature sets;
the second calculating unit 4025 is configured to calculate a similarity distance between each feature set according to the preference commonality information, and calculate a relevancy score between each user according to the similarity distance.
Specifically, the matching module 403 includes:
a construction unit 4031, configured to construct a user service representation model according to the relevancy score, and perform initialization processing on parameters in the user service representation model to obtain distribution parameters;
a grouping unit 4032, configured to generate grouping feature vectors according to the distribution parameters by using the user service representation model, group the grouping feature vectors, and reconstruct the grouping feature vectors;
a third calculation unit 4033 for calculating grouping errors based on the grouping result and calculating the reconstruction probability corresponding to the grouping feature vector based on the reconstruction result;
and the optimizing unit 4034 is configured to optimize the user service profile model by using a preset minimization model method, and obtain the user service profile until a sum of the grouping error and the reconstruction probability is smaller than a preset threshold.
Specifically, the matching module 403 further includes:
a matching unit 4035 for matching the interested products and the corresponding interested degrees of each user in the target users according to the user service representation;
a determining unit 4036, configured to calculate cosine similarities between the interested products according to the interest degrees, and determine recommendation scores between the interested products and the users according to the cosine similarities;
an adjusting unit 4037, configured to adjust the interested products corresponding to the users according to the recommendation scores, so as to obtain product recommendation information.
In the embodiment of the invention, the entity characteristics of the vehicle service information in three aspects of users, businesses and services are counted from the time dimension, then the entity characteristics are utilized to carry out characteristic fusion and characteristic screening, the target dimension characteristics related to product recommendation are selected, and the relevance score is used for classifying the service preference of each user so as to be used for carrying out product matching and recommendation according to the preference category of the user subsequently; in addition, a user service portrait model is pre-constructed through the similarity score to determine a calculation framework of the user service portrait, then parameters in the model are initialized, grouped and reconstructed, meanwhile, a minimization model method is adopted to optimize the user service portrait model, the optimization effect is measured through grouping errors and reconstruction probability, and finally the user service portrait is obtained for visualizing user service preference and improving the pertinence of subsequent product recommendation.
Fig. 4 and 5 describe the insurance product recommendation apparatus in the embodiment of the present invention in detail from the perspective of the modular functional entity, and the insurance product recommendation device in the embodiment of the present invention is described in detail from the perspective of hardware processing.
Fig. 6 is a schematic structural diagram of an insurance product recommendation device 600 according to an embodiment of the present invention, which may have a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 610 (e.g., one or more processors) and a memory 620, one or more storage media 630 (e.g., one or more mass storage devices) storing applications 633 or data 632. Memory 620 and storage medium 630 may be, among other things, transient or persistent storage. The program stored on the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations for the insurance product recommendation device 600. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the insurance product recommendation device 600.
The insurance product recommendation device 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input-output interfaces 660, and/or one or more operating systems 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, and so forth. Those skilled in the art will appreciate that the configuration of the insurance product recommendation device illustrated in figure 6 does not constitute a limitation of the insurance product recommendation device and may include more or fewer components than illustrated, or some components in combination, or a different arrangement of components.
The invention also provides insurance product recommendation equipment, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the insurance product recommendation method in the embodiments.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, which may also be a volatile computer readable storage medium, having stored therein instructions, which, when executed on a computer, cause the computer to perform the steps of the insurance product recommendation method.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. An insurance product recommendation method, comprising:
the method comprises the steps of obtaining vehicle service information of a target user, and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions;
enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and analyzing the service preference of the target user based on the target dimension characteristic to obtain a relevance score of the service preference;
constructing a user service portrait according to the relevancy score, and performing product matching on the target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information;
and recommending the product to the target user according to the product recommendation information.
2. The insurance product recommendation method of claim 1, wherein the extracting a plurality of entity features in the vehicle service information according to a preset service dimension comprises:
according to preset service dimensions, service indexes of the vehicle service information in each preset service scene are counted;
respectively calculating the classification attribute values of the service indexes in each service scene;
and identifying a plurality of entity characteristics corresponding to the service dimensionality in each service scene according to the classification attribute value.
3. The insurance product recommendation method of claim 1, wherein the enhancing each of the entity features according to a preset feature project to obtain corresponding target dimensional features comprises:
establishing a feature matrix by adopting each entity feature, and performing feature decomposition on the feature matrix to obtain a basic feature and a feature fusion factor;
performing feature reconstruction by using the basic features and the feature fusion factors to obtain fusion features, wherein the fusion features comprise single features and composite features;
and selecting a composite feature in the fusion features as the target dimension feature of each entity feature after enhancement.
4. The insurance product recommendation method of claim 1, wherein the analyzing the target user for service preferences based on the target dimensional characteristics to obtain a relevancy score of service preferences comprises:
distinguishing target dimension characteristics of each user in the target users to obtain a plurality of characteristic sets, and extracting preference commonality information among the characteristic sets;
and calculating the similarity distance between the feature sets according to the preference commonality information, and calculating the relevancy score between the users according to the similarity distance.
5. The insurance product recommendation method of claim 1, wherein said constructing a user service representation from said relevancy scores comprises:
constructing a user service portrait model according to the relevancy score, and initializing parameters in the user service portrait model to obtain distribution parameters;
generating grouping feature vectors by using the user service portrait model according to the distribution parameters, grouping the grouping feature vectors, and reconstructing the grouping feature vectors;
calculating grouping errors based on grouping results, and calculating reconstruction probabilities corresponding to the grouping feature vectors based on reconstruction results;
and optimizing the user service portrait model by adopting a preset minimization model method until the sum of the grouping error and the reconstruction probability is less than a preset threshold value, so as to obtain the user service portrait.
6. The insurance product recommendation method according to any one of claims 1-5, wherein the performing product matching on the target user based on the user service representation using a preset collaborative filtering algorithm to obtain product recommendation information comprises:
according to the user service portrait, matching interested products and corresponding interested degrees of all users in the target users;
calculating cosine similarity between the interested products according to the interested degree, and determining recommendation scores between the interested products and the users according to the cosine similarity;
and adjusting the interested products corresponding to the users according to the recommendation scores to obtain product recommendation information.
7. An insurance product recommendation apparatus, comprising:
the extraction module is used for acquiring vehicle service information of a target user and extracting a plurality of entity characteristics in the vehicle service information according to preset service dimensions;
the analysis module is used for enhancing each entity characteristic according to preset characteristic engineering to obtain a corresponding target dimension characteristic, and performing service preference analysis on the target user based on the target dimension characteristic to obtain a relevance score of service preference;
the matching module is used for constructing a user service portrait according to the relevancy score, and performing product matching on the target user by using a preset collaborative filtering algorithm based on the user service portrait to obtain product recommendation information;
and the recommending module is used for recommending the product to the target user according to the product recommending information.
8. The insurance product recommendation device of claim 7, wherein the analysis module comprises:
the decomposition unit is used for establishing a characteristic matrix by adopting each entity characteristic and performing characteristic decomposition on the characteristic matrix to obtain a basic characteristic and a characteristic fusion factor;
the reconstruction unit is used for reconstructing features by using the basic features and the feature fusion factors to obtain fusion features, wherein the fusion features comprise single features and composite features;
and the selecting unit is used for selecting the composite feature in the fusion features as the target dimension feature after the enhancement of each entity feature.
9. An insurance product recommendation apparatus, characterized in that the insurance product recommendation apparatus comprises: a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invokes the instructions in the memory to cause the insurance product recommendation device to perform the steps of the insurance product recommendation method of any one of claims 1-6.
10. A computer readable storage medium having instructions stored thereon, wherein the instructions, when executed by a processor, implement the steps of the insurance product recommendation method of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079954.0A CN114399367A (en) | 2022-01-24 | 2022-01-24 | Insurance product recommendation method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079954.0A CN114399367A (en) | 2022-01-24 | 2022-01-24 | Insurance product recommendation method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114399367A true CN114399367A (en) | 2022-04-26 |
Family
ID=81233594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210079954.0A Pending CN114399367A (en) | 2022-01-24 | 2022-01-24 | Insurance product recommendation method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114399367A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116738072A (en) * | 2023-08-15 | 2023-09-12 | 深圳大学 | Multidimensional recommendation method combining human factor information |
| CN116777530A (en) * | 2023-08-23 | 2023-09-19 | 山东四季汽车服务有限公司 | Automobile service system based on intelligent recommendation |
-
2022
- 2022-01-24 CN CN202210079954.0A patent/CN114399367A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116738072A (en) * | 2023-08-15 | 2023-09-12 | 深圳大学 | Multidimensional recommendation method combining human factor information |
| CN116738072B (en) * | 2023-08-15 | 2023-11-14 | 深圳大学 | Multidimensional recommendation method combining human factor information |
| CN116777530A (en) * | 2023-08-23 | 2023-09-19 | 山东四季汽车服务有限公司 | Automobile service system based on intelligent recommendation |
| CN116777530B (en) * | 2023-08-23 | 2023-11-07 | 山东四季汽车服务有限公司 | Automobile service system based on intelligent recommendation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108363821A (en) | A kind of information-pushing method, device, terminal device and storage medium | |
| CN109739844B (en) | Data classification method based on attenuation weight | |
| EP4322031A1 (en) | Recommendation method, recommendation model training method, and related product | |
| CN114399367A (en) | Insurance product recommendation method, device, equipment and storage medium | |
| CN112990386B (en) | User value clustering method and device, computer equipment and storage medium | |
| US20190080352A1 (en) | Segment Extension Based on Lookalike Selection | |
| EP3140799A1 (en) | An automatic statistical processing tool | |
| CN112232606B (en) | Service data prediction method, device, equipment and storage medium | |
| CN114612251A (en) | Risk assessment method, device, equipment and storage medium | |
| CN114997916A (en) | Prediction method, system, electronic device and storage medium of potential user | |
| CN114266443A (en) | Data evaluation method and device, electronic equipment and storage medium | |
| CN106372964A (en) | Behavior loyalty identification and management method, system and terminal | |
| CN112070126A (en) | Internet of things data mining method | |
| CN112990989B (en) | Value prediction model input data generation method, device, equipment and medium | |
| CN110674964A (en) | Search prediction system and method based on agricultural traceability information | |
| CN113327132A (en) | Multimedia recommendation method, device, equipment and storage medium | |
| CN113450004A (en) | Power credit report generation method and device, electronic equipment and readable storage medium | |
| CN112529319A (en) | Grading method and device based on multi-dimensional features, computer equipment and storage medium | |
| CN117237038A (en) | Commodity accurate exposure processing system based on flow engine | |
| CN115964944A (en) | Power industry industrial chain model generation method and device, storage medium and equipment | |
| CN112506930B (en) | Data insight system based on machine learning technology | |
| Shanti et al. | Machine Learning-Powered Mobile App for Predicting Used Car Prices | |
| CN115719244A (en) | User behavior prediction method and device | |
| Hamad et al. | Sentiment analysis of restaurant reviews in social media using naïve bayes | |
| Granov | Customer loyalty, return and churn prediction through machine learning methods: for a Swedish fashion and e-commerce company |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |